Born Together — Reared Apart

BORN TOGETHER—
R E A R E D A PA RT
The Landmark Minnesota Twin Study

NANCY L. SEGAL

Harvard University Press

Cambridge, Massachusetts, and London, England 2012
In memory of my parents, Al and Esther Segal,
and
To twins—reared apart, reared together,
and yet to meet.

Copyright © 2012 by the President and Fellows of Harvard College

All rights reserved
Printed in the United States of America

Library of Congress Cataloging-in-Publication Data

Segal, Nancy L., 1951–

Born together—reared apart : the landmark Minnesota twin study /
Nancy L. Segal.
p. cm.
Includes bibliographical references and index.
ISBN 978-0-674-05546-9 (alk. paper)
1. Twins—Psychology—Research. 2. Nature and nurture—Research. I. Title.
BF723.T9S436 2012
306.875—dc23 2011041636
Contents

Introduction 1

1 The Jim Twins (February–March 1979) 17

2 15,000 Questions × 137 Pairs 32

3 Early Findings (1979–1983) 65

4 Sexual Orientation, Cognition, and Medical

Traits (1984–1987) 82

5 Pivotal Papers: Personality and IQ (1988 and 1990) 97

6 Job Satisfaction, Cardiac Characteristics,

and More (1989–1990) 116

7 Psychopathology and Religiosity (1990) 136

8 Dental Traits, Allergies, and Vocational

Interests (1991–1992) 153

9 Creativity, Work Values, and Evolution (1992–1993) 171

10 Family Environments, Happiness, Sensation Seeking,

and the MMPI (1994–1997) 195

11 “Larks” and “Owls,” Ego Development, and

Authoritarianism (1998–2002) 221
vi CONTENTS

12 Twin Relationships, Social Attitudes, and Mental

Abilities (2003–2005) 246

13 Sexual Development, Fluctuating Asymmetry, Body Size,

and the Structure of Intelligence (2006 and Beyond) 266

14 Questions, Answers, and Twin Studies of the Future 298

Appendix A: Funding Sources 329

Appendix B: Glossary 333
Notes 339
Acknowledgments 397
Index 401
Introduction

Genes and glands are obviously important, but social learning

also has a dramatic role. Imagine the enormous differences that
would be found in the personalities of twins with identical
genetic endowments if they were raised apart in two different
families—or, even more striking, in two totally different cultures.
—Walter Mischel

C olumbia University professor Walter Mischel has been a

prominent figure in personality assessment and the study of emo-
tional control. He is also the 2011 recipient of the prestigious
Grawemeyer Award for contributions to psychology. The passage quoted
above, from his well-known undergraduate textbook, largely reflected
1970s’ and early 1980s’ views on factors affecting behavioral develop-
ment, but it would not appear in textbooks today.1 It has been replaced
by statements such as, “Currently, most kinship findings support a moder-
ate role for heredity. . . . Heritability research also reveals that genetic fac-
tors are important in personality.”2 This shift in perspective, from general
neglect or even contempt to increased acceptance of genetic contributions
to behavior, was not caused by the Minnesota Study of Twins Reared
Apart (MISTRA), but many claim that the study was instrumental in its
progression.
The classic twin method is a simple and elegant natural experiment for
examining genetic and environmental influences on behavior. The idea
of using twins to study factors affecting human behavioral variation was
first described by the British scholar Sir Frances Galton in 1875. Galton
recognized the usefulness of comparing twins who were “closely alike [in
childhood]” with those who were “exceedingly unlike.” He asserted that
“their history affords means of distinguishing between the effects of ten-
dencies received at birth, and of those that were imposed by the circum-
stances of their after lives; in other words, between the effects of nature
and nurture.”3 If what we now call monozygotic (MZ or identical) twins
show greater trait resemblance than dizygotic (DZ or nonidentical) twins,
2 INTRODUCTION

then genetic influence is likely. The biological bases of twinning had not
been established in Galton’s time, but he correctly surmised that “look-
alike” twins shared 100% of their genes and “less-alike” twins shared
fewer.
Richard Rende and his colleagues do not credit Galton with being
“Father of the Twin Method” because “he did not propose the compari-
son between identical and fraternal twin resemblance which is the essence
of the twin method.”4 They give this distinction to Curtis Merriman and
Hermann Siemens, whose 1924 writings described a more specific twin-
based approach with more familiar scientific terms. Director of the Min-
nesota Study of Twins Reared Apart, Thomas J. Bouchard Jr., and human
geneticist, Peter Propping, called Rende’s assertion a “disingenuous argu-
ment” based on ranking quantitative arguments above conceptual ones.
I would agree.5
The scientific study of twins did not become truly useful until the
fundamental distinction between twin types and the methods for distin-
guishing between them were established.6 These accomplishments were
made by a series of investigators between 1919 and 1925. In 1922, Leslie
Brainerd Arey proposed the labels “MZ” and “DZ” for “one egg” and
“two egg” twins, respectively.7 MZ twins share all their genes, having
split from a single fertilized egg (zygote) within the first two weeks after
conception. DZ twins share half their genes on average, by descent, hav-
ing originated from two separately fertilized eggs. In 1924, Siemens intro-
duced a similarity method for assigning pairs as one-egg or two-egg twins,
based on co-twin (twin pair member) comparison of hereditary traits.8
This method was widely used and is not unlike currently administered
physical resemblance questionnaires.
In 1956, geneticist Gordon Allen proposed changing the terms to
“monozygous” and “dizygous” because they capture the term “zygosis”
that refers to zygote formation. He also argued that these terms had one
less syllable, making them easier to write and say.9
Variations on the two types of twinning include MZ female twins who
sometimes differ in the expression of X-linked genetic conditions (due to
random inactivation of one X chromosome in each cell early in gestation),
and DZ twins who can be conceived by different fathers (when double-
ovulating women have multiple sexual partners close in time).10 These
twin pairs can help identify factors underlying co-twin differences in
behavior and physique.
The classic twin method is a powerful investigatory tool. Despite some
differences in the early development of twins and nontwins, Danish re-
searchers established that after the age of six years disease incidence and
 INTRODUCTION 3

mortality are comparable for twins and singletons (nontwins), a finding

confirmed for most behavioral and physical traits. Thus, most twin re-
search findings apply to the general population.11
The twin method rests on the equal environments assumption, the
premise that the environmental factors affecting similarity in a given trait,
such as sociability, are the same for MZ and DZ twin pairs. Although
MZ twins generally share more activities than do DZ twins, this differ-
ence is problematic only if (1) MZ twins are required to share school
activities to a greater degree than DZ twins, and (2) sharing activities
affects sociability in a meaningful way. If sharing activities increases socia-
bility, then the environments of the two types of twins would not be
equivalent with respect to this behavior. The equal environments assump-
tion would have been compromised in this hypothetical example. Here,
the environmental effect (the requirement that MZ twins share their ac-
tivities more often than DZ twins) is assumed to be an established school
policy, unrelated to the twins’ genetic proclivities.
Some critics have questioned the usefulness of twin studies, arguing
that MZ twins’ more similar rearing environments underlie their more
similar behaviors relative to DZ twins. However, these challenges and
others related to the equal environments assumption have been tested
empirically and have been refuted in most cases—similar treatment of
twins by others is not associated with similar behavioral outcomes. In
their landmark study of twins taking the National Merit Scholarship
Qualifying Test, John Loehlin and Robert Nichols showed that whether
co-twins were dressed alike, slept in the same room, or had the same
teachers was unrelated to their behavioral resemblance,12 a finding repli-
cated by others.13 In addition, more physically similar MZ twins (as rated
by parents) have not been perceived as being more alike in personality
than less physically similar twins.14 The lack of any biological connection
between physical appearance and psychological traits argues against the
former causally influencing the latter.15 However, twin studies of handed-
ness may violate the equal environments assumption because non-genetic
factors affecting handedness differ for MZ and DZ twins, as discussed in
Chapter 13.
Studying reared-apart MZ twins (MZA) and DZ twins (DZA) circum-
vents these challenges because the twins are raised in separate environ-
ments, often unaware of each other’s existence. The first scientific treat-
ment of MZA twins was a 1922 case study by psychologist Paul Popenoe
of twins Bessie and Jessie.16 The timing of this report is interesting because
the idea of studying twins raised apart had been brought to Galton’s at-
tention forty-seven years earlier. In a letter to Galton (circa 1874–75, the
4 INTRODUCTION

same period Galton published his classic paper on twins), M. Townsend

described his identical twin great-uncles who were “separated through
life” and who “rejoined one another at about 50.”17 Whether Galton tried
to study these twins or other separated twin pairs is unknown; they were
never mentioned in his writings.
Several case reports and studies of separated twins appeared in the
years that followed, such as Hermann J. Muller’s 1925 detailed analysis
of Popenoe’s twins.18 The last major reared-apart twin investigation was
conducted by Danish Professor Juel-Nielsen of Denmark in 196519—until
the MISTRA came into being.

The Climate of the Times

The MISTRA, initiated in March 1979, arrived at a critical juncture in
the history of psychology. Understanding the prevailing psychological
winds at that time is requisite to understanding why a reared-apart twin
study was so compelling yet controversial, why it attracted worldwide at-
tention, and why 1979 was an ideal time for it to start.
An environmentalist view, promulgated by the behaviorist James Wat-
son, dominated psychological theory and research from the 1930s through
the 1970s. Other than being born with a general capacity to learn, human
behavior was explained almost exclusively by forces outside the individ-
ual. Several important events were also instrumental in making the late
1960s and 1970s a time of heightened sensitivity to the origins of indi-
vidual and group differences. The 1964 Civil Rights Act outlawed dis-
crimination and segregation on the basis of color or race. The women’s
rights movement gained momentum in the 1960s, focusing on sex dis-
crimination in the classroom and workplace. Hovering above the scene
was the legacy of the Nazi regime (1933 to 1945) that promoted the gene-
tic superiority of some populations over others and racial purification by
eliminating individuals considered biologically inferior.20 It was in that
social climate that Arthur Jensen published the 1969 article “How Much
Can We Boost IQ and Scholastic Achievement?” which included a section
on possible genetically based race differences in intelligence test scores
(IQ).21 His discussion provoked angry outbursts from many academics,
educators, and others.22
The credibility of reared-apart twin research, published between 1943
and 1966 by British psychologist Sir Cyril Burt, was challenged by then
Princeton University Psychology Professor Leon Kamin and others in the
early 1970s, casting doubt over previous and subsequent twin studies.23
Kamin called attention to the implausibility of Burt’s .771 IQ correla-
 INTRODUCTION 5

tions, reported for three successive IQ analyses with samples of twenty-

one (1955), “over 30” (1958), and fifty-three (1966) MZA twin pairs.24 It
was also the case that no one had seen pictures of Burt with his twin sub-
jects or staff, although their absence is not proof of fraud. In fact, none of
the books by three early MZA twin investigators includes photographs
of the researchers with their twin participants, although photographs of
every pair appear in the American study headed by Horatio Newman
and a few pairs appear in the British study directed by James Shields.25
The Danish investigator Juel-Nielsen altered the names of the MZA twins
for whom he presented detailed case histories, so it is understandable that
he did not include any available pictures. In light of the Burt situation,
Bouchard would make it a practice to appear in pictures with every twin
pair we studied and every staff member we hired.
Charges that Burt fabricated both his data and his assistants have been
debated for years. The attention his case received is explained by Burt’s
leadership position in the psychological field (he was the first British psy-
chologist to be knighted) and his subject matter, about which scientists
cared greatly.26
Science historians Robert B. Joynson in 1989 and Ronald Fletcher in
1991 concluded their inquiries into the Burt affair by attributing the sus-
picious correlations to clerical errors and questionable aspects of his twin
sample to insufficient information about the participants and test condi-
tions.27 A 1995 study presented the views of five leading scholars in the
field, four of whom dismissed the charges but one of whom did not.28 The
study’s editor concluded that there was suspicion of fraud but no conclu-
sive evidence of guilt. The concerns over Burt’s twin research raised as
recently as 2007 are the same as those that were raised by James Shields
in 1978. Shields, who published a reared-apart twin study in England in
1962, noted the unlikely ease with which Burt recruited so many early
separated twins over a brief time period. Shields suggested (half seri-
ously) that Burt’s increasing MZA twin sample reflected a competitive
“one-upmanship” with Shields’s research.29
The Burt affair is unresolved at present; his data are no longer cited in
the psychological literature, but their omission has no effect on conclu-
sions over genetic influence on mental ability. The MISTRA, the Swedish
Adoption and Twin Study of Aging, the three early MZA twin studies, and
countless reared-together twin and adoption studies tell the same story.
Attention to a discredited study headed by New York psychoanalyst
Peter Neubauer of intentionally separated infant twins has occasionally
surfaced and accentuated the dim view of reared-apart twin studies.30 Be-
ginning in the 1960s, Neubauer gathered longitudinal data on the members
6 INTRODUCTION

of five MZA twin pairs and one MZA triplet set whose families were
purposefully not told they were adopting children of multiple births. The
controversial aspects of his data collection procedures attracted consid-
erable media attention in the early 1980s and still does. Recent attempts
by scientists and filmmakers to gain access to the original data have been
fruitless—prior to his death in 2008, Neubauer bequeathed the material
to Yale University’s Child Study Center with the stipulation that the file
remain closed until 2066.31 Psychiatrist Viola Bernard, consultant for the
Louise Wise Adoption Agency that placed the twins, favored their sepa-
ration. Bernard donated her twin-related documents to Columbia Uni-
versity, denying public access to the material until 2021.32
Most psychology and psychiatry textbooks in the 1960s, 1970s, and
early 1980s reflected environmentalist views. Others in the medical com-
munity also subscribed to a largely environmentalist paradigm in those
years. In an extraordinary case involving an accidentally castrated MZ
male twin infant during circumcision, the sex researcher, John Money,
argued that a child’s gender identity could be altered if surgical and psy-
chological interventions were introduced by eighteen months of age.33 His
findings went unchallenged publicly until 1997.34 Child psychiatrist Leo
Kanner blamed the apparently cold, conflicted practices of so-called re-
frigerator mothers for the onset of autism and other behavioral disorders
in young children.35 This belief persisted until the early 1970s.
In short, anyone suggesting that genes affected human behavior took a
huge academic risk. Harvard biologist Edward O. Wilson, famous for his
seminal work Sociobiology (1979), was doused with a bucket of water
as he approached the podium at the 1978 American Association for
the Advancement of Science (AAAS) convention in Washington, D.C.36
The tensions of those times are well captured in a review of essays on
race, social class, and IQ,37 and a book on the subject.38
At the same time, there were growing signs of discontent with the
prevailing environmentalist perspective. The roots of this dissatisfaction
came from a variety of sources. Research conducted in the 1950s and
1960s showed that pairings of some stimuli did not always produce pre-
dicted responses. An example is the Garcia Effect, in which rats learned
to avoid food associated with nausea but could not learn to avoid loud
noises or flashes of light when paired with the nausea.39 Thus, there ap-
peared to be biological constraints on behavior that stimulus-response
paradigms could not explain. The Garcia Effect was not fully appreciated
by psychologists until the late 1960s and 1970s.40
Continuing advances in human genetics strengthened the idea of gene-
behavior relationships. In 1959, investigators discovered that an extra
 INTRODUCTION 7

chromosome 21 was responsible for the behavioral and physical attri-

butes of Down syndrome (trisomy 21).41 In 1982, the detection of a link
between mental deficiency associated with phenylketonuria (PKU) and
lack of the enzyme phenylalanine hydroxylase further demonstrated a
genetic underpinning to behavior.42
Cognitive psychology, the branch of psychology concerned with how
people acquire, process, and store information, was also gaining ground in
the 1950s, 1960s, and 1970s.43 A biological view of language development
was introduced, claiming that language was too complex to be taught as
the behaviorists believed. Instead, the theory asserted that children were
born with the capacity to learn language and to master its rules almost
automatically.44 Computer models of information processing drew atten-
tion to how the mind might work to solve problems.45
Research summarized in an individual differences textbook by the late
Lee Willerman, published in 1979 (the year the MISTRA began), shows
the gradual but steady infusion of a genetic perspective into thinking about
behavior.46 Some of these studies predated those of many of the behavior-
ists. As early as the 1920s, Edward Tolman showed that rats could be
bred to be “maze-bright” or “maze-dull.” In the 1920s and 1930s, Barbara
Burks’s and Alice Leahy’s twin and adoption studies demonstrated a gene-
tic component to intelligence. In 1963, Nicki Erlenmeyer-Kimling and
Lissy Jarvik showed that IQ similarity varied positively with degree of ge-
netic relatedness. Several now classic twin and adoption studies by David
Rosenthal, Len Heston, and Irving Gottesman and James Shields indicated
a genetic basis for schizophrenia.47 The notion that a person’s behavioral
problems or learning disabilities were mostly learned from, and reinforced
by, his or her parents was weakening. The Behavior Genetics Association
came into being in 1970 in response to these trends, holding its first annual
meeting in Storrs, Connecticut, in 1971.48
By the late 1970s, the accumulating evidence against the behaviorist
perspective was becoming difficult to refute. Many individuals were
primed for a vision of human development that allowed for hereditary as
well as environmental effects. There was “just enough” doubt about the
primacy of the environment.49 It was this fluid academic climate of the
late 1970s and one investigator’s seizing of an opportunity to study a rare
separated twin pair that enabled the MISTRA to come into existence
when it did. The investigator was University of Minnesota Professor
Thomas J. Bouchard Jr.
8 INTRODUCTION

The MISTRA
The MISTRA came into being in 1979 and lasted for twenty years. It con-
ducted comprehensive psychological and medical assessments of eighty-
one MZA and fifty-six DZA twin pairs, with each assessment lasting for
an entire week. The study’s goal was to identify associations between differ-
ences in the twins’ life histories and the twins’ behavioral differences. The
MISTRA was not originally conceived from a developmental or longitudi-
nal perspective, but was exploratory in nature. Nevertheless, developmen-
tal data provided by the different pairs were revealing and relevant to the
ongoing longitudinal twin50 and adoption studies51 that were finding
genetic influences on behavioral and physical changes. Hypotheses were
not specified at the outset, except in a few instances.
Like Shields who had studied separated twins in the 1960s, the MIS-
TRA investigators took the view that the MZA twin design was a unique
and valuable research tool.52 It could generate ideas and predictions, and
provide results that could be used alongside results from other twin and
adoption studies. Findings of greater MZA than DZA twin similarity
would be consistent with, but not proof of, genetic effects on the traits
under study.53
I was associated with the MISTRA from 1982 to 1991, for three years
as a postdoctoral fellow and six years as Assistant Director of the Min-
nesota Center for Twin and Adoption Research. I scheduled twins’ assess-
ments, gathered psychological data on 86 of the 137 separated pairs, and
contributed to scientific reports. I revisited the University of Minnesota
in October 2009, April 2010, and October 2010, thirty years after the
MISTRA assessed the first reared-apart pair. The purpose of these visits
was to gather critical perspectives and personal reflections from the prin-
cipal MISTRA investigators. How did they view the study from a 2009–
2010 vantage point? How had the study impacted their area of expertise
in particular, and the psychological and medical fields in general?
Jay Samuels, professor of education at the University of Minnesota,
who developed the spelling and reading tests given to our separated twins,
concluded, “The timing was right. [The MISTRA] came about when be-
haviorism was losing popularity and cognitive psychology was gaining
rapidly. I would argue that in addition to [Noam] Chomsky’s attack on
language acquisition [by modeling and reinforcement], Bouchard’s work
had a major influence in bringing in a new paradigm.” Samuels’s views
were echoed by the over fifty collaborators and former students with
whom I spoke. Many also emphasized that the psychology department’s
academic networks made the University of Minnesota especially well
 INTRODUCTION 9

suited for a study of twins reared apart. Many of the collaborators had
professional connections to one another early in their careers.
Thomas J. Bouchard Jr. joined the University of Minnesota in 1969. He
was an industrial organizational psychologist who had earned his PhD
degree with Donald MacKinnon and Harrison Gough at the University
of California–Berkeley Institute for Personality Assessment and Research.
Irving I. Gottesman joined Minnesota’s faculty in 1966 after completing
his degree in that department in 1960. Gottesman had a rich history of
twin research involving studies of adolescent twins in Minnesota and
schizophrenic twins in the United Kingdom with James Shields that pre-
dated David Lykken’s twin research.54 Gottesman takes credit for “chang-
ing Bouchard’s career” from industrial organizational psychologist to
behavioral geneticist, and Bouchard agrees: “He put stuff in my mailbox—
I never took a course in behavioral genetics in my life.”55
Bouchard had considered starting a reared-together twin study before
the start of the MISTRA, but he did not want to encroach on his col-
league David Lykken’s area of research. Gottesman, however, suggested
to Bouchard that he launch a study of twins reared apart. Gottesman
pointed out that no one had done such investigations since Newman,
Freeman, and Holzinger’s 1937 study in Chicago, Shields’s 1962 study in
England, and Juel-Nielsen’s 1965 study in Denmark. Gottesman was
familiar with the Danish study and had authored the introduction to
Juel-Nielsen’s 1980 follow-up study of the twelve MZA twin pairs who
comprised the original 1965 sample. Gottesman wrote, “Identical twins
brought up apart are scarce, indeed, and their potential for stimulating
research ideas about the origins of normal and abnormal characteristics
is great.”56
Following discussions at the university’s faculty club with Gottesman,
Lykken, and psychologist Auke Tellegen, Bouchard had eventually given
up on the idea of a reared-apart twin study in Minnesota because there
was no obvious or systematic way to locate these rare separated sets.
Juel-Nielsen had obtained most of his twins through Denmark’s popula-
tion registry, but the United States did not have a comparable resource.
Finding separated twins through radio or television broadcasts, as New-
man and Shields had done, seemed impractical. Then Bouchard read about
the Jim twins in a newspaper article on February 20, 1979. His crumbling
maroon leather-bound calendar bears the following inscription for that
date: “Heard About Twins.” Consequently, the individual differences text-
book Bouchard was writing was never completed, and his work with
Gottesman on a book of readings, intended to update James Jenkins and
Donald Paterson’s 1961 individual differences volume,57 was discontinued.
10 INTRODUCTION

Instead, studying reared-apart twins became Bouchard’s first priority. As

Gottesman said, “We both got too busy.”

The Study in Context

Jim Lewis and Jim Springer are the separated twins whose February 1979
reunion was ultimately responsible for the launch of the MISTRA. The
press reported that both Ohio twins smoked Salems, chewed their finger-
nails, suffered from tension headaches, vacationed on the same Florida
beach, and had twice married women with the same first name (Linda
and Betty)—without ever having met.58 The only difference seemed to be
that Springer combed his hair down over his forehead while Lewis slicked
his back. This difference should have made it possible to tell them apart,
but it did not. Their angular features, unhurried speech, and laid-back
manner made differentiation nearly impossible for everyone outside their
families.
Twin researchers are accustomed to the matched personality and ap-
pearance of identical twins; however, the Jim twins’ similarities were es-
pecially striking because they were “MZAs”—monozygotic twins reared
apart—for thirty-nine years. Just by being themselves they could offer
unique insights into factors affecting variation in human traits. But until
the Jims could be observed and studied, the research team viewed some
of the similarities reported in the press with “healthy skepticism.”59
Not all MISTRA co-twins would be as similar as the Jims across all the
traits that we measured, but this fact does not discount genetic explana-
tions of behavioral and physical traits. Instead, it shows that common
genotypes can interact with different prenatal and postnatal environmen-
tal features to produce different outcomes. The MZA twins in some of
our pairs differed in IQ, religiosity, or body size, but the MZA twins as
a whole showed greater within-pair resemblance than the DZA twins,
indicating genetic effects.

Research Focus
Twin research operates at two levels. The first level includes studies di-
rected at the unique physical and behavioral aspects of twinship, such as
the effects of prenatal competition for nutrition and the experience of
growing up with a same-age sibling. Research has, for example, consid-
ered the effects of low birth weight on twins’ language development60
and the advisability of placing twins in the same or separate classrooms.61
The second level includes studies using twins as tools for estimating ge-
 INTRODUCTION 11

netic and environmental influences on physical and behavioral trait vari-

ation. This is accomplished by comparing resemblance between MZ and
DZ twin pairs, reared together and/or reared apart. Such studies provide
knowledge and insight for everyone, not just for twins and their families.
The MISTRA operated mostly at the second level, but some of our analy-
ses, such as our study of the twins’ social relationships discussed in Chap-
ter 12, operated partly at the first.
The MISTRA has had substantial impact on our understanding of in-
dividual differences in intelligence, personality, values, interests, religios-
ity, sexual orientation, mate choice, job satisfaction, sociality, and health.
It accomplished this in two ways: (1) by identifying a unique subject group
(MZA and DZA twins who were mostly adoptees) and (2) by adminis-
tering an extraordinarily comprehensive assessment battery to its partici-
pants. There is no other study of ordinary twins (or nontwins) for whom
such a wide range of psychological and medical findings is available.
Nearly every MISTRA personality paper was a “first” because a par-
ticular inventory or questionnaire had never been administered to twins
reared apart. This was important with respect to replicating prior findings
based on other kinships, but there were other forces driving our assess-
ment. More important was that researchers acquiring rare samples want
to study every behavioral and medical trait possible because great oppor-
tunities do not repeat themselves. We also had to stay current with devel-
opments in the field, such as assessing social attitudes and religiosity when
reared-together twin studies in those domains began appearing. Despite
our best efforts, some traits were never measured—Bouchard always re-
gretted that he didn’t record the twins’ favorite colors, or their shoe, hat,
and glove sizes.
Some results from the study were surprising—for example, the degree
of personality similarity was the same for MZA twins and MZ twins
reared together (MZTs). Furthermore, many of the findings showing ge-
netic influence on intelligence, special mental abilities, height, and weight
were comparable with and consistent with those from extant research on
twins reared together. How, then, is it possible to explain the constant
public and professional attention that the MISTRA attracted? First, the
findings were more compelling because the twins had grown up apart,
often without knowledge of the other’s existence. Second, the twins’ life
histories and surprise reunions appealed to the general public as well as
to academics and members of the press. People cared about these twins,
possibly because growing up apart runs counter to the general sense that
twins belong together. The parallels in the twins’ lives were fascinating,
even to those who clung to social views of human behavior.
12 INTRODUCTION

The consistently greater resemblance between the MZA than DZA

twins showed that genes affect virtually every measured trait. Reared-
apart twins also showed that environments matter, but not in the ways
most people thought they did. People are not passive recipients of ex-
perience but rather are active seekers and shapers of opportunities and
events that are compatible with their genes, assuming a normal range of
human environments. Some MZA twins, like Bridget and Dorothy from
Great Britain, ended up being very much alike despite being raised in
very different environments.62 One twin enjoyed rich educational oppor-
tunities in her home while her sister did not—still, their IQ scores were
quite close, and both twins were avid readers of historical novels: Doro-
thy read works by Catherine Cookson, whereas Bridget read works by
Caroline Marchant, Cookson’s pen name. Apparently, the less education-
ally advantaged twin actively sought new learning experiences, creating
her own environment from the sources available to her; for example, she
often visited the public library. Purposefully seeking activities and events
that are enjoyable is something that we all do. Thus, MZA co-twins’ fam-
ily environments can vary considerably without masking their genetically
influenced tendencies.63

Reared-Apart Twin Research

The MISTRA was not the first investigation of reared-apart twins. A 1937
Chicago study by Horatio Newman (biologist), Frank Freeman (educa-
tional psychologist), and Karl Holzinger (statistician) examined nineteen
MZA sets. Their work was followed by a 1962 study in Great Britain
by James Shields (psychologist) that included forty-four MZA sets and a
1965 study by Niels Juel-Nielsen (psychiatrist) in Odense, Denmark, that
examined twelve. Several additional case reports followed the Minnesota
study.64 Other investigations have since been launched in Sweden,65
Finland,66 and Japan,67 but none are as comprehensive as the MISTRA
in terms of the behaviors that were studied. And none of the other still
ongoing studies are as judicious in gathering the separated sets—some of
their twins, identified through registries, lived together until late child-
hood, whereas the MISTRA twins were separated much sooner.
The MISTRA was also the first study of its kind to include both MZA
and DZA twin pairs.68 Bouchard’s decision to use DZA twins as controls
was made in a very early memo to the “Twin Research Team.”69 This was
an important methodological improvement over past projects. Some early
investigators recruited only identical twins, so they may have unintention-
ally excluded the less-alike MZA sets. That is, prescreening of participants
 INTRODUCTION 13

may have led researchers to believe that MZA pairs with physical and/or
behavioral differences were DZA, so such pairs may have been over-
looked. In contrast, twins were accepted into the Minnesota study with-
out reference to their twin type, ensuring that the MZA pairs represented
a broader range of within-pair differences. In fact, twin type was not
confirmed by laboratory testing until the end of the assessment week or
after the twins had left Minnesota.

Compelling and Controversial

The MISTRA drew supporters and converts, but also critics. Gordon
Allen, then National Institute of Mental Health (NIMH) scientist and
president of the International Society of Twin Studies, told the New York
Times (December 9, 1979), “Such studies provide a lot of clues, if not
definitive answers, about all human beings.” In 1990, the late Daniel E.
Koshland Jr., a specialist in enzyme kinetics and former editor of the
prestigious journal Science, invited Bouchard to submit a paper reporting
the IQ findings from the study. Interesting sidelines to this story are re-
ported in Chapter 5.
At the same time, the idea of genetic influence on seemingly environ-
mentally driven traits such as intelligence, religious commitment, tradi-
tionalism, and work values rattled the status quo within the psychologi-
cal field. Critics outnumbered admirers, at least in terms of those publicly
venting their views. Frequent objections were that the twins’ matched
behaviors were explainable by similarities in their prenatal environments,
rearing homes, degree of contact prior to assessment, and/or physical at-
tractiveness rather than by their shared genes. Critics also charged that
the social contact between the twins while they were in Minnesota
increased the resemblance of their responses. Some believed that we fo-
cused mainly on behavioral similarities at the expense of differences.70
Another source of contention was the partial funding of the study by
the Pioneer Fund, an organization based in New York that has given
grants for research on race differences in intelligence. I will return to these
criticisms and to the study’s relationship with the Pioneer Fund in the final
chapter.
Others faultfinders decided that a large enough sample of reared-apart
twins would never be found. Some even accused the twins of going pub-
lic for financial advantage. In 1989, a talk show guest (and medical gene-
ticist) disparaged the MISTRA for publishing findings in the popular
press rather than in peer-reviewed journals. However, in 1989, ten years
after the study began, the MISTRA had published approximately twenty
14 INTRODUCTION

papers and book chapters, ten in the peer-reviewed scientific literature.

Because of the study’s comprehensiveness, the material was published
in a wide variety of medical and psychological journals, possibly giving
the appearance of no publications. The misperception that the MISTRA
findings were available only in the press or not at all has continued,
which is one reason why I decided to write this book.
The MISTRA continued until data were gathered on 137 reared-apart
pairs. The project has produced more than 150 scientific papers and book
chapters; a complete listing is provided on the Web site designed for this
book (www.DrNancySegalTwins.org). Most importantly, the MISTRA
stimulated dialogues among behavioral and medical science investiga-
tors, forcing everyone to reconsider their methods and interpretations.
Everyone’s understanding of behavioral phenomena benefited from this
process.

A Closer Look at the MISTRA

As I indicated, reports from the MISTRA are published in a variety of psy-
chological and medical journals. This fact attests to the comprehensive-
ness of the project. However, in pre-Internet days, and for individuals who
do not often conduct literature searches outside their fields, there have
been complaints regarding the lack of publication and impact.
A comprehensive overview of the scientific findings is needed to explain
how the study got started, who was involved, how the twins were identi-
fied, and how the study functioned. These are the goals of this book.
Some of the reared-apart twin pairs will also be introduced to illustrate
certain concepts and findings. Examining their life histories alongside the
scientific findings will make the data come alive for readers as it did for
me during the many weeks I spent with the twins.
Now that the data have been gathered, the papers published, and the
director retired—and the controversies quieted (somewhat)—it is time
to take a closer look at this important project. The material is presented
chronologically to capture events as they unfolded during the course of
the study, and to track repeated analyses as the sample size grew. This
form of presentation was also chosen to convey the excitement and con-
cerns that many of these events brought with them. However, this ap-
proach made choosing chapter titles tricky because unrelated topics are
sometimes grouped together. Surveying unrelated topics in the same chap-
ter may seem strange and somewhat disorganized, but it actually makes
perfect sense because the MISTRA gathered all kinds of data simultane-
ously. We took the twins for stress tests on the same day that they com-
pleted IQ tests, and we had them complete eye exams on the same day
 INTRODUCTION 15

that they filled in interest inventories. The various data were analyzed
when each investigator was ready and able to do so.
I joined the MISTRA in August 1982 as a postdoctoral fellow and
stayed for nine years. Working on the MISTRA and writing this book
came naturally to me because of my background in twin research. I
had completed my doctoral thesis at the University of Chicago in
1982, “Cooperation, Competition, and Altruism within Twin Sets: A Re-
appraisal,” and the MISTRA offered opportunities to reexamine some of
the ideas and findings generated by that work, using twins reared apart. I
am also a fraternal twin.
I have been given access to all files and documents associated with the
MISTRA. I have also interviewed nearly every investigator, many research
assistants, and some reared-apart twins who were part of this project.
Some twins’ names have been changed to preserve their identity.

The Twins
Many twins used their time in the Twin Cities to capture the fun they had
missed by not growing up together. For example, MZA twins Mark and
Jerry (both volunteer firefighters, hence their nickname “fireman twins”)
dunked their vitalogs (the activity monitors that continuously measure
heart rate and motion) in Coca-Cola after we had recorded their data.
The DZA twins Dewayne and Paul celebrated their first birthday together
in our laboratory at age seventy with party hats and favors.
Casual observations sometimes guided components of the assessment
schedule. Flirtatious exchanges between reunited opposite-sex co-twins
prompted us to add questions about sexual attraction to one’s co-twin.
Some reunited adult siblings as well as parents and their adult children
have experienced sexual feelings for one another, associated with having
been apart during children’s early years.71 I will return to this topic in the
discussion of the sexual life history study.
The more immediate rapport we observed among the MZA than the
DZA twins led to our development of a twin relationship survey. Associa-
tions between genetic overlap and social relatedness were timely topics
within the growing field of evolutionary psychology, and still are. More-
over, meeting one’s twin for the first time is an extraordinary life-changing
event. Many twins were overwhelmed with introductions to new in-laws,
nieces, nephews, and even parents. Their letters to Bouchard still move me:

I must tell you that it has been the most wonderful experience of my life to be
re-united with my sister and that when we met it was like a coming home.
(DZA female twins, met at age 56)
16 INTRODUCTION

You can imagine how amazed I was to learn [that I was a twin] 36 years
later . . . We have now had eleven happy months of getting to know each
other—and what happy months! (MZA female twins, met at age 36)

The reared-apart twins also learned a lot about their early circumstances
and medical life histories. They could finally understand that their me-
chanical aptitude or seasonal allergy had a partial genetic basis, which was
satisfying knowledge to adoptees without access to their biological past.
Most twins enjoyed each other’s company and stayed in touch, but even
those who did not felt gratified to have met their twin at last.
 CHAPTER ONE

The Jim Twins (February–March 1979)

O n February 19, 1979, the Lima News in Ohio reported the

reunion of thirty-nine-year-old identical twins Jim Lewis and
Jim Springer, separated at four weeks of age (Figure 1-1). The
twins, born to an unwed mother, had been placed in the Knoop Chil-
dren’s Home in Troy, Ohio, and had been adopted separately by two lo-
cal families. They grew up forty miles apart, Lewis in Lima and Springer
in Piqua.1
The couples who adopted the twins were Jess and Sarah Springer, and
Ernest and Lucille Lewis. Both families had been told that their child was
a twin whose brother had died. The source of this misinformation and
the reasons for it have never been determined. The Springers, who ad-
opted two weeks ahead of the Lewises, would have taken both twins had
they known both twins survived. In fact, several years later the Lewises
adopted identical twin boys, Larry and Gary, even though court officials
suggested that they take just one.2 Perhaps the court reasoned that rais-
ing one child would be easier for parents than raising two. By then, Lucille
Lewis knew that her son’s twin was alive, and she refused to separate
another set.
When Lewis was a toddler, his mother visited the Miami City, Ohio,
court regarding some adoption matters. A court officer blurted out that
the other family had also named their son Jim. Shocked by this informa-
tion, Lucille waited until her son Jim turned five to tell him he had a twin.
Over the years she encouraged him to find his brother, but he hesitated.
Then, shortly before his thirty-ninth birthday and for no apparent reason
“the time was right.” Lewis asked the Piqua court to contact his twin.
18 B O R N TO G E T H E R — R E A R E D A PA RT

Springer learned that he had a twin when he was eight, but he had
accepted the story that his brother had died. Then, at Jim Lewis’s urging,
the Piqua court told Jim Springer that his brother was alive and how to
contact him. Springer called Lewis and left a message with his call-back
telephone number. Lewis returned the call, and the twins met for the first
time.

Figure 1-1. The Jim twins: Jim Springer (left) and Jim Lewis. (Photo courtesy
Thomas J. Bouchard Jr. Unposed photographs taken in David Lykken’s
laboratory.)
 THE JIM TWINS 19

Three weeks after they met, Springer served as best man in the cere-
mony for his twin brother’s third marriage. Newspapers chronicled the
twins’ striking similarities, including their first names (Jim), their favorite
school subject (math), their dreaded school subject (spelling), their pre-
ferred vacation spot (Pas Grille Beach in Florida), their past occupations
(law enforcement), their hobbies (carpentry), and their sons’ first names
(James Alan and James Allan). The article about them was reprinted in
newspapers around the country, including the Minneapolis Star Tribune.
Jim and Jim were not the first separated twins to surface in the United
States since the publication of a series of case studies in 1949 by Barbara
Burks and Anne Roe.3 MZA twins Tony Milasi and Roger Brooks were
reunited in 1962, at age twenty-four, after Roger was mistaken for Tony
by a busboy in a Miami, Florida pancake house. Their reunion was cov-
ered in the press, and a book describing their lives appeared in 1969.4
When the book appeared, Bouchard had just joined the University of
Minnesota faculty and Lykken had not begun to study twins.
On February 20, 1979, the day that the Star Tribune article appeared,
University of Minnesota psychology graduate student Meg Keyes showed
the story to her professor, Thomas J. Bouchard Jr. She had taken Boucha-
rd’s course in individual differences in behavior, a class that included read-
ings about separated twins. A second copy of the Star Tribune article was
left in Bouchard’s mailbox by psychology professor Gail Peterson. Peter-
son appended a note saying, “I thought you would get a kick out of this.”
Meg, now Dr. Keyes of the Minnesota Twin Family Study, remembers
that Bouchard was “excited from the get-go—he moved fast getting funds,
staff, and tests together.” “I had to find the twins—I thought someone else
would go for the Jims, but no one did,” said Bouchard.
Every once in a while, a scientist encounters an unusually fascinating
research problem, issue, or situation that is irresistible. Such projects usu-
ally involve some degree of risk in terms of time, funding, and reputation,
but the process is exhilarating if the topic is right. Few people had the
opportunity to compare, firsthand, the behaviors of identical twins reared
apart from birth. Bouchard admitted that he had formed a “big plan” in
a few hours, but with no money.5
Bouchard telephoned the Associated Press in Cincinnati, Ohio, in an
attempt to locate the twins. He told the person on the other end that he
was prepared to “beg, borrow and steal” to support this case study. On
February 25, 1979, the New York Times ran the front-page story, “Iden-
tical Twins Seen as Offering New Clues in a Psychology Study; Searched
for Brother.” Bouchard was quoted as saying, “I’m going to beg, borrow
and steal and even use some of my own money if I have to. It is important
to study them immediately because now that they have gotten together
20 B O R N TO G E T H E R — R E A R E D A PA RT

they are, in a sense, contaminating one another.” The New York Times ap-
parently had obtained the story over the Associated Press wire; Bouchard
recalls hearing typing on the other end as he spoke, explaining why his
words were repeated.
Bouchard’s original letter of invitation sent to Jim Lewis and Jim Springer
introduced the investigators and provided an overview of activities the
twins would complete. The letter also explained that the project covered
airfare, food, lodging, some lost wages, and a modest honorarium to off-
set miscellaneous expenses. Bouchard also invited each pair to dinner one
evening.

The Core Research Team

Bouchard established a core research team consisting of psychologists
David T. Lykken (now deceased), Irving I. Gottesman, Auke Tellegen, and
psychiatrist Leonard L. Heston (Figures 1-2 and 1-3). Each had heard
about the Jim twins directly from Bouchard. At first, Lykken wondered if
studying just one pair was worth the trouble, but Bouchard told him,

Figure 1-2. Left to right: Leonard L. Heston, David T. Lykken, Niels

Juel-Nielsen, Elke D. Eckert, and Thomas J. Bouchard Jr. (Photo courtesy
Thomas J. Bouchard Jr. Photograph taken in May 1981 by a waiter in a
Minneapolis restaurant. The occasion was the visit by Dr. Niels Juel-Nielsen,
who had studied reared-apart twins in Denmark.)
 THE JIM TWINS 21

“You are my friend—humor me.”6 Lykken did and became intrigued.

Gottesman was immediately interested and suddenly confident that addi-
tional pairs could be found.
Heston felt that the Jim twins would make an “interesting case that was
worth doing,” but he doubted that the work would progress beyond the
single-study stage. His views changed once the group had assessed seven or
eight pairs, making him aware that the study had “real potential.” Heston
was also responsible for inviting University of Minnesota psychiatrist Elke
D. Eckert to become part of the core team. Tellegen spoke for both Lykken
and himself when he said that both thought studying the Jim twins would
be a “one-time opportunity” and that it required sophisticated idiographic
methodology—standard psychological procedures were not appropriate
for research on just one set. Initially, he and Lykken tried to develop statis-
tical techniques that would replace multiple subjects with multiple obser-
vations, but when the sample size grew this proved unnecessary.
The University of Minnesota’s psychology department had a long
history of individual differences research, much of it with a behavioral-
genetic bent. Therefore, studying development by way of twins reared
apart was natural for that time and place. Following Gottesman’s 1957
twin study dissertation, conventional identical-fraternal twin comparisons
had been ongoing since 1970 under the direction of David Lykken, who
developed the Minnesota Twin Registry between 1983 and 1990. The

Figure 1-3. Left to right: Irving I. Gottesman, Auke Tellegen, and MZA twin
Jim Lewis. (Photo courtesy Thomas J. Bouchard Jr. Photograph taken by a
media center staff member, March 1979.)
22 B O R N TO G E T H E R — R E A R E D A PA RT

Minnesota Twin Registry includes information on 901 twin pairs born

between 1904 and 1936, 4,307 twin pairs born between 1936 and 1955,
and 391 male twin pairs born between 1961 and 1964; see Twin Re-
search 2002. Bouchard called his department’s atmosphere “fabulous—
you cannot create it, it must exist.”
Lykken became Bouchard’s most frequent coauthor—according to
Bouchard, “We never really disagreed.” Gottesman was famous for his 1960s
twin studies of schizophrenia conducted with James Shields (who con-
ducted the 1962 British reared-apart twin study), and for having adminis-
tered the Minnesota Multiphasic Personality Inventory (MMPI) to adoles-
cent twins. Heston had published a seminal 1966 adoption study of schizo-
phrenia and was the father of identical twin daughters, Barbara and Ardis.
Heston and Gottesman had become acquainted in the mid-1960s when
Heston joined the Medical Research Council (MRC) Unit in Psychiatric Ge-
netics in London. Gottesman was already there and had encouraged
Heston to come.7 Eckert, a specialist in eating disorders, found the idea
of reared-apart twins fascinating: “It was a chance to look at genetic and
environmental influences on behavior, something I hadn’t studied much.”
Tellegen had recently developed the Differential Personality Question-
naire (now the Multidimensional Personality Questionnaire).8 The Multi-
dimensional Personality Questionnaire is an eleven-scale form for assess-
ing individual variation in personality. Tellegen had worked collaboratively
with Lykken on psychophysiological twin studies involving more than
five hundred pairs. Tellegen also credited a supportive department chair,
John (Jack) Darley Sr. for facilitating study of the Jim twins.9 Matt Mc-
Gue, now Regents Professor in the psychology department at Minnesota,
was a graduate student at the time, in charge of setting up the twins’ in-
formation processing tasks. “When I came back it was very exciting here.
The MISTRA put us on the map.”
Another factor favorable to the study’s success was that the univer-
sity’s psychology department and medical school were located on the
same campus, facilitating assessment of behavioral and health-related
traits. For the first time, reared-apart twins were undergoing cardiac,
pulmonary, dental, periodontal, ophthalmologic, and endocrinological
testing, as well as psychological assessment. Some medical tests, such as
the electrocardiogram (ECG), were administered from the beginning; oth-
ers, such as the periodontal examination, were added later.

The Assessment
Every colleague acknowledged that Bouchard’s extraordinary ability to
generate enthusiasm was key to starting the study and keeping it going.
 THE JIM TWINS 23

When the study began, Bouchard was investigating factors affecting gen-
eral intelligence and special mental abilities. His 1970s publications in-
cluded studies of spatial ability, problem-solving strategies, and intelligence
testing. He quickly put together a comprehensive assessment schedule that
required four days for the twins to complete. The Jim twins and the pairs
who followed became the first separated sets to complete psychological,
psychophysiological, psychomotor, life stress, and sexual life history pro-
tocols; previous reared-apart twin studies focused on general intelligence
and personality traits.
The research program could not be completed in four days, so the Jim
twins returned to Minnesota five months later. The assessment schedule
eventually expanded to fill five full work days, plus one evening and two
half-days.
An important activity of the first assessment day was a series of un-
posed photographs of the twins taken alone, then together. The MZA
twins’ typically similar body postures and the DZA twins’ typically dis-
similar ones were unexpected findings, suggesting that there is genetic
influence on these behaviors. We would also see genetic effects on body
movement and gestures suggested by the MZA twins during separate vid-
eotaped interviews held later in the week. Bouchard recalls a pair of Brit-
ish women who continually twirled a strand of pearls around their necks
as they spoke. Details of all the tests and inventories that were adminis-
tered are provided in subsequent chapters and in this book’s Web site.
The medical tests began on the first full study day and continued over
the next three days. Dr. Naip Tuna, then Director of Electrocardiography
and Non-Invasive Cardiology, first learned about the Jim twins from
Bouchard. Tuna was “very interested” but uncertain as to the importance
of heredity in adult cardiac functioning. “I thought genetics played a greater
role in children’s heart disorders,” he admitted. Nevertheless, Tuna was
intrigued and assumed the role of primary physician and cardiologist for
the study. Eckert and Heston administered the twins’ psychiatric inter-
views and medical life histories until clinical psychologist Will Grove re-
placed Heston, who joined the University of Washington in 1990.
Other physicians saw the reared-apart twins as a unique opportunity
to conduct research in their own fields. Dr. Michael Till, a pediatric den-
tist, read about the project in the university’s newspaper, the Minnesota
Daily. Till brought the study to the attention of his colleagues in adult
dentistry, but there was little interest. He attributes their cool response to
patient overloads, heavy teaching responsibilities, and emphasis on restor-
ative dental procedures. But Till was captivated. He contacted Bouchard,
whom he did not know at that time, and the addition of a dental compo-
nent to the project was decided over coffee. Till’s only concern was that
24 B O R N TO G E T H E R — R E A R E D A PA RT

the walls of his examining room were decorated for children and the ex-
amination chairs were on the small side.
Thanks to my own periodontal problems, a periodontal evaluation
was added to the project in 1986. While having my gums probed at the
university’s clinic, Dr. Mark Herzberg asked me about my twin research.
It was hard to talk, but I managed to describe what we were doing and
finding. He immediately recognized the research possibilities for his own
specialty and put us in touch with his colleagues Bruce Pihlstrom and
Bryan Michalowicz. During my 2009 visit to Minneapolis, Michalowicz
(like McGue) told me that the MISTRA had put the University of Min-
nesota’s periodontal department “on the map” in terms of defining gene-
tic risk for periodontitis.
Informal clinical interviews were offered to the first twenty-three twin
pairs as either a respite from the rigorous assessment schedule and/or an
opportunity to discuss personal life history issues. These interviews were
administered by Lloyd Sines, then professor in the university’s psychiatry
department and a colleague of Heston and Eckert, who brought him into
the study. I have never met Sines, but I have seen his name on our early
files. Over the course of a long telephone conversation in May 2011, I
heard a different take on what the reared-apart twins might tell us.
“I had no mission,” Sines told me. “I only had to talk to [the twins] in
a friendly, professional manner.” But there was more. “I did not share
the highly biological perspective of the others. I may have expressed
misgivings to them. Part of my motivation, in addition to the intrigue and
importance of the endeavor, was to do something different with the
twins—to approach them with a more clinical, subjective point of view.”
Sines recalled the well-known MZA twins Oskar and Jack, raised Catho-
lic in Nazi Germany and Jewish in Trinidad, respectively, and speculated
about the effects of their rearing situations on their personalities and
politics. “[Oskar’s opinion on World War II] vindicated my involvement
in the study—I knew I had found something that the biological approach
had missed.” I will return to our findings on Oskar and Jack in later chap-
ters of this book. Sines’s interviews were discontinued (as were some other
tests and activities) once the assessment schedule grew longer, but Bouchard
has always said that the twins appreciated them.
The University of Minnesota is situated in a large city that offers ac-
cess to many types of specialists. The Minneapolis War Memorial Blood
Bank, directed by Dr. Herbert F. Polesky, was just downtown.10 Polesky
analyzed the twins’ blood group profiles, classifying them as MZA or
DZA. (I will say more about this process in Chapter 2.) In 1985, we dis-
covered Tourette syndrome in all three members of an MZA/DZA triplet
 THE JIM TWINS 25

set. Dr. Maurice W. Dysken, a psychiatrist at the Minneapolis Veteran’s

Administration Hospital, immediately arranged a sleep study for one of
the triplets. This interesting case is described in Chapter 7.

Ultimately, Bouchard did not have to beg, borrow, or steal to bring the
Jim twins to Minnesota. He was serving on the university’s graduate
school research advisory committee when the Jim twins were found, and
he knew that this body dispensed faculty research grants, sometimes on an
emergency basis. He secured $6,500 within a couple of days, bringing the
Jim twins to Minnesota on March 11, 1979, just three weeks after they
were found. The funds were not intended to cover spouses, but the Jims
insisted on coming to Minnesota with their wives. This posed a financial
burden, but it also enabled comparative normative data to be gathered
on nontwins (and nonadoptees) by having spouses complete some of the
same tests as the twins. Analyses of spouse similarity (assortative mating)
could be performed on various traits of interest. As the study progressed,
it became clear that having spouses, children, or friends travel with the
twins provided an added incentive for the twins to participate. According
to Bouchard, a family member’s presence may have yielded a more repre-
sentative sample; otherwise, only the more adventuresome twins might
have participated.
In the wake of the publicity surrounding the Jim twins, Bouchard’s
department chair, John Darley, told him, “You will find more twins, so
you will need more grants.” Darley telephoned the Spencer Foundation
in Chicago, an organization that supports research on educational issues
and values. The next meeting of the awards committee was ten days
away, but Bouchard submitted a grant and received $32,000. “By then
we had a few more pairs, and things rolled.” Bouchard recalls his encoun-
ter with his chair as one of several unplanned events that moved the study
along. “Jack Darley was paying attention to what was happening in the
department. His actions became a model for me when I was chair.” At
about this time, Bouchard had also secured a $10,000 grant from the
publishing house Harcourt Brace Jovanovich. Both the Spencer and Har-
court funds covered the research period from spring 1979 to summer
1980. The funds covered the twins’ travel costs, some medical tests, the
assistants’ salaries, and other research-related expenses.

The Jim Twins

Excitement peaked in the psychology department when the Jim twins first
visited the university. Everyone involved in the research had undergone
26 B O R N TO G E T H E R — R E A R E D A PA RT

hours of preparation for what he or she thought would be a one-time

case study. Graduate students, Meg Keyes and Susan Resnick were re-
sponsible for most of the psychological data collection. Francie Gabbay
(now a research professor at the Uniformed Services University in
Bethesda, Maryland) was a visiting graduate student in David Lykken’s
psychophysiological laboratory. Bruce Hanson (now an internist at the
Minneapolis Veteran’s Administration Hospital) was a premedical stu-
dent at the time; he did not gather information directly from the twins
but helped process allergy data in a laboratory run by Dr. Malcolm Blu-
menthal. Blumenthal was a key figure in the twins’ allergy testing proto-
col. Hanson participated later in analyses of the twins’ ECG data.
Meg Keyes was struck by the Jims’ similar personalities. “They were
both so patient and kind, and also very serious.” Susan Resnick was in
the observation booth during the videotaped interviews: “Their gestures,
the way they crossed their legs, were remarkably alike. This was all new
to us at the time.” When the twins returned to Minnesota in August,
Lewis’s younger identical twin brothers (Gary and Larry) and their wives,
and Springer’s younger brother (also Larry) and his wife completed an
abbreviated version of the test battery. Gathering data from the twins’
unrelated siblings (with whom they had shared environments but not
genes) was another way to assess the relative effects of genetic and en-
vironmental influences on behavior. In fact, as the study progressed, a
number of the twins’ unrelated brothers and sisters were recruited for
participation.
Keyes recalls that the Jims, who were raised apart, seemed as much
alike as Gary and Larry who had been raised together. Furthermore, the
Jims resembled each other both physically and behaviorally more than
they resembled their adoptive siblings. As the data accumulated, these ca-
sual observations were confirmed; empirical quantitative analyses showed
very modest shared environmental effects on many traits. Behavioral simi-
larity appeared to be explained more by common genes than by common
environments.
Most of the MISTRA data were gathered by the six primary investiga-
tors, University of Minnesota physicians, and graduate student assistants.
The exception was the Wechsler Adult Intelligence Scale (WAIS) that was
administered to twins at the same time by separate examiners hired through
the university’s neuropsychological clinic. (The rationale for this proce-
dure is explained in Chapter 5.)
Virtually every colleague found studying the Jims and the other reared-
apart twins to be “fun” and even “exhilarating.” Nevertheless, it was rec-
ognized that the twins’ value lay not in providing definitive answers but in
 THE JIM TWINS 27

raising new ideas about the origins of human behavior. For example, both
of the thirty-nine-year-old Jims had gained ten pounds at the same time
for no apparent reason. Heston found this remarkable because, while
weight gain often occurs between the ages of thirty-four and forty, the
Jims’ weight change had occurred simultaneously. He suggested that what
appeared to be an “improbable coincidence” might reflect genetic influ-
ence on physical changes associated with aging.11 Heston was also capti-
vated by the Jims’ similar mixed headache syndrome (a tension headache
that turns into a migraine), which first affected both twins in their teens.
Lots of people have headaches, but in separate interviews the Jim twins
used the same words and concepts to describe their disability and pain.
Springer said, “It feels like somebody’s hitting you in the back of the neck
with a two-by-four,” and Lewis noted that “It’s centered in the back of the
neck, and it damn near knocks me out sometimes.”12
The Jim twins’ striking resemblance created intense interest among the
media, colleagues, and, of course, the Minnesota investigators. Members
of the research team said many times that they did not expect the twins
to be as similar as they were. Relying on statistical findings, Bouchard
discovered that on most of the twenty-three vocational test categories, the
Jims were as alike as the same person taking the same test twice.13 And
their California Psychological Inventory scale scores were so alike that
one twin’s profile could be superimposed upon the other’s. The twins’
performance (nonverbal) IQ scores were just one point apart, although
their full scale scores differed by nine points. Bouchard himself told the
New York Times that “If someone else brought the material to me and
said, ‘This is what I’ve got,’ I’d say I didn’t believe it.”14
The twins’ physical measures and other medical characteristics were
also highly matched. Lewis weighed 154.90 pounds and was 70.90 inches
tall, while Springer weighed 154.59 pounds and was 71.40 inches tall.
Springer had had two heart attacks, and Lewis had been hospitalized for
a suspected heart problem. The average male age for having a heart at-
tack is sixty-six, so the thirty-nine-year-old twins’ heart problems were
unusual.15 Their biological family’s health history may have placed them
at higher risk, but this information was unknown. The twins’ smoking
habits and visual acuity were also alike.
There is something to be said for being present in order to appreciate
and absorb what was taking place during the reared-apart twin assess-
ments. Bouchard once explained that the MISTRA actually took place
on two levels, statistical and anecdotal. The statistical level involved com-
paring the magnitude of correlations and concordance measures between
the MZA and DZA twin pairs. Such data are gathered objectively,
28 B O R N TO G E T H E R — R E A R E D A PA RT

systematically, and uniformly across sets. The anecdotal level involved

examiner observations and participant reflections specific to given pairs.
Observation-based information was revealed in a casual way over the
course of each study week, often during lunches and coffee breaks.
James Shields, author of the 1962 reared-apart twin study, also ac-
knowledged these two approaches, calling them biometrical and case
history, respectively.16
Biographical sketches, appended to the three reared-apart twin studies
that preceded the MISTRA, included descriptive data on the twins’ rear-
ing circumstances, personality traits, and social relations.17 This informa-
tion was a useful complement to the quantitative outcomes. Anecdotal
findings, so striking to those who observe them directly, are rarely wel-
come in professional journals, but they can offer fresh ideas about behav-
ioral phenomena. We saw beer cans held with pinky fingers underneath
for support, heard about hand washing before and after using the toilet,
and observed quarter pieces of toast left uneaten on two plates. No one
suggests that specific genes are responsible for these behaviors, but gene-
tic effects might underlie positioning of the bones, tendencies toward
tidiness, and feelings of satiety or restraint.
Juel-Nielsen, investigator of the Danish reared-apart twin study, vis-
ited the University of Minnesota in 1981. Bouchard specifically asked
him if he had observed unusual habits or quirks in his MZA twin pairs.
He said that he had, but he had not published the information because it
might have detracted from his quantitative findings. He also worried that
no one would believe it.
Before the Jim twins arrived, Bouchard had wondered as well if the
newspapers had exaggerated the twins’ similarities. Once they left, he con-
cluded that they had not.

Jim Twins: The Aftermath

The original plan was to conduct a single clinical case study of the Jim
twins. However, the stream of news reports and television shows kept
the Jim twins in the public imagination. For many, reared-apart identical
twins who end up alike despite their different backgrounds pose a great
challenge to explanations of behavior. The Jim twins’ stories of separation
and reunion were also poignant and moving—they both spoke of empti-
ness in their lives that disappeared after meeting one another. Both twins
agreed that finding each other was the best thing that had ever happened
to him. Eight years later, both twins characterized their relationship as
“closer than best friends.”
 THE JIM TWINS 29

Bouchard began receiving letters and calls from reared-apart twins

across the country and around the world, due almost exclusively to the
media’s response to the Jim twins. Bouchard relished the publicity be-
cause he might have spent millions of dollars on recruiting new separated
sets with little yield, but the press was far-reaching. Many people, not
just academics, knew about the study. McGue agreed: “Bouchard used
the anecdotes as a way to find twins. IQ correlations just do not capti-
vate people.”
The single case study plan was revised to allow for the study of several
separated twin sets. The sample grew to fifteen by February 1980, and
the MISTRA formally came into being. By early 1983, the MISTRA had
examined more than fifty pairs. According to Bouchard, “By then I was
convinced that I had something. The genetic effects were sufficiently strong,
showing that developmental studies reporting correlations between par-
ents and children in support of environmental effects were misleading.”
Parents and children share both their genes and their environments, so
correlations between them do not necessarily imply environmental influ-
ence on behavior. Sports-minded parents may offer swimming and tennis
lessons to children, but children may also inherit genetic predispositions
for good athletic performance.
The MISTRA lasted for twenty years, but it might have lasted longer.18
Just as some events favored its inception and early progress, continuing
controversies and a few disappointing—even absurd—grant reviews
worked against it. For example, one grant reviewer suggested that the
investigators “spend their money hiring private detectives” to verify the
backgrounds of twins Dorothy and Bridget, who arrived in Minneapolis
wearing “seven rings on seven fingers.” Both twins, in fact, were wearing
seven rings, three bracelets, and a watch when Bouchard met them at the
airport. Their long slender finders and hands, plus the ability to afford
such adornments, may partly explain their attraction to jewelry (Figure
1-4).19 (Dorothy and Bridget also were the twins who enjoyed books by
the same author.) Another grant reviewer proposed a “feasibility study”
to secure a sample of 150 MZA and 150 DZA twin pairs. Reared-apart
twins are rare—in 1981, about twenty pairs had been studied, and
Bouchard’s grant proposal called for assessing seven additional sets.
There were also laudatory reviews—“The topic of this proposal is
certainly one of the most important issues in psychology,” and “[the twin
sample] provides the strongest methodological basis for estimating the
form and extent of environmental influence on given dependent vari-
ables.” The clash among reviewers hurt the study’s federal funding pros-
pects, causing the research team to rely on private sources.
30 B O R N TO G E T H E R — R E A R E D A PA RT

Looking back over the history of reared-apart twin studies, it is clear

that Bouchard’s decision to include DZA twin pairs was correct. (Shields’s
1962 study had included limited data on eleven DZA twin pairs.) Not
one DZA twin set rivaled the Jims as the most alike pair, although there
were strong candidates among the MZAs. But even if an exceptional DZA
set had shown the Jims’ extraordinary resemblance, it would not have al-
tered the study’s conclusions. The gene-environment balance is best gauged
by comparing as many MZA and DZA twin pairs as possible—only then
can “informed weights” be assigned to the various sources of influence.20
Single sets cannot provide definitive answers, not even the Jims.
The Jim twins triggered a chain of events, including newspaper cover-
age, televised interviews, an invitation to Bouchard to address the Amer-
ican Adoption Congress and, consequently, the identification of new
reared-apart twin pairs. Bouchard’s daughter Elizabeth recalled that “that
first summer was heady, with all the talk of possibilities of where this re-
search might go.” Bouchard claims “no responsibility” for this, seeing the
process as serendipity.21
If the Jim twins had not been as alike as they were, the MISTRA may
have gone in a different direction—or in no direction at all. Not all MZA

Figure 1-4. Reared-apart British twins’ long, slender hands and fingers,
showing their shared taste for jewelry. (Photo courtesy Thomas J. Bouchard Jr.
Photograph taken in David Lykken’s laboratory.)
 THE JIM TWINS 31

twin pairs were as alike as the Jims—if one of these pairs had reunited
first, there might have been less media attention, and fewer separated sets
would have surfaced. It was fortunate that the Jims were as similar as
they were. Without them, science and society might have been deprived
of the wealth of data gathered from the 137 reunited sets and the over
150 publications that resulted. Others are still being written.
 CHAPTER T WO

15,000 Questions × 137 Pairs

T he MISTRA evolved rapidly following the Jim twins’ 1979

assessment, especially with respect to methodology—how we
found the twins, what we asked them to do, and where we had
them do it. This chapter provides detailed descriptions of the partici-
pants, test schedule, genetic terms, and statistical concepts needed to un-
derstand the inner workings and quantitative findings of the project.
From the start, each twin completed protocols such as a life history
interview (covering their separation, reunion, contact, residential history,
and rearing family composition), personality and interest question-
naires, and the Wechsler IQ test. We added tests and questionnaires (such
as the Woodcock-Johnson Reading Mastery Test and the Neuroticism-
Extraversion-Openness Inventory) and eliminated some procedures in the
interest of time. The changes were often prompted by new twin-based
findings in the scientific literature. We added the Public Opinion Inven-
tory and the Wilson-Patterson Conservatism Scale in 1986 when Nick
Martin and his team published a provocative reared-together twin study
showing genetic effects on radicalism, tough-mindedness, and conserva-
tism, behaviors measured by those instruments.1 Thinking at that time
was that social attitudes were mostly due to family influences, so these
new findings could completely revise our understanding of the origins of
people’s opinions. It was important for us to study attitudes and opinions
in reared-apart twins to see if we would also find genetic influence using a
different twin method.
We estimated that the twins completed 15,000 questions during the
study week. This many questions completed by a modest size sample raises
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 33

the possibility of obtaining significant results by chance alone. An impor-

tant methodological feature of the study was including more than one
test of many of the traits we studied. This enabled within-study replica-
tions of our analyses to see if the results were consistent. Furthermore, by
using questionnaires and inventories administered previously to reared-
together twins we could compare our findings with those from other stud-
ies, and we found very consistent patterns with some exceptions.

Participant Identification
Identifying participants for the MISTRA was almost as exciting as the
study itself. We mostly learned about reunited twin pairs from news-
papers, but we also received tips from colleagues, family members, and
acquaintances. We studied DZA as well as MZA pairs to avoid the criti-
cism of possibly attracting the most similar sets. We studied DZA opposite-
sex as well as same-sex pairs to determine if male-female pairs were less
alike on some traits than their same-sex counterparts.
It was more exciting to find twins by oneself than from our other
sources. On December 6, 1979, when I was a graduate student, I summa-
rized my dissertation research on twins for the Children’s Home and Aid
Society of Illinois and mentioned the MISTRA. A social worker con-
fessed to me that her agency had separated female twins, Barbara and
Judy, thirty-one years earlier, yet neither twin knew they were part of a
set. The social workers believed that separating the twins was advisable
because Barbara (the fussy twin) might not receive the attention she
needed if placed with Judy (the engaging twin). I pushed for information
about the twins, but the society suggested that I have Bouchard contact
the Executive Director, Edwin Millard. Bouchard did, and nine months
later, in September 1980, Barbara and Judy were reunited in Minneapo-
lis at age thirty-two. However, the process of finding these twins was
complex.
When I spoke with the society’s former research director, Joan
DiLeonardi, in April 2010, she recalled the case immediately. “If my stu-
dent and I had decided not to do it, it would never have happened. The
1940s were the ‘heydays of secrecy in adoption,’ but by the 1980s things
had changed.” Fortunately, DiLeonardi’s field student, Ellen Smith, was
fascinated by the possibility of bringing twins together, and searched hun-
dreds of Chicago-area records. When the twins were finally located,
DiLeonardi placed several calls to each one, first inquiring about their
respective adoption experiences. In her second call to Barbara, DiLeonardi
told her, “You have a special sibling.” “Oh?” “You have an identical twin.”
34 B O R N TO G E T H E R — R E A R E D A PA RT

Barbara screamed with excitement. It was no different for Judy, who

said, “I will call her right now!”
For a while, Barbara and Judy held the status of “best MISTRA pair”
because they had had no contact prior to assessment and were previously
unaware of having been born a twin. Bouchard, who witnessed the
twins’ reunion, was captivated by them—“It was as if they had known
one other all their lives.” Bouchard recalled that “not a nanosecond went
by between the time that one twin spoke and the other twin answered—
I’ve been married thirty years and I don’t do this with my wife!” DiLeon-
ardi was also at the airport, having flown to Minneapolis to witness the
twins’ reunion. She recalled that Bouchard immediately examined the
shapes of their ears, which typically match in MZ twins, and believed
they were identical. She also said that by the time they arrived at the
hotel the twins were finishing each other’s sentences. “I have identical
twin cousins who did this, and I thought it was because they had grown
up together.”
Another wonderful moment concerned opposite-sex twins, Iris Johns and
Aro Campbell, New Zealanders who met at age seventy-five (Figure 2-1).

Figure 2-1. Aro Campbell (left) and Iris Johns. The twins, separated at six
weeks of age, met for the first time at age seventy-five. (Photo credit Nancy
L. Segal.)
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 35

I learned about them accidentally in April 1991 when I checked the

Guinness Book of World Records to see if one of our DZA pairs, De-
wayne and Paul, who were reunited at sixty-nine, were recognized as the
longest separated pair in the world. But Dewayne and Paul were the
longest separated “American pair,” whereas Iris and Aro set the world’s
record. Bouchard did not think we would find the twins, but after two
phone calls made within five minutes I proved him wrong.
Iris and Aro were delighted to come to Minnesota. Each had grown up
believing that the other twin had died, but Iris had harbored doubts. She
appeared on Missing, a TVNZ television program for people searching
for family members. Someone who knew Aro and had seen the show
was responsible for bringing the twins together. “We had to wipe away so
many tears that night,” Iris said.2

Introducing the Twins to the MISTRA

Once we located twins, our next step was to call them on the telephone
to invite them to campus. I loved making these calls because the twins
often provided interesting information that the press hadn’t reported.
Ann Blandin and Barbara Parker were such a pair. Three of us found
them simultaneously—I called Bouchard from New York when I saw the
story in a New York newspaper during a visit home; Bill Fine, a friend of
Bouchard’s, had seen the story in a San Francisco newspaper; and jour-
nalist Lew Cope from the Minneapolis Star Tribune had also sent word.
Barbara and Ann each described herself as a “picky bedspread picker-
outer”—so when Ann peeked into Barbara’s bedroom for the first time
she was astonished to see that they had chosen the same unusual bed-
spreads. Perhaps it was the color, the texture, or the price of the spread
that both twins found appealing.
Ideally, researchers would want to find separated twin sets in national
population registries that record all occurrences of multiple births and
rearing circumstances. Such resources are available in the Scandinavian
countries, which allowed Juel-Nielsen to assess the complete population
of reared-apart Danish twins that met his criteria.3 Population-based
twin registries are also available in large regions of China4 and in other
nations, but a national twin registry does not exist in the United States.
However, localized registries, such as the Minnesota Twin Registry at the
University of Minnesota and the Mid-Atlantic Twin Registry at Virginia
Commonwealth University have been created by investigators to facili-
tate their research. Other more specialized rosters include the Vietnam
Era Twin Registry, the Chronic Fatigue Twin Registry, and the Interna-
tional Registry of HIV-Exposed Twins.
36 B O R N TO G E T H E R — R E A R E D A PA RT

Lacking a national registry, the MISTRA studied a collection of cases

because our reared-apart twins surfaced in many ways and at any time
(Table 2-1). The twins themselves and the people who heard about them
contacted us because they knew about the MISTRA largely through the
media attention the study had attracted. Our professional colleagues also
knew about the study because of publications, presentations, and the
usual academic word of mouth. We also identified twins through adop-
tion registries, social service agencies, and search groups. The majority of
pairs were identified by self-referral, but most of the twins learned about
the study through media reports.
We studied 137 reared-apart twin pairs between March 1979 and
April 1999. Close to seven new pairs, on average, were studied each year
that the study was ongoing, and forty-four pairs returned for follow-up
study beginning in 1987. Table 2-2 shows that the number of new twin
pairs was highest during the first four years of the study, then tapered off
somewhat as the study progressed. Most twins accepted our invitation to
participate. The fewer numbers of reared-apart pairs studied after 1994
may have reflected the dwindling of such pairs as well as our limited
funding. However, after the MISTRA ended in 1999 several new twin
pairs, separated mostly because their families could not care for one
or  both of them, were identified. It is likely that better search options,
many available on the Internet, allowed these twins to meet. The actual
numbers of reared-apart twins may not have declined as much as we had
thought.
That twin registries were unavailable does not mean that the group
of twins we studied was unrepresentative of twins generally—in fact,
the reared-apart twins resembled ordinary reared-together twins in many
ways. Table 4 in our 1990 Science paper, discussed in Chapter 5, shows

Table 2-1. Sources of Reared-Apart Twins from the MISTRA

Source Number of Pairs (N = 137) Percentage of Total

Media reports 11 8.0

Adoption services/Search groups 26 19.0
Referrals from colleagues 26 19.0
Self-referrals 50 36.5
Previous twin participants 16 11.7
Other 8 5.8

Note: These values reflect the most direct source of information. The majority of
self-referrals and referrals from colleagues and previous participants were in response to
media coverage of the study. Other sources included the twins’ attorney, the Guinness
Book of World Records, and the interested public.
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 37

Table 2-2. Reared Apart Twin Pairs Studied Each Year

Month/Year Pairs (N = 137)

3/1979–2/1980 14
3/1980–2/1981 12
3/1981–2/1982 15
3/1982–2/1983 10
3/1983–2/1984 3
3/1984–2/1985 10
3/1985–2/1986 8
3/1986–2/1987 13
3/1987–2/1988 4
3/1988–2/1989 9
3/1989–2/1990 4
3/1990–2/1991 6
3/1991–2/1992 7
3/1992–2/1993 4
3/1993–2/1994 6
3/1994–2/1995 3
3/1995–2/1996 1
3/1996–2/1997 2
3/1997–2/1998 4
3/1998–4/1999 2

that MZA and MZT twins display about the same degree of resemblance
on most measured physiological and psychological traits. For conve-
nience, however, I will refer to our twins as a sample (a representative
subset of the members of a population).

Separation and Contact

In order to participate in the MISTRA, twins had to have been separated
by four years of age and have spent their formative years apart. Several
pairs met as children, but most met as adults, and several met for the first
time in Minneapolis. Some contact did not disqualify twins from partici-
pating; in fact, the variable amounts of contact between co-twins pro-
vided opportunities to evaluate associations between degree of contact
and similarity in intelligence, personality, and other behaviors. It turned
out that contact was not associated with resemblance between the twins
across virtually all the behaviors that we studied. For example, time spent
apart until the first reunion correlated only 0.08 with the within-pair (co-
twin) difference in IQ, showing that twins who met at earlier ages were
not more alike in intelligence than were twins who met at later ages.
38 B O R N TO G E T H E R — R E A R E D A PA RT

The twins’ mean age at separation was 218.21 days (standard devia-
tion = 343.59) and ranged from 0.00 to 1,644 days. Expressed in years, the
mean age at separation was 0.60 years (standard deviation = 0.94) and
ranged from 0.00 to 4.50 years.

Determining Twin Type

Determining the zygosity (twin type) of most pairs was accomplished by
the comparative study of blood group systems. Blood samples were drawn
from each twin during the middle of the study week and forwarded
to the Minneapolis War Memorial Blood Bank. A standard protocol in-
volved testing for the following blood group systems: ABO, Rh (rhesus),
MNSs, Kell; Lewis, Duffy, Kidd, P; serum proteins: Gc (group specific
factor), Tf (transferrin), Hp (haptoglobin), AK (adenylate deaminase),
and 6-PGD (6-phosphogluconate dehydrogenase); and red cell enzymes:
Cp (ceruloplasmin), PGM1 (phosphoglucomutase), AcP (acid phospha-
tase), EsD (esterase-D), and GLO (glyoxalase I). Based on their identity
across these factors, twins were classified as MZ with over 99 percent
certainty, and discordance in any of these blood factors indicated that the
twins were DZ.
The specific blood groups, proteins, and enzymes that were assayed
changed over the years. Current twin studies decide zygosity based on
within-pair concordance or discordance across 13–15 DNA short tandem
repeat (STR) markers or DNA segments. Each individual inherits one copy
of an STR from each parent. The number of repeats in each marker shows
considerable variation among people, making these markers useful for
establishing the twins’ zygosity. The chance that two nonidentical twins
would match for 13 STRs is one in one billion.
In later years, testing of standard DNA markers was used to classify the
MISTRA twins. Three DNA markers (probe MS43, locus D12S11; probe
V1, locus D17S79; and probe SLi989, locus D7S467) were added to the
analysis in 1994, and new ones were added by the time the study was
completed.5 The reliability of this procedure is high, and the current ex-
pense is low (approximately $150/pair). DNA can be derived from buccal
smears, obtained by gently swabbing the inner cheek, a procedure that is
less invasive than drawing blood.
The blood group data, together with fingerprint resemblance (Slater’s
Z score, a measure of co-twins’ similarities and differences in total ridge-
count or number of lines across each fingertip), ponderal index (body
height/cube root of body weight), and cephalic index (head breadth/head
length) yielded a probability of misdiagnosis of less than 0.001.6 Thus,
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 39

the possibility of misclassifying twin pairs was extremely small. Opposite-

sex pairs were assigned as DZ based on their sex difference, although
their blood groups were analyzed in the interest of completeness. The
zygosity of one MZA twin pair was determined by their concordance for
six red blood cell markers and similarities in fingerprint and anthropo-
metric traits. The zygosity of one DZA twin pair was determined by their
discordance for highly heritable physical features such as hair color.
These latter pairs did not visit Minnesota but were assessed in England
by Bouchard and assistant Dr. Alison MacDonald from the Institute of
Psychiatry in London.7
Zygosity determinations were not available until the end of each study
week or until after the twins left Minneapolis. Nevertheless, our global
impressions of the twins and our inspection of their highly heritable char-
acteristics, such as hair color, eye color, and ear shape, were informative
visual clues to twin type. Race and Sanger, who collaborated with James
Shields on his twin studies, noted that “the blood groups practically never
contradict the opinion of such a skilled observer of twins.”8 I confirmed
this observation in a 1984 study by showing that my immediate impres-
sions of twin type, based upon the twins’ similarity in appearance, showed
96 percent agreement with blood type determinations.9
We were generally correct in our informal zygosity assessments, but in
one exceptional case we were wrong. The DZA twins Kerrie and Amy
looked enough alike to be MZA twins—after all, it was mistaken identity
that had reunited them—but they looked just different enough to make
us wonder (Figure 2-2). By the end of the study week, we were convinced
that Kerrie and Amy were “similar-looking DZAs” rather than “dissimilar-
looking MZAs,” and the blood-typing laboratory work proved that our
later impressions were right.
The physical similarity of this pair was enhanced by their concordance
for brittle hair and sparse hair growth. The twins were examined by a
University of Minnesota dermatologist who confirmed the limited hair
growth in one twin and ectodermal dysplasia in the other twin. Ectoder-
mal dysplasia refers to a group of conditions involving abnormalities of
the skin, hair, nails, teeth, and sweat glands. The twins had apparently
inherited this trait from their biological father, but to different degrees—
when their biological mother saw their reunion picture in the newspaper,
she called them to say, “Girls, you have your father’s hair.” One twin had
transmitted the condition to two of her three children.10
Kerrie and Amy were adopted as infants by families in different parts
of Vermont. When Kerrie was eighteen, she moved to a larger city in
another part of the state. Suddenly people started calling her Amy. When
40 B O R N TO G E T H E R — R E A R E D A PA RT

Kerrie turned twenty-five, she met a man at a party who said he’d bet
a  million dollars that she and someone named “Amy” were twin sis-
ters.11 After a series of telephone calls, the twins met and were struck by
their resemblance. Their adoption agency eventually confirmed what
they already knew. Kerrie is angry that she and Amy missed their grow-
ing-up years together. “They didn’t realize what [separation] can do to
somebody,” she said.
Kerrie and Amy were our only DZA twins to be reunited because one
twin was confused for the other. Their experience underlines the wide
variability among DZ (and DZA) twins and full siblings with respect to
similarities in appearance and behavior.
The question of biased data collection based on visual impressions of
twin type is a concern in every twin study. We handled this issue in vari-
ous ways, based on the different tests and activities we administered, but
general rules were followed to maintain impartiality. For example, the IQ
tests were administered by different examiners who were not part of the
research team, the twins were kept apart while they completed question-
naires, and each twin received the same protocols at the same time as his
or her co-twin (or one immediately after the other) to avoid exchange of
information. We also requested that the twins refrain from discussing the
tests until both had completed them. There were exceptions to these rules.
Bouchard, for instance, noted that twins given the special mental ability

Figure 2-2. Look-alike DZA twins reunited by mistaken identity. (Photo credit
Thomas J. Bouchard Jr.)
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 41

battery by different examiners showed larger score differences than twins

who were tested together. Subtle differences in examiner explanations
and/or rapport were the likely causes. Thus, the decision was made to
test the twins together, seating them on opposite sides of the room facing
away from one another. There were some other exceptions to our rules,
and their possible effects on the data will be considered.

The Final Reared-Apart Twin Sample

There were two final reared-apart twin samples. The individual sample
(N = 262 individual twins or triplets) counted each participant just once
and was used for descriptive and correlational analyses. For example, the
individual file was used to examine correlations between different mea-
sures such as parental education and twins’ IQ score.
The paired sample (n = 137 pairs) was reserved for quantitative data
analyses, such as calculating IQ intraclass correlations (measures of twin
resemblance) and estimates of personality heritability (degree to which
personality traits were influenced by genetic factors). The paired file was
also created to accommodate two MZA male triplet sets and two MZA/
DZA triplet sets; the MZA/DZA triplet sets were each composed of two
MZA females and a DZA male co-triplet. Depending on the measure, the
two MZ triplet sets were variously entered as one MZ set composed of
two triplets chosen randomly (triplets 1–2 or triplets 1–3), as three pairs
each (triplets 1–2, triplets 2–3, and triplets 1–3), or as one group of three
(triplets 1–2–3). In my analysis of the twins’ social relationships I entered
the MZA triplets as three separate twin pairs because the relationship
qualities among the different pairings could differ. The two MZA/DZA
triplet sets were usually entered as one MZ female twin pair and two DZ
opposite-sex twin pairs, although the two male-female sets were some-
times eliminated to avoid data duplication.
Three MZA pairs (two male and one female) who were under eighteen
years of age were, with some exceptions, routinely omitted from the data
analyses because we wanted to report findings on adult twins. An addi-
tional MZA female pair that was discordant for severe cardiac difficulties
and other physical complications, presumably due to fetal transfusion syn-
drome (shared prenatal blood circulation), was also generally excluded. The
data from these pairs have, however, been included in the descriptive tables
presented in this chapter in the interest of completion. The zygosity and sex
composition of the two complete samples are displayed in Table 2-3.
The twins’ mean (average) age at assessment and contact measures are
presented in Table 2-4. Contact measures refer to how much time the
twins spent together before and after their separation. The variability
42 B O R N TO G E T H E R — R E A R E D A PA RT

Table 2-3. Zygosity and Sex Composition of the Individual and Paired Reared-Apart Twin Samples

Individual Sample Paired Sample

Zygosity (individuals) (pairs)

MZA 156 81
Male 58 32
Female 98 49
DZA (same sex) 76 38
Male 24 12
Female 52 26
DZA (opposite sex) 30 18
Male 16 —
Female 14 —
Total 262 137

Note: The paired sample in the table includes all possible pairings of the members of
each of the four triplet sets. The MZA female twins with a male co-triplet (two sets) are
entered in the MZA female individual sample, explaining why the number of individual
female opposite-sex twin individuals is lower than the number of males.

(standard deviation [SD]) and range (minimum and maximum values) of

each measure are also provided.
The MZA twin pairs were younger at assessment than the DZA twin
pairs, a difference associated with their younger age at reunion (r = 0.72,
P<.001). These differences are a likely reflection of the greater ease with
which MZA co-twins find each other, due to their similar appearance;
consequently, the period of time between separation and first contact was
shorter for MZA twins than for DZA twins. Note that the MZA twins
were separated at an earlier age than the DZA twins, so they had less
early contact. However, the length of time from reunion to assessment
was longer for MZA than DZA twins, a finding that explains their greater
contact before coming to Minneapolis. This makes sense because MZA
twins met sooner than DZA twins, giving them more opportunity to in-
teract. Still, the percentage of lifetime living apart was nearly identical for
the MZA and DZA twin pairs (95 percent and 97 percent, respectively).
Critics have focused on MZA-DZA differences in (1) time between re-
union and assessment and (2) contact prior to assessment to explain the
greater behavioral resemblance of the MZA than DZA twin pairs. Clini-
cal psychologist Jay Joseph, a harsh critic of behavior genetics,12 has also
argued that length of contact may be less critical than the age at which
the contact occurs.13 He claims that contact between twins in infancy or
childhood cannot be equated with contact in adulthood, which he says
would have a greater impact on behavior. Joseph’s objection fails because
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 43

Table 2-4. Age at Assessment and Contact Measures for MZA and DZA Twins

MZA DZA Total

a
Age at assessment (years)
N (individuals) 156 106 262
Mean 39.75 45.71 42.16
(SD) (13.48) (13.45) (13.76)
Range 11–68 22–77 11–77
Time together before
separation (days)b
N (pairs) 81 56 137
Mean 146.10 322.52 218.21
(SD) (246.7) (429.69) (343.59)
Range 0–1,461 3–1,644 0–1,644
Age at reunion (years)c
N (pairs) 78 55 133
Mean 31.45 42.35 35.95
(SD) (15.46) (13.71) (15.67)
Range 0.33–65.11 20.92–75.11 0.33–75.11
Separation to first contact
(years)d
N (pairs) 78 55 133
Mean 31.29 41.71 35.60
(SD) (15.63) (13.71) (15.68)
Range 0.50–75.17 17.00–75.17 0.50–75.17
Reunion to assessment (years)e
N (pairs) 78 55 133
Mean 8.01 3.12 5.99
(SD) (13.15) (6.88) (11.23)
Range 0–47.25 (0–30.00) 0–47.25
Length of contact (years)f
N (pairs) 80 56 136
Mean 1.65 0.96 1.37
(SD) (3.46) (1.16) (2.77)
Range 0.02–23.67 0.02–4.52 0.02–23.67
Lifetime apart (percentage)
N (pairs) 80 56 136
Mean 95.62 97.57 96.42
(SD) (7.52) (2.52) (6.05)
Range 52–99 86–99 52–99

Note: Contact means either telephone or personal contact, not contact by letter. A few
complex cases were omitted from some measures. SD: standard deviation.
a. MZA < DZA [t(260) = −3.51, P < .001].
b. MZA < DZA [F = 18.18, P < .001, t(80.07) = −2.77, P < .01].
c. MZA < DZA [t(131) = −4.19, P < .001].
d. MZA < DZA[t(131) = −3.98, P < .001].
e. MZA > DZA [F = 18.29, P < .001, t(122.18) = 2.79, P < .01].
f. MZA > DZA [F = 6.61, P < .01; the mean values do not differ significantly].
44 B O R N TO G E T H E R — R E A R E D A PA RT

most contact measures had little influence on the twins’ behavioral out-
comes and some associations were in a counterintuitive direction. For
example, in my analysis of the twins’ social relationship, I discovered a
slight but significant correlation (r = 0.16) between days before separa-
tion and feelings of familiarity at the time of assessment. This implied
that the twins who were together longer felt less familiar with one an-
other than those who were separated sooner! This counterintuitive as-
sociation was possibly a chance finding.
The MISTRA sample also included the twins’ spouses, children, par-
ents, siblings, and siblings’ spouses, for a total of 214 nontwin individu-
als. These nontwins completed a shorter version of the assessment bat-
tery that included IQ tests, life history interviews, and inventories. Their
data were often included in the age- and sex-correction procedures we
performed prior to data analysis.
Age- and sex-correction procedures remove the effects that shared
age and sex might have on trait similarity. For example, two women aged
fifty-six years who were born in the 1950s would most likely have been
childhood readers of the Bobbsey Twins, teenage fans of the Beatles, and
adult users of calcium supplements to offset bone loss—more so than two
people of differing age and sex. If reading preferences, favorite songs,
and medical treatments are associated with age and sex, then it is impor-
tant to control for them. We did this according to the statistical proce-
dures described by McGue and Bouchard.14 Data from the nontwins also
allowed us to see if the different measures were affected by assortative
mating. The age and sex of the nontwins are shown in Table 2-5.
Country of origin, summarized in Table 2-6, was considered the place
where twins resided at the time of the study. Most twins came from the
United States (45.4 percent) and England (32.4 percent). The study was
generally limited to twins from English-speaking countries, although sev-
eral twins originally from Germany, China, and the Netherlands partici-
pated with the assistance of interpreters and/or protocols translated into
their native languages. Several twins were unable to travel to Minneapolis,
so they completed an abbreviated portion of the study in their hometowns.
In such cases, the protocols were administered by the study staff and by
other trained individuals, as in the case of the British twins described
earlier.
A small subgroup of pairs included twins raised in different countries—
Germany and Canada, China and the United States, and England and
Australia, to name a few. But the twins with the most exotic rearing differ-
ences were Jack Yufe and Oskar Stohr (Figure 2-3), who were mentioned
briefly in Chapter 1.
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 45

Table 2-5. Age and Sex of Nontwin Participants

Percentage
Relationship Number Age (SD) Range Female

Spouse of twin/triplet 144 45.76 43.1

(11.91)
21–79
Child of twin/triplet 4 28.75 50.0
(14.15)
15–45
Adoptive parent 8 46.50 87.5
(5.01)
39–53
Sibling of twin 28 38.96 67.9
(14.58)
20–68
Spouse of sibling 8 37.75 75.0
(13.92)
23–65
Other 22 43.91 72.7
(14.67)
17–72
Total 214 44.09 52.3
(12.84)
15–79

Table 2-6. Country of Residence for Reared-Apart Twins

Number
Country Percentage (individuals)

Australia 5.3 14
Canada 1.9 5
England 32.4 85
Germany 1.9 5
Israel 0.4 1
Netherlands 0.8 2
New Zealand 0.8 2
South Africa 0.4 1
Sweden 1.1 3
United States 45.4 119
Mixed/Uncertain 10.0 26
Total 100.0 262
46 B O R N TO G E T H E R — R E A R E D A PA RT

Jack and Oskar were born in 1933 to a Romanian Jewish father (Josef)
and a German Catholic mother (Liesel) who had met on a ship headed to
Trinidad. The couple separated when the twins were six months old—
Josef kept Jack in Trinidad and raised him Jewish, while Liesel took
Oskar to Germany and raised him Catholic. Jack grew up fearful of his
German roots in British-controlled Trinidad, while Oskar concealed his
Jewish ones in Nazi Germany. Both twins dealt with these fears in com-
plementary ways—Jack by becoming “very British” and Oskar by be-
coming “very German.” The twins met briefly in Germany in 1954 at age
twenty-one. It wasn’t a friendly reunion because, aside from their lan-
guage barrier, they regarded each other with suspicion. They didn’t meet
again until 1980 after Jack’s wife read about the MISTRA in People
magazine. When the twins met at the Minneapolis International airport,
both were wearing light-blue shirts with epaulettes on the shoulders and
wire-rimmed glasses.
There was a lot of public fascination surrounding this pair. The reunion
of “Nazi and Jew” appeared in newspaper headlines, but this was an ex-
aggeration because Oskar was only twelve years old when World War II
ended. Like other boys his age, he had been required to join the Hitler
Youth League, an activity he enjoyed mostly for its sports opportunities.
The twins’ different rearing circumstances extended beyond their
extraordinary cultural divide. Jack was raised by a father, and Oskar by

Figure 2-3. MZA twins Jack (left) and Oskar. (Photo courtesy Thomas
J. Bouchard Jr. Photograph taken in David Lykken’s laboratory.)
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 47

a grandmother (his mother was often absent). Jack joined the Israeli
navy and worked on a kibbutz; Oskar worked in the coal mines in the
Ruhr. But their Minnesota Multiphasic Personality Inventory (MMPI)
personality profiles matched almost completely, and they displayed a
number of unusual similarities—washing their hands before and after us-
ing the toilet, collecting rubber bands around their wrists, and intention-
ally sneezing loudly in crowded elevators. Oskar savored Jack’s spicy
Caribbean cuisine, so different from his German diet of bread and brat-
wurst. These twins are a great example of how common genes can lead to
shared behaviors even when environments diverge dramatically. Oskar
and Jack were fascinated by their shared habits and tastes, but repelled by
their contrasting political and religious beliefs. They engaged in a love-
hate relationship until Oskar’s death in 1997, a very sad event for Jack.15
The substantial percentage of British twins is due to interest taken in
the MISTRA by the late John Stroud, former Assistant Director of Social
Services for Hertfordshire, about ten miles outside London. Tall, heavy-
set, and imposing with a bushy beard, twinkling eyes, and dry sense of
humor, Stroud took great pleasure in reuniting families, especially twins,
and he developed a national reputation for doing so. Stroud’s death in
1989 was a personal and professional blow to everyone who knew him
and to twins still hoping to meet some day.16
Passage of the Children Act in 1975 gave adopted British children
the right to know their biological parents. This act, and the fact that time
of birth (for twins only) is recorded on birth certificates in England and
Wales, facilitated Stroud’s efforts.17 However, twins in one British DZA
female pair did not benefit from this practice because they were born in
different months—one on October 31, the other on November 1—a situ-
ation that delayed their reunion for some time. In another case, a clerk
reviewing a record for a twin searching for biological kin inadvertently
placed her thumb over the time of birth while inspecting the birth certifi-
cate, hiding the fact that the person in the search was a twin.
Stroud commented, “The aspects I enjoy are the detective work which
goes into these cases . . . and the sheer happiness that comes when such
separated twins meet each other. I think every twin I have helped has
used the same phrase, ‘I am over the moon.’ ”18 During our life history
and twin relationship interviews, some participants told me they discov-
ered their twinship while searching for their biological parents. They ad-
mitted that, once this was known, finding their twin became more impor-
tant than finding their mothers and fathers.
48 B O R N TO G E T H E R — R E A R E D A PA RT

Why Twins Were Separated

The MISTRA twins were separated because of parental divorce (7.3 per-
cent), parental death (6.6 percent), illegitimate birth (44.5 percent), phys-
ical incapacity of parents to provide child care (3.7 percent), financial
limitations (13.1 percent), and miscellaneous causes (13.9 percent). The
reasons for separation were unknown in 10.9 percent of the cases. Several
unusual separation circumstances are of particular interest. A set of MZA

Figure 2-4. Reared-apart MZA triplets Eddy Galland, David Kellman, and
Robert Shafran (left to right) who participated in the MISTRA at age nineteen.
(Photo credit Thomas J. Bouchard Jr.)
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 49

male triplets from New York City, Eddy, Dave, and Bob (Figure 2-4), were
placed in different adoptive homes at the advice of Columbia University
psychiatrist Viola Bernard. Bernard believed that separate rearing bene-
fited multiple birth children because they would enjoy undivided parental
attention and a special place in their families. These triplets (and five
MZA twin pairs who were not in the MISTRA) were part of the discred-
ited 1960s study of separated twins by Neubauer that resulted from these
placements, as mentioned in the Introduction.19
Other unusual circumstances kept some twins apart. Canadian MZA
twins Brent and George were inadvertently switched with a nontwin
infant, Marcus, while being moved to a second foster home. Members of
MZA male set Bill and Bruce were separated because their parents al-
ready had a daughter and had planned on having only two children in
total—the ideal American family—so they gave Bruce away. Professor
William Tucker, in questioning the size and characteristics of Cyril Burt’s
MZA twin sample, found it “a tad preposterous that a middle class cou-
ple, unexpectedly finding themselves the parents of twins, pick one infant
to send to an orphanage so as not to put a crimp in their life style.”20
Nevertheless, we had such a situation in our study.
When I asked Bouchard to name the most memorable reared-apart pair
we studied, he didn’t hesitate. Jim and Jim were “the most spectacular”—
so were Oskar and Jack as well as Bridget and Dorothy, the twins who
each wore seven rings, three bracelets, and a watch. Mark and Jerry
(both volunteer firemen) were “pretty interesting.” But Bill and Bruce
were “the greatest pair on earth.” One evening, Bouchard invited the
eighteen-year-old twins and their two families to his former home on
Lake Minnetonka in Wayzata, Minnesota. While he was entertaining
their parents, he noticed that the twins had walked to the dock holding
hands—“they wanted to talk,” he said.
I remember being in the testing room, watching Bill’s parents watching
Bruce. They seemed greatly in awe and very uncomfortable seeing a rep-
lica of their son, the twin they had given away.
Twins are, presumably, separated less often today than in the past be-
cause adoption agencies have been sensitized to the importance of twin
and sibling relationships. Most jointly placed brothers and sisters do as
well as, or better than, singly placed children or siblings, though a few
do worse.21 Current research recommends placing siblings together,
while evaluating each situation on a case-by-case basis in the event of
exploitation or bullying of one sibling by another.22 The MISTRA twins’
life stories and reunions, as well as research documenting the significance
of the twin relationship, have been used to support common placement in
cases involving twins’ custody.23
50 B O R N TO G E T H E R — R E A R E D A PA RT

Joan DiLeonardi, who helped reunite twins Barbara and Judy, recalled
that a former executive at Illinois’s Children’s Home and Aid Society felt
that separating twins was an acceptable practice. Years later, when she
met the former executive at a reception, he said that he was puzzled over
efforts to bring twins together, implying that there was nothing special
about twin children. The social stigma attached to illegitimate birth has
lessened in many societies, allowing unwed mothers to raise their infants
comfortably. However, twins continue to be reared apart for other rea-
sons. Assisted reproductive technology often results in multiple births, a
situation some families are emotionally or financially unable to manage.
An estimated tens of thousands of female infants were abandoned under
China’s One-Child Policy, indirectly resulting in an unknown number of
separated twins sets, some of which I am studying.24 Several switched at
birth twin cases have also been identified over the years.25 Given inadequate
procedures for linking mothers and newborns in some hospitals, it is
likely that other cases exist but may never be detected.26

Rearing with a Parent or Family Member

A large proportion of the MISTRA twins (70.2 percent) were reared by
an unrelated family. The remainder were raised in other circumstances,
including biological or nonbiological caretakers or multiple placements
with relatives or nonrelatives. Individuals who had multiple early place-

Table 2-7. Reared-Apart Twins’ Rearing Circumstances

Number
Rearing Circumstance Percentage (individuals)

Unrelated family 70.2 184

Biological mother 3.8 10
Biological father 1.5 4
Grandparents 0.8 2
Uncle/aunt 3.1 8
Multiple families (relatives) 1.5 4
Multiple families (nonrelatives) 5.3 14
Biological mother and father 2.7 7
Unrelated, stable by age 5 1.5 4
Multiple relatives, stable by age 5 .8 2
Multiple nonrelatives, stable by age 5 1.5 4
Complex case 7.3 19
Total 100.0 262

Note: Individuals in four triplet sets are each counted once.

 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 51

Table 2-8. Rearing Circumstances of Reared-Apart Twin Pairs Organized by Zygosity (Pair Data)

MZA DZA

% n (pairs) % n (pairs)
Rearing Circumstance 81 56

Unrelated-unrelated 70.4 57 62.5 35

Related-related 7.4 6 5.4 3
Unrelated-related 14.8 12 12.5 7
Complex case–other 7.4 6 19.6 11

Note: Rearing categories in this table were collapsed across the specific biological and
nonbiological rearing circumstances listed in Table 2-7.

ments with both relatives and nonrelatives or who had no primary care-
takers were assigned as “complex” cases. These data are summarized in
Table 2-7.
Some critics have claimed that the reared-apart twins’ behavioral simi-
larities are enhanced when co-twins are reared by biological relatives.
However, those conducting the analyses have a different perspective on
this question. In 1958, Shields, who was in the early stages of his own in-
vestigation, emphasized the advantages of having twins variously raised
by related and unrelated families “because of the possibility for making
internal comparisons.”27 (This issue is examined again in the next chap-
ter.) The co-twins in most of our pairs were both reared by unrelated in-
dividuals (67.2 percent). The next most frequent arrangement was one in
which one co-twin was reared by biological relatives and the other was
reared by nonrelatives (13.9 percent), as shown in Table 2-8.

Family and Environmental Measures

The twins’ family and environmental measures were gathered as part of the
life history battery. Specific measures included the mean rearing parents’
years of education, the highest occupational status attained by mothers and
fathers, the twins’ number of residences, and the presence of forty-one dif-
ferent physical facilities (e.g., a world atlas and a pet dog or cat) in the rear-
ing home. The parents’ highest occupational status was coded with the
Duncan SEI system using the methods of Charles W. Mueller and Toby L.
Parcel.28 The largest percentage of fathers for whom information was avail-
able were managers, officials, proprietors/craft, and the largest percentage
of mothers were homemakers. The occupational data are presented in
Table 2-9 for the full sample of individual twins, organized by zygosity.
Another common criticism of the study was that similarities in the rear-
ing home might largely explain the MZA twins’ behavioral similarities.
52 B O R N TO G E T H E R — R E A R E D A PA RT

Table 2-9. Rearing Parents’ Best Occupation (Duncan SEI) for MZA and DZA Twin Individuals

Rearing Father

MZA DZA

Occupation n % n %

Professional/technical 29 20.1 9 9.2

Managers/officials/ 36 25.0 28 28.6
proprietors
Clerical 8 5.6 7 7.1
Sales 5 3.4 4 4.1
Craft 35 24.3 28 28.6
Operatives 17 11.8 10 10.2
Service 7 4.9 1 1.0
Nonfarm labor 4 2.8 6 6.1
Farmer/farm manager 3 2.1 4 4.1
Farm laborer — — 1 1.0

Rearing Mother

MZA DZA

Occupation n % n %

Professional/technical 20 14.4 10 11.0

Managers/officials/ 8 5.8 10 11.0
proprietors
Clerical 30 21.6 16 17.6
Sales 11 7.9 4 4.4
Craft 3 2.2 3 3.3
Operatives 18 12.9 11 12.1
Service 23 16.5 9 9.8
Nonfarm labor 1 0.7 — —
Farmer/farm manager 1 0.7 — —
Homemaker 24 17.3 28 30.8

Note: MZA females in two MZA/DZA triplet sets are entered as individuals in the
MZA sample only. Data were incomplete for rearing parents in some cases.

We found that the percentages of pairs in which the co-twins’ rearing

parents held similar jobs were comparable for the MZA (rearing father:
20 percent, rearing mother: 24 percent) and DZA twins (rearing father: 17
percent, rearing mother: 20 percent). Therefore, if the parents’ occupa-
tion did affect their children’s intellect, the effect would be the same for
both types of twins. Most importantly, however, for this criticism to be
meaningful, it is first necessary to demonstrate that a specific home feature
is trait-relevant. That is, a causal link between the parental occupation
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 53

and the children’s behavioral outcome (such as children reared by law-

yers excel in verbal skills) must be demonstrated if the argument is to be
taken seriously.

The Reared-Apart Twins’ Reunion Circumstances

Information on how twins found each other was available for all but five
pairs (four MZA and one DZA). Most twins (n = 99, 75.0 percent) lo-
cated each other after one or both contacted adoption agencies, placed
notices in the media, called names in telephone books, or visited old
neighborhoods. Other events responsible for the twins’ reunions in-
cluded mistaken identity (n = 18, 13.6 percent) and family knowledge
of the other twin (n = 13, 9.9 percent). The remainder (n = 2, 1.5 per-
cent) met due to various other circumstances. For example, MZA twins
Caroline and Margaret Shand met when they recognized each other at
a church.
The Shands grew up just twenty miles apart in Scotland. Each knew
that she had a twin, but each had been told not to ask questions about
their birth. They were interested in one another, but they never searched.
Then, at age sixty-four, Caroline was working in a church, and one day
she heard a knock at the door. When she finally answered, the two visitors
had walked to a nearby gravesite. Caroline approached them and saw
herself in one of them. “Are you my sister?” she asked Margaret. They
“hugged, laughed and, of course, cried.”29
Shields, author of the 1962 British MZA twin study, assessed twins who
had met under somewhat similar circumstances. MZA twin girls Jessie
and Winifred were drawn to one another at age two, but their meetings
were not encouraged by Jessie’s family. The twins lived within a few hun-
dred yards of one another, and when they became school friends at age
five, Jessie’s mother enrolled her daughter in a different school. At this
juncture, the girls were told they were twins. When they turned eight,
they ended up at the same school for administrative reasons, becoming
Shields’s youngest separated pair. Shields wondered if there might be a
biological basis to MZA twins’ attraction, “based on likeness . . . not just
a matter of cultural attitudes to twins.”30

Comment on Recruitment
Several recruitment examples are worth describing because they show
that the more similar MZA twin pairs did not always participate in the
MISTRA. One case concerned MZA female twins who were both
54 B O R N TO G E T H E R — R E A R E D A PA RT

extremely fearful of flying to Minneapolis from the East Coast. The

twins declined Bouchard’s offer to be driven to Minnesota by staff mem-
bers or to have testers visit their homes. According to Bouchard, they
looked “strikingly alike” and displayed the same “high stress reaction.” A
second example concerns an MZA male twin nearing retirement who
initially declined our invitation. Bouchard visited the twin while travel-
ing and learned that this individual had a heart problem that in his state
was considered work-related and would earn him a higher retirement
pension. If, however, the problem was found during the study and proved
to be genetically linked, then his higher pension would be denied. We
solved this situation by delaying the assessment until the twin’s retire-
ment was final, an option that had not been raised in previous communi-
cations. Thus, both participation and lack of participation by individuals
could be explained in many ways.
Bouchard maintained a file of twins who were searching for their co-
twins. Approximately eight twins had not been reunited with their twin
brothers and sisters when the study ended.

The Assessment in Full

Once twins were scheduled for the assessment, they received a packet of
materials with information about the study and a list of items to bring,
such as adoption documents and childhood photographs. All travel and
hotel expenses were covered by the project. Twins were also given a mod-
est honorarium for miscellaneous expenses ($75.00 to $275.00), al-
though various other costs such as child care, pet sitting, and up to $200
in lost wages were reimbursed directly. The project also paid for one
individual (usually a spouse) to accompany twins to Minneapolis and to
complete a short form of the psychological assessment.
Twins traveling to Minneapolis from the United States or Canada ar-
rived the day before the study began (Saturday); twins traveling from other
countries arrived two days earlier (Friday) so that they could recover from
jet lag. Twins departed on the final day of the study or remained in the
United States at their own expense.
By February 1980, nearly one year after the study began, the standard
assessment schedule (“twin week”) included five full days and two half-
days. The first half day was a Sunday, and the second half-day was the
following Saturday. One evening was set aside for sexual life history in-
terviews for twins only.
Bouchard usually drove the twins from their hotel to the university on
Sunday afternoon and escorted them to a large room (the “twin room”)
on the fourth floor of the Psychology Department, located in Elliott Hall.
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 55

Two or three assistants (myself included) would already be present, hav-

ing prepared the test materials, inventory booklets, and refreshments. It
was exciting—actually thrilling—-to meet the twins and to discover what
they were really like.
We first reviewed the week-long schedule with the twins as part of the
informed consent procedure. We encouraged them to complete all the
medical and psychological tests we administered, but we assured them
that they were free to decline any activity they preferred not to do. We
explained to the twins that we wanted to keep the data as free from bias
as possible so we asked them to refrain from discussing any of the tests or
procedures with one another until they both had completed them. We also
assured them that all the data we gathered from them would be kept con-
fidential and that we would not release findings to their twin, spouse, or
anyone else without their written consent.
Beginning on Monday, Bouchard or a staff member picked up the twins
at their hotel each weekday morning at 7:45 and brought them back be-
tween 5:00 and 6:00 each night. Each participant’s progress was recorded
on a checklist that was updated continually (Figure 2-5). The schedule
worked well because responsibilities and assigned times were decided in
advance and were consistent across pairs.

Figure 2-5. Bouchard updating the twins’ checklist, May 1985. A pair of MZA
male twins from Great Britain and their wives were undergoing assessment at
the time. (Photo credit Nancy L. Segal.)
56 B O R N TO G E T H E R — R E A R E D A PA RT

We examined the twins in testing rooms located in Elliott Hall, the

building that housed the Psychology Department, and in the University of
Minnesota’s hospital complex. Time was set aside for completion of sched-
uled inventories and questionnaires that we had prepared and bound
into four (later five) spiral notebooks (Psychological Batteries I–V). Twins
completed these forms in Elliott Hall and in hospital lounges while wait-
ing for medical appointments. It is estimated that of the 15,000 questions
answered by the twins, 4,000 came from the inventory booklets. This
number may seem high, but most twins completed the full battery during
their assigned week; occasionally, twins took remaining questionnaires
home and returned them by mail.
Journalists, writers, and television producers were often interested in
interviewing twins. We always asked the twins in advance if they would
agree to media coverage and we arranged this only with their consent.
Actual data collection sessions were never filmed.
A complete annotated list of the psychological, physical, and medical
tests appears on the Web site. Information on administration, results, and
conclusions are provided in subsequent chapters as this research story
unfolds.

Genetic Terminology
The field of human genetics has many terms and concepts, but only some
require definition for the purposes of this book. Some terms will be defined
as they occur in the text, and others are further explained in the Glossary.
Individuals familiar with the genetics field and/or with quantitative genetic
analysis may wish to skip the next two sections of this chapter.
Genes are the hereditary units that occupy a specific location (locus)
on one of our twenty-three pairs of chromosomes. Chromosomes are the
structures in the nucleus of each of our cells that contain our genetic ma-
terial, or DNA. DNA stands for deoxyribonucleic acid and is the double-
helical molecule that carries genetic information. Different forms of a
gene at a certain locus are called alleles.
Each parent transmits 50 percent of his or her genes to each child. The
egg and sperm (sex cells or gametes) each contain one copy of each of the
twenty-three pairs of chromosomes; thus, each child has a maternal allele
and a paternal allele at each locus. Full siblings (and DZ twins) have a
50 percent chance, on average, of sharing the same gene at a given locus.
Some siblings (and DZ twins) will share over 50 percent of the same
genes, and some will share less. This may depend upon chance, but it
may also depend upon whether mothers and fathers match on genetically
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 57

influenced traits (assortative mating). Assortative mating can lead to

greater than expected trait similarity among siblings and DZ twins. How-
ever, assortative mating will not affect trait similarity between MZ twins
because they share 100 percent of their genes.
The genotype of an individual refers either to all the genes he or she
carries or to the genes that are relevant for a specific trait that he or
she carries. The phenotype refers to a measurable or observable trait. For
example, a person with type A blood (phenotype) may carry either two
genes coding for type A, or one gene coding for type A and one gene cod-
ing for type O (genotype). This is because the gene for type A blood is
dominant, meaning that it is expressed regardless of other genes with
which it is paired. The gene for type O blood is recessive because it is not
expressed if paired with a dominant gene; people with type O blood must
have two type O (recessive) alleles.
Close correspondence of genotypes and phenotypes does not always
occur as it does in the blood type example given above. Some phenotypes
may change depending upon the environment. A person who is allergic
to penicillin will not show an adverse reaction unless he or she receives
the medication. A person with a tendency toward weight gain may add
or lose pounds depending upon his or her diet and exercise.
Discrete traits are those that are affected by genes at one location, or lo-
cus, on a chromosome. Discrete traits include the major blood groups (A, B,
or O), phenylketonuria (an inborn error of metabolism), and Huntington
disease (a lethal condition linked to a specific gene on chromosome 4). Such
traits have been called “either-or” traits because either you have them or
you do not. In contrast, continuous traits (also called polygenic or biometri-
cal traits) are those associated with many genes on different chromosomes.
Examples include verbal ability, running speed, and body weight. These
traits differ from either-or traits because everyone (within the normal range
of human development) has verbal ability, can run, and has weight. The dif-
ferent values of these traits, representing fluency, quickness, and size, form a
population distribution along which each individual is located.
Some traits (continuous or discrete) are affected by epistasis or interac-
tion between several genes at different loci (chromosomal locations). An
example in humans is the Bombay phenotype in which individuals lack a
protein called the H antigen. Such individuals will show the type O blood
phenotype even if they carry type A genes.31 Epistasis is likely to reduce
full sibling and DZ twin resemblance because the odds that a second
sibling (or DZ twin) will inherit the same multiple gene combination as
his or her co-sibling (or co-twin) is slim (25 percent for the same maternal
allele and the same paternal allele at a given locus, but less for multiple
58 B O R N TO G E T H E R — R E A R E D A PA RT

alleles at different loci). This also explains why MZ twin resemblance for
complex polygenic traits (those involving many different genes) should be
high, and full sibling and DZ twin resemblance should be relatively low.
MZ twins inherit 100 percent of the same genes and, consequently, all
their unusual gene combinations, whereas DZ twins inherit 50 percent of
the same genes, on average, by descent.
Lykken introduced the term emergenesis to describe complex traits that
show large resemblance differences between MZ and DZ twin pairs (i.e.,
when the DZ twin correlation is less than half the MZ twin correlation).
Emergenesis can be considered a grand version of epistasis; it is discussed
further in Chapter 9. Epistasis and emergenesis are likely to reduce parent-
child resemblance in some traits because the gene combinations under-
lying these traits in the parental generation become disentangled in the
children’s generation.

Quantitative Analyses
The MISTRA and other twin studies use intraclass correlations and bio-
metrical modeling to assess genetic and environmental contributions to
trait variation (individual differences). These terms are used through-
out the book in conjunction with the specific analyses. Additional details
about the different procedures are available in behavioral genetics text-
books32 and twin methods manuals.33
Correlations express the extent to which two measures vary or change
with one another. Correlations vary from –1.0 to +1.0, with 1.0 indicating
a perfect positive relationship, −1.0 indicating a perfect negative relation-
ship, and 0.0 indicating no relationship at all. For example, traditionally
the number of candles on our birthday cakes go up by one for every year
that we age. The correlation between candles and age would be positive
(and 1.0) since they both change in the same direction and by constant
amounts. In another example, the weight of one’s clothing is likely to go
down as the outside temperature increases. The correlation between cloth-
ing weight and temperature would, therefore, be negative because these
measures are changing in opposite directions. The correlation for these
two measures might conceivably be −0.80. The relationship between them
might not yield a perfect correlation because some people wear protective
hats and scarves in the sun, thereby adding weight, and because changes
in the two probably do not occur at a constant rate.
Some measures show no relationship—that is, both may increase or
decrease in ways that are unrelated to each other. For example, it is un-
likely that height and hair color are related in a meaningful way. Tall and
short people are probably equally likely to have dark or light hair.
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 59

Ordinary correlations that express the relationships between two mea-

sures, as in the examples, are not ideal for assessing twin similarity. For
one thing, twin (and sibling) similarity is concerned with the agreement in
the same test or inventory given to two different people. In addition, twin
researchers are interested in finding the amount of trait variation that
is shared between co-twins. This can be accomplished by using a special
type of correlation called the intraclass correlation. Intraclass correlations
(ri) express the ratio of the between-pair variation (differences among
twin pairs) to the total variation (differences among twin pairs plus differ-
ences within twin pairs). These types of correlations are calculated sepa-
rately for MZA and DZA twin pairs and compared. Relatively greater
differences between pairs yield higher intraclass correlations, reflecting
greater resemblance within pairs. Relatively lower differences between
pairs yield lower intraclass correlations, reflecting reduced resemblance
within pairs.
Suppose that the members of twin pair A in one group obtained IQ
scores of 100 and 105, and the members of twin pair B obtained scores
of 125 and 120. Next, suppose that the members of twin pairs C and D
in a second group obtained IQ scores of 100 and 105, and 102 and 107,
respectively. In both groups, the average within-pair difference is 5. How-
ever, in the first group the co-twins in each pair score closer to each other
than they do to the members of the other pair. Here, the between-pair
differences would be high relative to the within-pair differences, yielding
a high intraclass correlation. However, in the second group, the co-twins
score closer to the members of the other pair than they do to their own
co-twin. Here, the between-pair differences would be low relative to the
within-pair differences, yielding a lower intraclass correlation than found
for the first group. Of course, researchers would calculate intraclass cor-
relations using much larger samples!
Genetic effects on a trait are demonstrated when MZ (or MZA) intra-
class correlations exceed DZ (or DZA) intraclass correlations.34 Emergenic
effects would be suggested when the DZ (or DZA) intraclass correlation is
less than half the size of the MZ (or MZA) intraclass correlation.

Two other concepts—genotype by environment interaction and genotype-

environment correlation—are important to think about with reference
to twins reared apart. Genotype × environment interaction (G × E) means
that how the environment affects a phenotype depends on the genotype.
For example, strenuous activity (environment) may strengthen the body
(phenotype), but the resulting strength will vary with individuals’ general
body build (genotype). And if one basically fit MZA twin was reared in a
home emphasizing a healthy diet and his or her co-twin was reared in a
60 B O R N TO G E T H E R — R E A R E D A PA RT

family that feasted on junk food, both twins might still end up with
similar health histories because the twin with the poorer diet may have
been resistant to the effects of the diet and other adverse environmental
conditions—up to a point.
Genotype-environment correlation (G-E) refers to associations between
individuals’ genes and environments, of which there are three types. Pas-
sive G-E involves parental transmission of both genes and environments
to children. Musically gifted parents might transmit both genes and envi-
ronments conducive to musical talent to their children. Reactive (evocative)
G-E involves reactions to individuals based on their expressed behaviors.
The teachers of bright MZA twins might be likely to provide them with
intellectually stimulating activities. Active G-E involves individuals’ seek-
ing out opportunities that support their interests and talents. Sports-
minded MZA twins might independently seek opportunities to practice
their skills. Passive G-E would apply to MZT twins but not to MZA
twins who are reared apart. However, teachers might react the same way
to two bright MZA twins despite their separate rearing. And two musi-
cally talented MZA co-twins might both seek opportunities to perform
even in very different environments.
The reared-apart twin study by Newman, Freeman, and Holzinger il-
lustrates another important interpretive issue regarding differences within
and between pairs. The MZA ri s for the Binet IQ score was 0.67, indicat-
ing substantial genetic influence.35 Thus, the MZA co-twins were more
like their co-twins in IQ than they were like the members of the other
pairs. At the same time, the ordinary correlation expressing the relation-
ship (within pairs) of the co-twins’ IQ differences and their educational
differences was 0.79. This meant either that (1) the higher scoring twin
had had more schooling or better schooling than his or her co-twin, or
(2) the higher scoring twin had sought out more educational experiences.
Some people have interpreted the second result as discounting genetic ef-
fects on IQ, but this is a mistake because the co-twins in each pair were
generally more similar to one another than they were to the members of
the other pairs.

People differ across traits due to the genetic and environmental differences
among them. Heritability is the proportion of population differences in
intelligence, height, or running speed that is associated with genetic dif-
ferences among the members of a population. To say that running speed
has a 33 percent heritability in ten-year-old females means that about
one-third of the population differences in that trait are explained by the
genetic differences among the young female members.36 It does not mean
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 61

that one-third of a particular young girl’s running speed is associated with

her genes. Heritability does not refer to individuals for whom genes and
environments are inextricably intertwined.
Heritability tends to go up when people’s trait-relevant environments
are more uniform: when environments are alike, the differences between
people are more closely tied to their genetic differences. Conversely, heri-
tability tends to go down when people’s trait-relevant environments are
more variable: when environments differ, the differences between people
can reflect these effects. (This is what makes MZA twins’ similarities
striking.) Furthermore, heritability estimates characterize only a particu-
lar population at a particular point in time, so they cannot be generalized
to other populations. The heritability of a trait can also change depend-
ing upon the environment, although some traits are more resistant to
change than others. A healthier, more uniform diet can increase the aver-
age height of a population—in doing so it could also increase the herita-
bility of height because individuals’ height differences would more closely
reflect the genetic differences among them. The extent to which individ-
ual differences in a trait are explained by genetic and environmental fac-
tors can be estimated using the biometrical techniques described later in
this section.
There are two types of heritability: narrow and broad. Narrow herita-
bility is associated with genes that work in an additive fashion. Additive
genetic variation refers to the independent effects of genes adding up
to produce a given trait. In other words, the genetic effect is a simple
summation—one gene does not affect the expression of other genes. If
the effect of one allele is “1” and the effect of the other allele is “1,” the
total effect is “2.” Additive genes are responsible for most of the genetic
resemblance between family members. Nonadditive genes are those that
interact with one another and may involve two genes (dominance), sev-
eral genes (epistasis), or many genes (emergenesis). In other words, they
do not add up. MZ twins share their nonadditive genes; recall that DZ
twins and siblings have a 25 percent chance of inheriting the same pater-
nal and maternal alleles (at a given locus) as their co-twin or co-sibling,
but less chance of sharing alleles across many loci.
Broad heritability is associated with all sources of genetic effects (addi-
tive and nonadditive). The MZA intraclass correlation directly estimates
broad heritability because MZA twins share all their genes but do not
share their rearing environment.37 In other words, MZA co-twins have
only their genes in common, so their observed similarities reflect their
shared genes. The DZA intraclass correlation directly estimates half the
heritability because DZA twins share 50 percent of their genes, on average;
62 B O R N TO G E T H E R — R E A R E D A PA RT

therefore, doubling the DZA correlation also estimates broad heritability.

This calculation assumes that there is no assortative mating—if there were,
the DZ twins’ increased resemblance would reduce heritability because
their degree of similarity would be closer to that of MZ twins.
Narrow heritability refers only to additive genetic effects. It can be indi-
rectly estimated by calculating the difference between the MZT ri and
the DZT ri and doubling the difference (2[MZT ri − DZT ri]), known as
Falconer’s formula.38 Based on a survey of many studies, the overall
MZT ri for IQ is 0.86, and the DZ ri is 0.60.39 The IQ heritability would
then be equal to 2(0.86 − 0.60) or 0.52. Falconer’s formula may overes-
timate narrow heritability in the presence of nonadditive genetic
variance.
Intraclass correlations can be interpreted as causal because twins are
experiments of nature (individuals who share 100 percent of their genes
or share 50 percent of their genes, on average) and experiments of society
(rearing together in the biological family or rearing apart in a nonbiologi-
cal family).40 For example, both MZT and DZT twins share their rearing
environment, but DZT twins share half as many genes, on average. Thus,
higher MZ than DZ intraclass correlations demonstrate genetic influence
on a trait. MZA and MZT twins both share 100 percent of their genes, but
only MZT twins share their rearing environment. IF MZA and MZT twins
show the same degree of resemblance for a trait, this shows that sharing a
family environment does not make family members alike.

Age- and sex-corrections, statistical procedures for removing the effects

of age and sex on a trait, have been applied in virtually all the MISTRA’s
data analyses. MZ twins and approximately two-thirds of DZ co-twins
are the same age and sex, factors that can inflate trait similarity between
them just as it would between two unrelated people who are both twenty
and both female. The age- and sex-adjustment procedures used in the
MISTRA have been described by McGue and Bouchard.41 Briefly, test
scores are corrected for the effects of age, sex, age2, age × sex, and age2 × sex,
and analyses are performed on the residuals (the values remaining after
the various age and sex effects are removed).
Most of the MISTRA’s heritability analyses used biometrical modeling
techniques as well as intraclass correlations. The simple comparison of the
MZ (or MZA) and DZ (or DZA) intraclass correlations is an important
first step in behavioral-genetic analysis because this demonstrates whether
or not there is genetic influence on the trait. Biometrical modeling, devel-
oped since the 1970s, is a more informative approach to understanding
the origins of a trait because it allows genetic and environmental factors
 1 5 , 0 0 0 Q U E ST I O N S × 1 37 PA I R S 63

affecting traits to be estimated simultaneously. The genetic component

can be subdivided into additive variance and nonadditive genetic vari-
ance, as we saw earlier, whereas the environmental component can be
subdivided into shared and nonshared environmental effects. Shared
environmental effects are the common events or experiences that lead
to similarities between family members. For example, siblings in a family
share meals, so they might develop similar tastes in food. Nonshared en-
vironmental factors are events or experiences that affect only one person
in a family, making that person different from his or her relatives. If one
sibling took piano lessons and the other sibling took acting lessons, their
musical and theatrical abilities might diverge.
Biometrical methods can test how well a model fits a given set of data.
Model fit refers to the discrepancy between the actual data (observed val-
ues) and the results implied by the model (expected values). The simplest
such models assume that shared genes underlie similarity between rela-
tives, mating occurs at random (is not assortative), genetic effects are addi-
tive, genetic and environmental effects are independent from each other,
and genetic and environmental effects combine additively.
Additive genetic effects, nonadditive genetic effects, and shared envi-
ronmental effects cannot be estimated simultaneously with only MZT
and DZT twin pairs.42 Fortunately, multiple kinships such as reared-apart
and reared-together twins, or twin families consisting of twins, their par-
ents, and/or their children can be incorporated into the same biometrical
model, leading to improved estimates of genetic and environmental ef-
fects. Investigators can compare the fit of several models and decide, for
example, if a model that includes genetic and nonshared environmental
components fits the data better than a model that also includes a shared
environmental component. Random error measurement is often consid-
ered as a nonshared environmental effect, but with appropriate data can
be modeled separately as well.
The early researchers of reared-apart twins did not have the benefit of
the latest sophisticated analytical techniques that were developed over the
last several decades. These methodological tools are among the most im-
portant contributions made by behavioral geneticists to the fields of
psychology and medicine.

Beyond the Laboratories

The human interest of the reunited twins extended beyond their similari-
ties and differences. An especially satisfying part of each assessment week
was watching the twins’ social relationships evolve. Many lives changed
64 B O R N TO G E T H E R — R E A R E D A PA RT

for the better once the twins met and became better acquainted. Sixty-
four-year-old Scottish twins Margaret and Caroline Shand, both unmar-
ried, became friends and housemates. Thirty-nine-year-old British twins
Elaine Alin and Mary Holmes became close companions and confidantes,
especially when Elaine developed breast cancer. Thirty-two-year-old bach-
elors and firefighters Mark Newman and Jerry Levey swapped firehouse
stories over Budweiser beers, each placing a pinky finger underneath
the can. And Tony and Roger enjoyed an all-you-can-eat pizza entrée for
the price of one—these tall, dark-haired, food- and fun-loving twins pre-
tended to be the same person until a waitress baffled by the number of
“seconds” requested by her patron refused to bring more.
 CHAPTER THREE

Early Findings (1979–1983)

Almost any experiment that one might think of doing with

human subjects will be more interesting and yield more valuable
results if one does it with twins.
—David Lykken

Our First Findings

The third International Congress on Twin Studies, held in Jerusalem,
Israel, in 1980, was an opportunity to present preliminary findings from
the MISTRA and to introduce the study to twin research colleagues from
around the world. The first two presentations in Jerusalem were largely
qualitative in nature, quite different from the quantitative treatment of
the data as the sample grew. These early cases suggested that possible
life-history factors were associated with the co-twins’ differences in se-
lected behaviors. For example, twins in one MZA pair did not share their
fear of snakes, a difference traceable to one twin having been bitten as a
child. Periods of heavy drinking were reported by just one twin in several
pairs, an apparent response to major situational stress. Both behavioral
similarities and differences were of interest, and helped to identify topics
for future study.
A paper co-authored by Bouchard and colleagues described the impe-
tus for the project (the discovery of the Jim twins) and outlined a sample
assessment schedule with descriptions of inventories and interviews.1
Some developmental findings and other life-history data were captured
in several illustrative cases. These observations, as they were reported
during the project’s earliest stages in 1980, follow. I have presented two
of these cases to illustrate how data were compiled when the project
began, before it transitioned from a small series of reared-apart pairs to
the MISTRA. This change will become clearer as the story of the study
is told.
66 B O R N TO G E T H E R — R E A R E D A PA RT

Case 1. MZA males, age 23, separated at age 5 days, and reunited at
age 22 years.2
• Both twins were overweight until junior high school, then
became quite thin.
• Both twins were openly and actively homosexual prior to
meeting.
• Both twins had speech problems for which they, respectively,
received therapy in kindergarten and grade school.
• Both twins were diagnosed with hyperactivity in kindergarten or
first grade.
Case 2. MZA females, age 57, separated at age 6 weeks, and reunited
at age 10 years.3
• Both twins had experienced enuresis (bed-wetting) until age 12 to
13 years.
• Both twins had nightmares beginning in their teens, lasting for 10
to 12 years. The dream content included doorknobs and fish-
hooks in their mouths and feelings of being smothered.
• The twins had similar marital and educational backgrounds.
The twins’ similarities in complex behaviors, such as temperamental
traits, sexual orientation, speech problems, and nightmares, were not ex-
pected by Bouchard and the others. “Having been familiar with the litera-
ture on the heritability of temperament, we were not ready for what we
found.”4 The lack of methods for capturing these observations quantita-
tively and objectively placed a limit on evaluating these early findings.
Therefore, one of our future goals was to try to explain these behaviors
with reference to specific features of each twin’s rearing circumstances,
such as their parental relationships, educational opportunities, and home
atmosphere.
In the absence of clear rules or guidelines, deciding what constitutes
co-twin concordance (resemblance) versus discordance (nonresemblance)
for some traits can be difficult and somewhat subjective. We observed
differences between MZA co-twins, but they appeared mostly as “varia-
tions on a theme,” suggesting quantitative rather than qualitative dis-
crepancies. One twin might appear less extraverted, fearful, or fastidious
than his or her co-twin but still display some level of that trait. Designa-
tions of the MZA co-twins as extraverted-introverted, fearful-fearless, or
fastidious-untidy were, therefore, less likely. This dilemma, which occa-
sionally confronts all twin investigators, was revisited by Len Heston in
November 2009 during my interview with him at his home in Palm Springs,
California.
 E A R LY F I N D I N G S 67

Heston coauthored our controversial 1986 paper on the reared-apart

twins’ sexual orientation, which included six twin sets. One MZA male
pair was especially vexing to categorize. One twin was very open about
being gay—he had had two brief heterosexual encounters in his late
teens before becoming exclusively homosexual at age nineteen. He went
on to have sexual contact with seven men, one for a prolonged period of
time. In contrast, his co-twin regarded himself as exclusively heterosex-
ual. He was married and had four children, but he had had an affair with
an older man between the ages of fifteen and eighteen before having his
first heterosexual experience with his future wife at age twenty.
Heston speculated that the twins’ rearing circumstances might have
affected their sexual identities and behaviors. The first twin was raised in
a large city where opportunities for sexual diversity were available and
relatively acceptable. The second twin grew up in a small farming com-
munity in which homosexuality was probably highly stigmatized, mak-
ing him more sexually restrained. “I still don’t know how to count this
[pair],” Heston admitted. Classifying individuals as gay or straight, autis-
tic or nonautistic, or schizophrenic or nonschizophrenic depends on the
diagnostic criteria (strict, intermediate, or broad) that the individual in
question must meet. Knowing that someone has an affected identical twin
can potentially bias this process. Therefore, professional judgments of
behavior are ideally made by several individuals who are blind to the
twin status of the person in question. One of the famous Genain quadru-
plets, studied at the National Institute of Mental Health in the early 1960s,
might not have been diagnosed as schizophrenic if not for her more se-
verely affected sisters.5 A firm decision about the sexual classification of
the reared-apart male twin pair previously described was never made.
I will say more about our sexual behavior studies later.
The second MISTRA paper presented in Jerusalem described preliminary
findings of psychiatric traits among the first fifteen MZA twin pairs (seven
female and eight male).6 The twins ranged in age from sixteen to fifty-seven
years. Each twin was interviewed separately, one on Monday afternoon and
the other on Tuesday afternoon, by psychiatrists Heston and Eckert in ses-
sions lasting from one and a half to three hours. The interview followed a
semistructured format, but Heston and Eckert posed additional questions
about specific issues that the twins described. This procedure changed to a
structured format in 1982 following our decision to administer the Diag-
nostic Interview Schedule to each co-twin by different psychiatrists.
Our early analysis revealed similarities in psychiatric traits among some
pairs, but it was not uncommon for one twin to be more severely affected
than his or her co-twin. Sample findings based on three pairs are presented
68 B O R N TO G E T H E R — R E A R E D A PA RT

in Table 3-1 as they appeared in the original publication. Collectively, we

observed that MZA co-twin differences in psychiatric traits often show
quantitative versus qualitative variation. Statistical analyses of the psychi-
atric data did not occur until 1990 when the reared-apart twin sample
was larger.
The MISTRA participants were not clinical referrals. Years later, Eckert
recalled that “there was not much that was striking in the medical or psy-
chiatric interviews—these were ordinary people who probably represented
the larger general population [with respect to psychiatric symptoms].”
None of the twins showed eating disorders (anorexia or bulimia), Eckert’s
area of expertise. A similar view was also voiced by periodontist Bryan
Michalowicz, who detected little advanced gum disease in the twins.
Most of the MISTRA twins were adoptees, for whom many studies
suggest a higher rate of behavioral problems relative to nonadoptees.7
However, recent work is revising this view. Several studies have reported
that adoptees may be at only slightly greater risk for behavioral disorders
than nonadoptees, and may even score more favorably in some problem
areas (such as social difficulties).8 Regardless, the extent to which the
reared-apart twins represented the greater population was important to
consider if the results were to be generalized. This topic will be addressed
further in coming chapters.
Bouchard presented a third, solely authored paper at the Jerusalem
twin conference.9 It was not about the MISTRA specifically, but some of
his points were relevant to the project. Bouchard argued that general in-
telligence is the major mental ability underlying a variety of mental ability
tests, although he recognized the presence of special mental skills such as
perceptual speed and spatial visualization. As such, he favored using the
Wechsler Adult Intelligence Scale (WAIS) over the Raven Progressive
Matrices as a general ability measure. (The WAIS consists of six verbal
and five nonverbal or performance subtests, whereas the Raven is a non-
verbal test of problem-solving ability and thus is less comprehensive.)
Bouchard also supported using multiple measures of general and special
skills in behavioral genetic studies, especially the MISTRA. He decided to
have the reared-apart twins complete two general intelligence measures
(WAIS and Raven/Mill-Hill composite) plus two special mental ability
batteries (Hawaii Ability Battery and Comprehensive Ability Battery). This
approach would allow assessment of the stability or consistency of par-
ticipant behavior, thereby increasing confidence in the findings.
The climate surrounding genetic studies of IQ in general, and reared-
apart twin studies in particular, was highly contentious. Throughout the
1970s, university faculty members who presented such topics in their
Table 3-1. Preliminary Summary of Psychiatric Traits in Three MZA Twin Pairs

Alcohol,
Drug Fears, Speech Enuresis
Pair Twin Age Sex Abuse Phobias Problem (bed-wetting) Other Psychiatric Traits

5 A 35 F + + Adjustment reaction with anxious

mood
B + ++ Same as above
10 A 16 M + + Emotionally stable
B + ++ Moody, nervous, sociopathic traits,
hyperactive
13 A 24 M + Emotionally stable
B + Emotionally stable

+ = Abnormality present; ++ = Abnormality more pronounced.

Source: Adapted from Elke D. Eckert et al. (1981), “MZ Twins Reared Apart: Preliminary Findings of Psychiatric Disturbances and Traits,” in Twin
Research 3: Part B. Intelligence, Personality and Development, ed. Luigi Gedda, Paolo Parisi, and Walter Nance, 179–188 (New York: Alan R. Liss, Inc.,
1981).
70 B O R N TO G E T H E R — R E A R E D A PA RT

classes were subject to ridicule, even harassment. “Fire Bouchard” had

been painted on the outside of a building close to Elliott Hall.10 The cam-
pus newspaper referenced the attack on Bouchard from the Students
for a Democratic Society in an editorial defending freedom of speech on
campus.11 Bouchard persisted in teaching and in researching these topics,
as did many of his colleagues.12
The 1974 Behavior Genetics Association meeting in Minneapolis orga-
nized an impromptu session in which four members (including Bouchard)
discussed their experiences teaching behavioral genetics concepts and
findings as they applied to IQ. Once behavioral genetics entered the main-
stream of psychology in the 1980s, there was less need for such events, al-
though evidence of genetic influence on intelligence was still challenged.13
However, a planned 1992 conference on genetics and criminality was
forced to reschedule in 1995 due to protests over possible misuse of the
findings.14

Constructive Replication: Challenges to Reared-Apart Twin Research

The MISTRA’s inclusion of multiple measures of general intelligence
allowed attempts at constructive replication, a concept developed by
Lykken.15 Constructive replications are efforts to confirm findings when
one or more conditions of the original experiment or analysis differ, such
as the specific test or the participant sample. Such replications may be
conducted by the original researcher or by an independent investigator.
In 1980, Princeton University sociologist Howard Taylor published The
IQ Game, in which he reanalyzed IQ data from the three previous reared-
apart twin studies.16 Taylor was interested in whether MZA co-twins were
more alike in intelligence if they were separated after 6 months, reunited
prior to testing, reared by biological relatives, and/or raised in similar en-
vironments. Based on his analysis of the reared-apart twin data from the
three early studies, Taylor concluded that genes contributed little to indi-
vidual differences in measured general intelligence.17
Taylor’s conclusion was stunning, coming as it did in the midst of the
IQ debate over whether, and to what extent, genes affect mental ability.
It turned out that the studies by Newman and colleagues and Juel-Nielsen
had included a second major general intelligence measure. In 1983,
Bouchard applied Taylor’s criteria for partitioning the early reared-apart
pairs according to separation, rearing, reunion, and contact, but with
reference to the alternative general intelligence measure. Bouchard was
unable to replicate Taylor’s findings—time of separation, age at reunion,
relatedness of rearing relatives, and environmental similarities were un-
related to IQ.18
 E A R LY F I N D I N G S 71

In 1981, clinical psychologist Susan L. Farber continued Taylor’s thesis

that increased contact should lead to increased MZA IQ resemblance.19
Her book was both highly acclaimed and solidly rejected by many in the
field, a good reflection of the intense divisiveness on questions concern-
ing the genetic basis of ability and the soundness of twin methodology.20
As was true of Taylor, Farber’s views are still embraced by critics of twin
methodology, despite the available data that challenge those views.21
However, her book remains a useful compendium of reared-apart twin
data organized into chapters covering intelligence, personality, psychopa-
thology, and physical complaints.
As recently as 2007, statistics professor Jack Kaplan, in a 2007 letter
to the New York Review of Books, raised many of the same objections as
Taylor.22 Author and science historian Frank Sulloway responded, “[Ka-
plan] seems unaware that these issues have all been addressed empirically
during the last two decades and are no longer regarded as ‘highly ques-
tionable.’ ”23 A more recent challenge to twin studies of IQ heritability
appeared in a 2009 book by University of Michigan psychology profes-
sor Richard Nisbett.24 Nisbett claimed, “The direct estimate of heritabil-
ity based on the correlation between the IQs of identical twins reared
apart makes a tacit assumption that is surely false—namely, that the twins
were placed in environments at random.”25 Nisbett’s other objection,
that MZ twins’ shared intrauterine environment enhances their behavioral
similarity, is also raised frequently.26 In fact, adversities posed by MZ
twins’ common prenatal environment more often have the opposite ef-
fect.27 That is, MZ twins experiencing shared blood supply, fetal crowding,
or other unfavorable circumstances are likely to differ in size and health.28
The finding that some reared-apart twins may have had similar rearing
environments does not negate the notion of random placement. In fact,
the expectation is that some co-twins would be reared in similar environ-
ments if placement were truly random. The MISTRA twins did not meet
strict requirements for complete randomization of placement, but their
environments were far less alike than those of twins reared together. Our
analyses showed mostly negligible placement effects on the different traits
and on co-twin resemblance in these traits.
Furthermore, many twin and adoption studies show that common rear-
ing contributes only modestly to intellectual development and only during
childhood. I found that forty-three virtual twin pairs (same-age unrelated
children, raised together since infancy who mimic twinship but without
any genetic link) showed an IQ correlation of 0.30 at age 5.1 years, and
a correlation of 0.11 at age 10.7 years.29 Other research has shown that
the IQ correlation for different-aged adoptive siblings approaches 0.00
when they reach adolescence.30
72 B O R N TO G E T H E R — R E A R E D A PA RT

The MISTRA at Age Two

In the year 1981, data had been gathered on fourteen reared-apart twin
pairs and two reared-apart triplet sets. The first quantitative analysis of
MZA twin data was presented at the eleventh annual meeting of the
Behavior Genetics Association, held at the State University of New York
in Purchase. That paper described the twins’ information processing bat-
tery, discussed the twins’ similarities in these measures, and contrasted
them with similarities on a comparable psychometric battery.31 (An ex-
panded version of this paper, published in 1984, is discussed in the next
chapter.) Twenty-four MZA twin pairs had been assessed when the con-
ference took place, so, with only one interesting exception in 1990, the
case studies from the MISTRA were replaced by statistical presentations
of grouped data.
The year 1982 continued the transition from case descriptions to
quantitative summaries of early findings, mostly embedded in empirical
papers, review articles, and book chapters as the number of reared-apart
twin pairs increased. Early findings on IQ and on the clinical and con-
tent scales of the Minnesota Multiphasic Personality Inventory (MMPI)
were presented at the 1982 Behavior Genetics Association meeting in Ft.
Collins, Colorado. These results relied on small samples, but they started
a trend that evolved into a more stable pattern of genetic effects.32
Many of our early findings also came from Lykken’s psychophysiological
laboratory.

Lykken’s Laboratory
David Lykken’s 1981 presidential address to the Society for Psychophysi-
ological Research was a comprehensive overview of reared-together twin
studies that included preliminary findings from the MISTRA.33 Putting
these different sources of data in the same tables provided hints as to
how much genes and shared environments affected behavioral variation.
Lykken began by summarizing age, age at separation, and age at reunion
for the three previous twin studies and the MISTRA, as shown in Table
3-2. By 1982, our MZA sample was second in size only to Shields’s. Twins
in the MISTRA were separated earlier and reunited later, on average, than
those in the other three studies.
The MZA and MZT correlations for fingerprint ridge count and for
height, shown in Table 3-3, were quite similar, and both were much higher
than the corresponding DZT correlations. These data conformed to pre-
dictions based on a model in which the genetic effects reflect the adding
 E A R LY F I N D I N G S 73

Table 3-2. The Four Major Studies of MZA Twins Reared Apart

Age When Age at Age at

Study Studied Separation Reunion

Newman et al. (1937), n =19 26.1 1.6 12.5

Shields (1962), n = 44 38.8 1.4 11.0
Juel-Nielsen (1965), n = 12 51.4 1.5 15.9
MISTRA, n = 30 36.1 0.3 23.9

Source: Adapted from Lykken, “Research with Twins: The Concept of Emergenesis,”
Psychophysiology 19 (1981): 361–373.

Table 3-3. Intraclass Correlations for Three Anthropometric Variables (1982)

DZT MZT MZA

Variables (n = 146) (n = 274) (n = 30)

Fingerprint ridge count 0.46 0.96 0.98

Height 0.50 0.93 0.94
Weight 0.43 0.83 0.51

Note: The reared-apart twins’ height, weight, and fingerprint data were gathered in
David Lykken’s laboratory. Height, weight, and fingerprints were also gathered in Thomas
Bouchard’s laboratory; and height and weight were additionally measured during the
general physical examination. The handprints that were also obtained in Lykken’s
laboratory are not presented here.
Source: Adapted from Lykken, “Research with Twins: The Concept of Emergenesis,”
Psychophysiology 19 (1981): 361–373.

up of many genes. In contrast, the .51 heritability estimate for weight,

based on the MZA twins’ weight correlation (a direct estimate of herita-
bility), was considerably lower than the 0.80 heritability estimate result-
ing from doubling the difference between the MZT and DZT intraclass
correlations. This finding suggested possible violation of the equal envi-
ronments assumption,34 but, as I have indicated, the equal environments
assumption has been upheld in most twin studies.35 The discrepancy be-
tween the MZA and MZT correlations suggested that weight is affected
by shared environmental factors.
The most important feature of our early findings was that they were
generally similar to those from reared-together twin studies. We found
this to be very encouraging because our twins were not gathered systemati-
cally as were most twins in the reared-together studies. Bouchard pointed
out that because the MISTRA participants were chosen as they were it was
“crucial” to show that they were not aberrant on any measures, relative
to twins from other studies.36 We were also able to show that the MZA
74 B O R N TO G E T H E R — R E A R E D A PA RT

and MZT twins were comparable for the electroencephalogram (EEG)

parameters (brain wave spectra) that we measured.
Each twin spent three to four hours in Lykken’s psychophysiology
laboratory. Both twins arrived at Lykken’s laboratory on a Monday af-
ternoon for individual and joint photographs. Lykken liked taking his
own pictures because he enjoyed studying the twins’ physical features in
detail, and anyone who visited Lykken’s laboratory will recall the rows
of matched faces, eyes, and ears that decorated his walls, appearing on
the wall opposite his photographs of the reared-apart pairs. (This com-
pelling display is reproduced in Figure 3-1.) However, only one twin was
assessed on Monday afternoon; the second twin, who was studied the
following day, was escorted to the University of Minnesota hospital com-
plex for medical and psychiatric interviews. Findings did not differ be-
tween twins as a function of their order of assessment.
A significant component of Lykken’s assessment was the EEG read-
ings.37 Some earlier work, including a 1974 study by Lykken, showed a
genetic influence on EEG spectra, but little data were available on twins
in general and very little on separated pairs.38 Twins in just one of Shields’s
sets had been tested, and both showed the same spike and wave patterns,
despite their discordance for epilepsy. Nine of Juel-Nielsen’s pairs had
been tested and showed “complete concordance.”39
Lykken used a six-channel polygraph to record the EEG data from
each twin. Five channels were dedicated to brain wave recording from
the occipital lobes and at the temporal lobes. The sixth channel was used
to remove eyeblink artifacts from the EEG recordings.40 Participants sat
alone in a shielded booth. A resting EEG was taken for the first five min-
utes while the twin sat quietly with eyes closed. Next, twins listened to
five minutes of jazz music, followed by a Shafer auditory event-related

Figure 3-1. Reared-apart

twins’ eyes and ears.
(Photographs by David
Lykken.)
 E A R LY F I N D I N G S 75

potential41 and a visual event-related potential. (Event-related potentials

are brain activities during cognitive processing.) A five-minute resting EEG
was recorded again at the end of the session.
The data were examined for four parameters characterizing the EEG
spectrum. The four parameters were defined by the mean activity within
each of the classic EEG frequency bands: delta, theta, alpha, and beta.
A fifth parameter, known as phi, is the median frequency of alpha activ-
ity. These measures were obtained at the start and at the end of each
session.
The EEG data for the first four measures generally suggested a simple
additive model because the DZT correlations (0.30–0.50, n = 53) were
approximately half the size of the MZT (0.72–0.86, n = 89) and MZA
(0.61–0.93, n = 25) correlations. The mean correlations for the four clas-
sic parameters were .40, .81, and .81 for the DZT, MZT, and MZA twin
pairs, respectively. The exception was phi, which Lykken explained by
emergenesis (the effects of multiple genes on complex traits), due to the
marked correlational difference between the DZT twins (0.13) and both
the MZT (0.81) and MZA (0.89) twins. The MZA EEG data mimicked the
MZT EEG data, showing that the two groups were comparable on this
measure.
Bill Iacono, University of Minnesota Regents Professor of Psychology
and Lykken’s former colleague, believes that Lykken’s 1982 published
address had a real impact on the way psychophysiologists thought about
their measures at that time. That is because his findings drew attention to
a heritable component that was difficult to ignore. Iacono was less enthu-
siastic about the lasting influence of the actual EEG findings because the
methods used today are quite different—today one would apply multi-
variate imaging techniques.
Lykken’s psychophysiological assessment also included studies of elec-
trodermal response (skin conductance after exposure to a series of auditory
stimuli). He presented only brief descriptive MZA findings in his 1982
paper—we are told that of the twenty-seven MZA twin pairs who were
tested, only four pairs showed discordant electrodermal habituation
curves. A more comprehensive analysis of these data, based on a larger
twin sample, was completed by Lykken in 1988 and is discussed in Chap-
ter 6.42
Another component of Lykken’s test battery (that allowed constructive
replication within the project) was the Raven Progressive Matrices and
Mill-Hill. Both tests were administered by a PDP-12 computer as un-
timed (“power”) tests.43 Their sum yielded a separate IQ estimate, based
on norms from over 130 twin pairs (Table 3-4).
76 B O R N TO G E T H E R — R E A R E D A PA RT

Table 3-4. Intraclass Correlations for the Raven and Mill-Hill Intelligence Tests for Reared-Apart and
Reared-Together Twin Pairs

MZA MZT DZT

Intelligence Measures (n = 29) (n = 71) (n = 42)

Raven 0.58 0.66 0.19

Mill-Hill 0.78 0.74 0.37
IQ (R + M) 0.71 0.78 0.14
Raven/Time 0.64 0.72 0.39
Mill-Hill/Time 0.71 0.76 0.49
IQ(R/T + M/T) 0.71 0.80 0.39

Source: Adapted from Lykken, “Research with Twins: The Concept of Emergenesis,”
Psychophysiology 19 (1981): 361–373.

The MZA correlations did not differ substantially from the MZT
correlations, and both exceeded the DZT correlations. The composite IQ
(R + M), like the Raven, produced DZT correlations that were much less
than half the MZA and MZT correlations, suggesting emergenic effects.44
The Raven and composite IQ scores, when converted to “rate-of-processing”
scores by dividing each twin’s score by their average response time, pro-
duced correlations consistent with an additive genetic model.
These data showed that the MZA twins were not atypical in general
mental ability compared with the MZT twins. The MZA twin sample was
small at this stage in the study, but the 0.71 composite IQ correlation
would remain stable as we tested new pairs and replicated findings from
the three early reared-apart twin studies.

Auke Tellegen, one of the primary MISTRA investigators, had spent

twelve years, from 1970 to 1982, devising the Differential Personality
Questionnaire. This form (now the Multidimensional Personality Ques-
tionnaire) yielded scores for eleven personality traits. A full analysis will
be presented later, but an early look at selected scales suggested that so-
cial potency (forceful and persuasive versus letting others take charge)
and positive affect (well-being and achievement versus depression and
nonpleasurable engagement) were emergenic traits (Table 3.5). It was also
striking that the MZA twins (0.74, n = 28) were more alike in Conserva-
tism than the MZT twins (0.58, n = 247), a finding we did not obtain for
the other traits. There was no immediate explanation for this result, but
it was intriguing in 1982 because of the following relevant findings in the
literature.
In 1964, Gerrit J. S. Wilde reported that MZT twins living apart for
five years were more alike in psychosomatic complaints than twins living
 E A R LY F I N D I N G S 77

Table 3-5. Intraclass Correlations for Four Differential Personality Questionnaire Measures for
Reared-Apart and Reared-Together Twin Pairs

MZA MZT DZT

DPQ Measure (n = 28) (n = 247) (n = 122)

Social potency 0.67 0.65 0.07

Conservatism 0.74 0.58 0.44
Positive affect 0.57 0.63 0.02
Negative affect 0.65 0.67 0.43

Source: Adapted from Lykken, “Research with Twins: The Concept of Emergenesis,”
Psychophysiology 19 (1981): 361–373.

together, with the reverse found for extraversion.45 Ten years later, Scot-
tish psychologist Gordon Claridge reported that MZT twins living apart
for five years were more alike in extraversion and divergent thinking
than MZT twins living together.46 Shields had found that his MZA twins
were more alike in extraversion and neuroticism than his MZT twins.
Shields and Claridge suggested that because twins living apart do not
need to differentiate from their co-twin, they could express their gene-
tically based potentials more freely than those living together. Whether
this effect was trait-specific and whether it would appear in later MISTRA
data analyses were of interest. However, with larger samples in 2003 and
2004 no difference in conservatism between the MZA and MZT twins
was found, making it likely that the high MZA correlation in 1982 re-
flected small sample fluctuation.
Lykken devised a Recreational Interests Survey that yielded eight differ-
ent interest factors, four of which he discussed in his presidential address
(Table 3-6). The MZA twins (0.65–0.80, n = 28) were more alike than the
MZT twins (0.46–0.55, n = 140) on three of the four measures (blood
sports: hunting and trapping; intellectual: reading and self-educating;
and husbandry: making and fixing things; but not Sierra Club: camping
and wilderness trips), a finding Lykken attributed to sampling bias. But
as Shields suggested, perhaps twins living apart did not feel pressured
to differentiate from their co-twin. Both the MZA and MZT twins
were more alike than the DZT twins, with the exception of the Blood
Sports measure, for which the DZT twins (0.68, n = 70) were somewhat
more alike than the MZT twins (0.54). Mutual influence may have af-
fected the DZT twins on this measure. Organizing the twins by sex
would have been of interest with reference to Blood Sports, but the
sample size was too modest at this stage. These data were analyzed again
in 1993.
78 B O R N TO G E T H E R — R E A R E D A PA RT

Table 3-6. Intraclass Correlations for Four Recreational Interests for Reared-Apart and Reared-Together
Twin Pairs

MZA MZT DZT

Interest (n = 28) (n = 140) (n = 70)

Blood sports 0.80 0.54 0.68

Sierra Club 0.49 0.57 0.28
Intellectual 0.69 0.46 0.42
Husbandry 0.65 0.55 −0.07

Source: Adapted from Lykken, “Research with Twins: The Concept of Emergenesis,”
Psychophysiology 19 (1981): 361–373.

Pulmonary Studies
Concern over the physical consequences of smoking cigarettes was a ma-
jor concern in 1982, and it still is. The MISTRA enabled the first pulmo-
nary function study of twins reared apart in an attempt to address the
genetic factors.47
The twins underwent thirteen pulmonary tests administered by the same
technician. They blew into a spirometer, a device measuring the volume of
air that is inhaled and exhaled, and completed a respiratory symptom
questionnaire. Six twin pairs were concordant for nonsmoking, three
pairs and the triplet set were concordant for smoking, and six pairs were
discordant for respiratory problems (five pairs for smoking and one pair
for asthma).
Only two of the thirteen respiratory tests—activity in the lower air-
ways48 and change in lung volume over time49—separated the smokers
from the nonsmokers. However, a third test measuring the volume of air
expelled during the first second of forced exhalation50 provided the most
sensitive measure of difference between smoking and nonsmoking twins
within a pair. Genetic factors affecting this third test were indicated by
small pulmonary differences within the smoking-concordant and non-
smoking concordant twin pairs, in contrast with the marked differences
between twins in the smoking-discordant pairs.
Our findings and those of others demonstrated that genetic factors af-
fect some measures of susceptibility to airway obstruction from cigarette
smoke.51 However, it was puzzling that the groups of smoking concor-
dant and discordant pairs did not differ on the other pulmonary mea-
sures. Given the vast research linking respiratory problems to smoking, it
would be wrong to conclude from just one study that cigarettes did not
interfere with respiratory functioning.
 E A R LY F I N D I N G S 79

Visit from Niels Juel-Nielsen

There is a videotape that could become a classic in the history of psychol-
ogy, but probably no one watched it until I converted the clunky three-
quarter inch tape to DVD in November 2011. In May 1981 Bouchard
had invited Dr. Niels Juel-Nielsen, the Danish reared-apart twin investi-
gator, to visit the MISTRA, at which time a discussion between the two
was recorded. Like Bouchard’s, Juel-Nielsen’s sudden discovery of a set of
reared apart co-twins captivated his interest in studying their behavioral
similarities and differences. And as happened in Minnesota, the media’s
attention to this first case attracted several others, although Juel-Nielsen
drew most of his pairs from the Danish Twin Registry. Finally, like
Bouchard, Juel-Nielsen was “impressed” by the twins’ unexpected simi-
larities and struggled with defining their differences in a practical sense.
Bouchard was rarely in the position of interviewer as he was in this
particular setting. Still, he explained that his reading of Newman et al.’s
study convinced him of the beauty of the twin research design—“cutting
and definitive”—although “the neatness of the design is overwhelmed
by the complexity of the twins’ lives.” By then, Bouchard had studied
twenty-one MZA twin pairs and “had not settled down and absorbed”
the indication that MZAs were about as similar in many ways as MZTs,
challenging current views of shared environmental effects on behavior.
“How long will it take me?” he asked.

New Twin Pairs and the New Minnesota Center

for Twin and Adoption Research
Even while we were all paying more attention to the statistical findings,
we remained fascinated by the compelling stories of the newly identified
twin pairs. A lasting contribution to the new reared-apart twin “culture”
had occurred in the early years with the 1981 publication of a Charles
Addams cartoon. Undoubtedly inspired by MISTRA findings reported in
the press, the cartoon depicts the now famous, fictional Mallifert twins,
who were identical in every way (Figure 3-2). None of our pairs were as
alike as the Malliferts across all measured traits, but the individual ele-
ments of this scene were accurate. For example, Lucky and Dianne, red-
headed and vivacious, shared a love for dogs: one showed dogs profes-
sionally, and the other taught obedience classes. Barbara and Daphne
had the same naturally crooked pinky fingers and distaste for politics.
Several DZA twins also displayed striking similarities—male twins
Bouchard discovered in 1984 had both decorated their bodies with sixteen
80 B O R N TO G E T H E R — R E A R E D A PA RT

tattoos—but such observations were seen less frequently among the DZA
twins.
When Kimerly (Kim) Wilcox and I arrived in 1982 as MISTRA post-
doctoral fellows, we both felt extremely lucky to be at exactly the right
place at the right time. The data we worked with was fresh and new, and
people cared about it.
When I had been in Minnesota for about a year, some faculty and staff
organized a birthday party for Bouchard. Kim and I designed a t-shirt
for him with the number “.771.” This number was the critical MZA IQ
correlation that had been repeated in several of Sir Cyril Burt’s updated
analyses, which was still regarded with suspicion. Very early on and for a
brief time, the MISTRA IQ correlation was 0.771 as well. Our presenta-
tion of Bouchard’s t-shirt was a success, although some eyes rolled. That
figure is worth mentioning because it comes up again.

Figure 3-2. “Separated at birth, the Mallifert twins meet accidentally.”

(© Charles Addams, with permission from the Tee and Charles Addams
Foundation.)
 E A R LY F I N D I N G S 81

In 1983, Bouchard established the Minnesota Center for Twin and

Adoption Research, a named unit within the Department of Psychology.
Its goals were to provide information about twins and adoption, assist
separated twins in search of their co-twins, provide administrative sup-
port to twin-based projects, and encourage understanding of the biologi-
cal and social aspects of twinship. Most importantly, the center gave sepa-
rated twins the resource they needed to find one another.
 CHAPTER FOUR

Sexual Orientation, Cognition,

and Medical Traits (1984–1987)

I n 1984, I had been working on the MISTRA for two years. Every
pair I studied provided a unique take on nature-nurture questions.
MZA twins Bill and Tim showed the same side-to-side sway as
they walked. MZA twins Stan and Mitch were both homosexual despite
being reared apart their entire lives. Greater insight into the twins’ be-
havioral and physical similarities and differences began to emerge from
the analyses we conducted between 1984 and 1987. Many of our find-
ings continued to be reported in the press.
The MISTRA had matured greatly since my arrival in 1982. The num-
ber of empirical papers in scientific journals, relative to descriptive ac-
counts in book chapters and review papers, was growing. By the start
of 1984, we had studied fifty-four reared-apart twin pairs (thirty-eight
MZA and sixteen DZA). Interesting trends were emerging in information
processing, cardiac functioning, immunology, ophthalmology, and sexual
preference, all of which we reported in psychological or medical journals.
Chapters and reviews now included tables with preliminary statistical
findings on physical characteristics (height, weight, and fingerprint ridge
count) and psychological traits (general intelligence, special abilities,
personality, and interests). Previous studies of twins reared apart and to-
gether offered a backdrop against which to appraise these early results.
The Minnesota Center for Twin and Adoption Research, established in
1983, was in its first full year of operation. By the end of 1984, we had
added seven new twin pairs and one new triplet set to the study. The sam-
ple now included sixty-four separated sets (forty-four MZA and twenty
DZA), the largest number of twin pairs studied by any research team.
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 83

Five papers and three book chapters were published between 1984 and
1987, including one of the first on information processing.

Information Processing
Studies in the early 1980s had found relationships between the rate of
performing tasks involving memory scanning and psychometric cognitive
ability measures, although the reasons were uncertain.1 Genetic effects on
information processing parameters and on cognitive measures were also
of interest, but had never been examined using twins reared apart. The
MISTRA had the data to address both these issues.2
The twins completed three information processing measures: Posner
Letter Identification,3 Sternberg Memory Search,4 and Shepard-Metzler
Cube Rotation.5 The Posner measure presents two letters—in a physical
identity condition (e.g., A, A) or in a name identity condition (e.g., A, a)—
that participants must identify. Name identity minus Physical identity is
the difference in time between presentation of the problem and response
in the two conditions. The Sternberg measure asks participants to memo-
rize one, three, or five digits from 0 to 9. Participants then indicate if a
number that appears on the screen several seconds later is part of that
series.6 The Shepard-Metzler measure presents pairs of three-dimensional
cubed figures on a two-dimensional screen. The two figures may be the
same but shown from different angles, or they may be different figures.
Participants decide if the pairs are the same or different.7
Each twin was tested separately, usually by the same examiner. Given the
straightforward instructions, push-button response procedure, and correct/
incorrect response format, meaningful tester bias was unlikely. The or-
der in which the tasks were presented was the same across participants:
(1) Posner, (2) Sternberg, and (3) Shepard-Metzler. Testing usually took
place over two days, with most items administered on the first day (ap-
proximately 90 minutes) and the second set of Shepard-Metzler items ad-
ministered on the second day (approximately 30 minutes). Testing never
began until the twins were comfortable with the concepts and procedures.
Graduate students Dan Moloney and Mary Moster administered the
Shepard-Metzler Rotations to many of the twins between 1984 and 1990.
Dan liked doing this because he was good at these problems and solved
them silently as the twins worked. When I spoke with him by telephone
in April 2010, he said that he regretted not tape-recording the twins during
these sessions. “I sat quietly behind them,” he said. “The strategies were so
different between twin pairs, but within the MZA pairs they were so
similar. Both twins vocalized or turned around or stared at the screen or
84 B O R N TO G E T H E R — R E A R E D A PA RT

solved the problems quickly. It was amazing. I smiled to myself when I

saw these things, thinking no one would believe me. When the twins were
successful or frustrated they both reacted a certain way and maybe there
is a genetic component to such behavior.”
In addition to the information processing tests, the twins completed the
Wechsler Adult Intelligence Scale (WAIS), and selected subtests from
the Hawaii Study of Cognition and Kit of Factor-Referenced Cognitive
Tests. The availability of these mental ability data let us analyze the twins’
information processing abilities as a two-part study. Part 1 examined re-
lationships between the information processing measures and the special
ability and general intelligence measures. Part 2 looked at the relative con-
tributions of genetic and environmental factors to individual differences in
information processing skills. These analyses included twenty-eight to
thirty-one MZA twin pairs and eleven to thirteen DZA twin pairs. Many
detailed results were presented in the information processing paper, so
only a selective sampling is provided here.
The information processing data were organized into three compo-
nents: overall speed of response (29.8 percent), speed of information pro-
cessing (20.8 percent), and speed of spatial processing (16.7 percent); the
percentages reflect the proportion of variation in information processing
explained by these components. Overall speed of response correlated
significantly with the general cognitive ability measures (r = −0.31), ver-
bal reasoning (r = −0.18), and perceptual speed and accuracy (r = −0.20).
The negative correlations mean that faster responding was associated
with higher test scores. Speed was associated with both general and spe-
cific mental abilities, supporting the view that processing speed is a major
part of intelligence. Speed of information (memory) processing correlated
significantly with general cognitive ability (r = −0.32) and with verbal
reasoning (r = −0.27). Speed of spatial processing correlated mostly with
spatial processing measures (r = −0.32).
Significant MZA twin resemblance for information processing mea-
sures tapping general speed of response was found (ris = 0.37 to 0.60). The
MZA correlations were also significant for the percentage correct on the
Sternberg (ri = 0.79) and Shepard-Metzler tasks (ri = 0.62). In contrast, ge-
netic influence was not indicated for measures reflecting the speed of spe-
cific cognitive processing.8 The DZA twin sample was too small to allow
reliable conclusions regarding genetic effects.
The reared-apart twin data enabled many information processing anal-
yses beyond the MZA-DZA comparisons. For example, the presence of a
verbal-spatial distinction for the information processing measures, similar
to that found for the ability measures, was supported. Thus, the reared-
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 85

apart twins contributed to the debate over the nature of information

processing components and their relationship with general intelligence
and special mental skills. The significant MZA intraclass correlations for
speed of response measures were exciting to see, supporting the idea that
processing speed contributes importantly to general intelligence.

Cardiac Functioning
Franz Halberg is founder and director of the Chronobiology Laboratory
at the University of Minnesota. Chronobiology is the study of cyclical
phenomena and their adaptation to lunar- and solar-related rhythms.
Halberg had heard about the MISTRA from psychiatrist Elke Eckert and
was immediately interested in studying the twins. When we began work-
ing with Halberg, the cyclical features of the human heart had been well
documented, but less had been done regarding genetic factors.
Longitudinal measurements of the human heart have established the
presence of several components, such as ultradian rhythms (having a
frequency of greater than one cycle in twenty hours), infradian rhythms
(having a frequency of less than one cycle in twenty-eight hours), and
circadian rhythms (having a frequency of one cycle in approximately
twenty-four hours). Even when people are in social isolation the heart
maintains its circadian rhythmicity, consistent with genetic origins. Only
one very small twin study of circadian rhythms, showing that pulse was
genetically influenced, was available when the MISTRA began.9 Thus,
the reared-apart twins presented “a new challenge” to our colleagues in the
departments of laboratory medicine (Bruce Hanson), chronobiology
(Franz Halberg), and cardiology (Naip Tuna).10
Every Wednesday morning, twins visited the University of Minnesota’s
Variety Club Heart Hospital to be hooked up to a twenty-four-hour Holter
monitor that records heart activity continuously.11 The device hangs by
a strap over the shoulder, much like a travel bag. Twins were also given
booklets for recording their meal times and significant changes in physical
activity, such as walking up stairs. Showering was forbidden during this
period. Each twin of a pair was monitored during the same week, and usu-
ally on the same day.
We compared the twins’ resemblance in the mesor (rhythm-adjusted
mean), amplitude (the extent of rhythmic change), and acrophase (timing
of rhythm in relation to a reference point).12 The sample included eigh-
teen MZA twin pairs (eleven male and seven female) and four DZA twin
pairs (two male and two female); three MZA male pairs were generated
from a set of triplets.
86 B O R N TO G E T H E R — R E A R E D A PA RT

The MZA intraclass correlations for the three cardiac parameters

ranged from 0.61 to 0.68, showing that genetic factors explained about
60 to 65 percent of the variance in these measures. (The small DZA twin
sample could not provide reliable results.) When we organized the data
by sex, the mesor and amplitude correlations stayed significant for males,
but the mesor correlation dropped to near zero for females. The possibil-
ity that genetic effects on cardiac measures work differently for men and
women was raised by these data, but needed to be confirmed with larger
samples. Age was eliminated as a factor affecting twin similarity because
the correlations between unrelated same-age twins created from the sam-
ple were negligible.
Thus, the previously indirect indication of genetic influence on circa-
dian rhythms among individuals in isolation was confirmed directly.
Hanson and colleagues concluded that the detection of genetic effects in
“a variable as noisy as heart rate,”13 studied over a brief period and with
a small sample, is methodologically meaningful. When I spoke to Hanson
in 2009, he wondered whether the MZA twin data were driven by the
twins’ similar activities during their study week. Of course, the DZA twins’
activities were also synchronized, but the small DZA sample prevented
this comparison. We did establish that rhythm parameters did not differ
between twins studied on the same versus different days.
Halberg continued studying genetic effects on the circadian amplitude
of the human heart rate with a larger number of reared-apart twin pairs
(thirty-five MZA and twenty-two DZA). In a 2003 study on circadian
findings, he included a graph showing a significant heritable effect (MZA
ri = 0.55).14 In contrast, the DZA correlation was negligible, suggesting
an emergenic effect. Halberg was also interested in blood pressure and
regretted that we had recorded the twins’ measurements only over one
night rather than the full week. However, extended blood pressure re-
cordings would have been difficult with the large and noisy machines of
the time (recordings can be made less intrusively today). In fact, Halberg
insists that people should wear blood pressure monitors every day of
their lives—“womb to tomb” is what he said, and I believed he was seri-
ous. I am certain that he would have liked monitoring the twins’ heart
activities for the entire week as well.
One DZA female twin (not included in the present analysis) decided to
have sex on the night that she wore her Holter monitor. Her twin sister
did not, even though her spouse was also present. To the extent that our
heart rates are affected by sexual activities, these DZA twins would have
shown differing numbers during the nighttime hours, partly associated
with their different choices that night.15 I wish I had taken a closer look
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 87

at the printouts they handed to us the next morning, but we were rushing
to get the twins to breakfast and to the allergy laboratory.

Immunology
Only a handful of conventional twin studies examining immunological
regulation were available when the Jim twins were found. Genetic fac-
tors in total levels of immunoglobulin or antibodies had been reported,
but these measures had never been studied in reared-apart twins prior to
the MISTRA.16 Total immunoglobulin refers to the basic protein struc-
ture of the immune system and is expected to be similar in MZ twins.
Antibody titers reflect different challenges to the immune system.17 If one
co-twin had hepatitis and the other co-twin did not, then their different
antibody titers should reflect this difference.
The immunological information often took weeks to process—a portion
of the blood sample used in the immunological studies was processed in
Denver where our collaborator, Peter Kohler, had his laboratory. The five
classes of immunoglobulins (IgG, IgM, IgA, IgD, and IgE) provide resis-
tance by the release of antibodies. The immunoglobulin (Ig) measures in-
cluded total IgM, IgG, and IgA levels. Twin similarity in specific antibodies
to tetanus toxoid and to polyvalent (several antibodies that counteract
specific antigens) and other substances was also assessed.18 Sera (blood
with clotting factors removed) from twenty-six MZA twin pairs and ten
DZA twin pairs were used in the first paper from this laboratory.
The results from our twin analysis were compelling: “In spite of differ-
ent environmental exposures to antigens, the predominant factor(s) de-
termining total immunoglobulin and isotypic antibody levels in these
twins was genetic rather than environmental.”19 Occasional differences
within selected MZA twin pairs were linked to marked variation in teta-
nus toxoid immunizations and exposure to Streptococcus pneumococci,
as expected.
There is a bigger story here: the MZA twins’ greater similarity in total
concentration (Ig: MZA 0.80 to 0.82, DZA −0.40 to 0.35) than antibody
titers (IgA: MZA 0.25, DZA 0.17; and IgM: MZA 0.59, DZA 0.54), and
their reduced similarity in response to specific antibodies suggested that
genetic regulation plays a greater role in the former. Based on this work,
future research would probably show a stronger genetic effect from anti-
body responses to immunization with defined antigens. If MZA co-twins
had both been vaccinated for smallpox or hepatitis B then their reaction
to specific antibodies should be similar. Knowing that genetic factors
can affect the body’s response to toxic substances may improve disease
88 B O R N TO G E T H E R — R E A R E D A PA RT

prevention and treatment, especially for people whose relatives show

resistance to certain vaccines or sensitivity to specific antigens.

Ophthalmology
The MISTRA twins’ visual characteristics were studied by the late Dr.
William H. Knobloch, from the University of Minnesota’s Department
of Ophthalmology. Eye examinations were scheduled for 1:00 p.m. on
Wednesday afternoons and took several hours to complete. The twins’
eyes were dilated during the course of this visit, which made them unable
to read and complete the questionnaires. They left the hospital wearing
protective eyeglasses and were taken by Bouchard on a tour of the Twin
Cities’ lovely lakes. How much scenery they actually saw has never been
determined.
The twins’ eyes were examined at the same time by separate ophthalmo-
logic technicians. Each twin completed a brief eye history, including ques-
tions about their visual health during childhood. Next, assessments were
made of their visual acuity (clearness of vision), refraction (ability of the eye
to bend light so an image is focused on the retina), intraocular pressure (fluid
pressure inside the eye), and muscle balance (ability of the eyes to work to-
gether). Dilation of the eyes enabled inspection of the anterior segment
(front third of the eye) and fundus (interior surface opposite the lens), pro-
cedures completed by an ophthalmologist. Stereo photographs of the optic
nerve (nerve connecting the eye to the brain), macula (small part of the ret-
ina providing the best vision), and standard fields (area that is seen at a given
moment) were also made; the twins’ identities were masked prior to evalu-
ation of these photographs. A 1985 article reported the findings for eigh-
teen MZA and eight same-sex DZA twin pairs.20

• Refractive error. Concordance rates for refractive error were 75

percent for the MZA twin pairs and 50 percent for the DZA twin
pairs. These values, which agree with both past and current studies,
demonstrate genetic effects on visual acuity.
• Motility. Concordance for esotropia (a condition in which the
visual axes are nonparallel such that the eyes appear to be looking
in different directions) was observed in both members of three
MZA twin pairs and in one member of one DZA twin pair. Most
striking to Knobloch was the “biologic clock” demonstrated by the
three concordant MZA pairs: twins in pair A began wearing glasses
at ages seven and eight; twins in pair B developed estropia at an
early age and wore eyeglasses during their preschool years; and
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 89

twins in pair C had had strabismus surgery (to correct the inward
turning of the eyes), one twin at eighteen and the other at nineteen.
• Optic nerve and stereo photographs. The fundus refers to the eye’s
interior surface and includes the retina, optic disc (eye region where
the optic nerve emerges from the eyeball), macula, fovea (center of
the macula and responsible for sharp vision), and posterior pole
(back of the eye). Similarity of the fundus was measured by the cup
to disc ratio (area in the center of the optic disc), based on mea-
surements from the stereo photographs. Concordance was 90
percent for the MZA twins and 50 percent for the DZA twins,
indicating a genetic influence. Marked discordance in one MZA
twin pair was noted.

The most striking eye findings were the genetic effects on refractive
error and estropia. The coordinated appearance and treatment of eye con-
ditions in selected pairs suggested the presence of genetically mediated
timelines underlying these conditions. This phenomenon was well illus-
trated by fifty-eight-year-old MZA male triplets (not included in our 1985
paper) in which one member showed macular degeneration (breakdown
of the visual structure surrounding light-sensitive cells), while his two
brothers showed only early signs. A final insight from the ophthalmology
study (communicated informally by Knobloch at a later date) was that
he could predict twin type almost perfectly from the similarity of the fine
structures on the retina, a bit of information that has never been reported
in a scientific journal.21
Little knowledge of the twins’ birth histories was available. This is unfor-
tunate because administering oxygen to premature infants can impair the
visual apparatus and twins are at increased risk for early birth.22 Thus, it
is possible that concordance among some MZA (and DZA) pairs reflected
this early treatment, reducing the contribution from genetic factors.
Twins continued to visit the eye clinic until 1995, yielding a sample of
approximately 125 pairs. Unfortunately, Knobloch passed away in 2005,
and the full data set has never been analyzed.

Sexual Life History

The MISTRA was the first study to systematically gather comprehen-
sive data on the separated twins’ sexual life histories in general, and on
female twins in particular. The MISTRA’s 1986 article describing sexual
orientation findings for six MZA twin pairs suggested that “male homo-
sexuality may be associated with a complex interaction, in which genes
90 B O R N TO G E T H E R — R E A R E D A PA RT

play some part.”23 That article was not the first to suggest that sexual
orientation has a genetic basis. In 1968, Heston and Shields had pub-
lished a twin and family study of homosexuality that suggested genetic
influence,24 and a handful of other twin studies, published between 1953
and 1971, reached the same conclusion. Heston, acknowledging that his
1986 conference paper on this topic “could not support a sweeping con-
clusion” due to the small number of pairs, decided to “describe the find-
ings and let people draw their own conclusions.” His presentation of these
data at an international congress generated considerable discussion, not
all positive.
The genetic and environmental origins of sexual preference continue to
be debated. Twin studies conducted in the early 1990s, using self-selected
and targeted samples, yielded higher concordance values than those con-
ducted more recently. A summary of studies published between 1952 and
1993 show median concordance rates of 52 percent for MZ twins and
17 percent for DZ twins. In contrast, median concordance values for twin
registry and population-based studies conducted since 2000 yield concor-
dance values of 24 percent for MZ twins and 13 percent for DZ twins.25
A 2000 study, using a large anonymous Australian sample and applying
strict criteria for sexual orientation, found concordance figures of 20 per-
cent for MZ male pairs, 24 percent for MZ female pairs, 0 for DZ male
pairs, and 10.5 percent for DZ female pairs.26 The different recruitment
methods used in the older and newer studies most likely explain the dif-
ferent findings.
Evidence from a growing number of twin and adoption studies (includ-
ing additional separated pairs in the MISTRA) support moderate genetic
influence on male sexual orientation, and little or none in females. The
Australian study reported heritabilities of 0.45 for males and 0.08 for fe-
males. A 2010 population-based study of nearly 4,000 Swedish twin pairs
reported somewhat comparable figures of 0.34–0.39 and 0.18–0.19 for
male and female sexual preference, respectively.27
Homosexuality has also been examined with reference to prenatal ef-
fects on behavior. Hormonal explanations suggest that the level of prena-
tal androgen exposure may influence the development of brain structures
relevant to sexual orientation. Higher birth order in males with older
brothers has been linked to homosexuality, possibly reflecting maternal
immune reactions to testosterone. Pregnancy stress has also been linked
to homosexuality in some but not all studies. Psychosocial theories in
which later-born male children develop feelings of inadequacy or are raised
by overprotective mothers, leading to homosexuality, have been pro-
posed but not substantiated. It is possible that parents may react in spe-
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 91

cific ways to children who display gender nonconformity, although it is less

likely that they create it.28 These sources of influence could be associated
with MZ twin concordance or discordance for sexual orientation.
Previous reared-apart twin researchers did not investigate sexuality,
perhaps because they worried that such a personal topic would make
some twins uncomfortable. Nevertheless, homosexual experiences were
noted in two male reared-apart pairs studied by Juel-Nielsen, and in one
female pair studied by Shields. However, the information is sparse, mak-
ing these pairs difficult to classify in terms of concordance or discordance
for sexual preference.29
The MISTRA’s Sexual Life History assessment took place one evening
after dinner, mostly in psychology department facilities. Co-twins com-
pleted these interviews at the same time, but in different locations and
with different examiners. Spouses and other individuals who accompa-
nied the twins to Minneapolis did not participate in this phase of the
study—we wanted twins to answer freely and honestly, so it was impor-
tant that their companions be unaware of the questions that we posed.
The Sexual Life History battery consisted of a Sexual Meaning Survey,
Sexual Life History Timeline, Sexual Behavior Questionnaire, and Semi-
Structured Interview. The different components were developed by Drs.
Joseph Bohlen and Margaret Sanderson from the University of Minne-
sota. The Sexual Meaning Survey was presented as a semantic differen-
tial in which twins indicated the extent to which words in fifty-one pairs
(e.g., strong-weak; exciting-boring) were meaningful to them in a sexual
sense. Two questions concerning the personal importance of sex and the
importance of sex in their current relationship were also included. The
Sexual History Timeline listed various events (e.g., kissing, intercourse)
for which twins indicated if such events had happened in their lives and,
if so, at what age. The Sexual Behavior Questionnaire asked about the
frequency of sexual activities (e.g., sexual dreams, sexual fantasies). The
Semi-Structured Interview allowed twins to expand upon sexual behav-
iors and themes that were relevant to them.
The estimated percentage of Americans who identify as gay is 1.7 per-
cent.30 Therefore, the number of pairs in which one or both twins were
homosexual was understandably small—the sample included only six
pairs (two male and four female) out of the 55 pairs that had been studied
when the paper appeared in 1986.31 But sexual behaviors in these unique
pairs were of interest.
The data posed some curious dilemmas while also presenting new
ideas. Twins in one male pair were clearly concordant for homosexuality.
They were reunited after one twin was mistaken for his brother when he
92 B O R N TO G E T H E R — R E A R E D A PA RT

visited a gay bar in another town. Both twins had been active homosexu-
als since age thirteen, although one twin had had several heterosexual
encounters. Most reared-together homosexual twins deny sexual inter-
est in one another; however, as indicated in the MISTRA’s 1986 report,
these twins became sexual partners following their reunion.32 Shortly
thereafter, due to our observations of flirtatious behavior between DZA
opposite-sex twins, we began asking all the twins if they had sexual inter-
est in one another. We were especially interested in the answers from the
opposite-sex pairs—as indicated earlier, it was conceivable that in their
case their separate rearing prevented development of the Westermarck
Effect.33
Genetic sexual attraction is the counterpart to the Westermarck Effect,
and may develop in its absence. It refers to the strong sexual connection
experienced by some biological mothers and sons, fathers and daughters,
and brothers and sisters who were separated since birth and reunited as
adults.34 The Westermarck Effect cannot occur in such cases because it
requires early cohabitation, thereby allowing possible sexual feelings to
flourish. Consistent with this view, mate selection studies show that people
are generally attracted to others with whom they are similar in some
physical and behavioral features such as height, intelligence, or values.35
Interestingly, experimental evidence shows that females prefer the odors
of males with whom they share intermediate numbers of paternally trans-
mitted histocompatibility genes (genes controlling the immune system)
compared with males with whom they share few.36 These findings suggest
an optimal balance between inbreeding and outbreeding. The few re-
ported cases of attraction and marriage between reared-apart opposite-
sex twins are comprehensible in light of these effects.37
Sexual feelings between reunited MZA twins shows that genetic sexual
attraction can occur between reared-apart same-sex as well as opposite-
sex siblings. Homosexual MZA twin pairs are rare, yet mutual sexual
attraction was demonstrated in one of our male sets, and sexual attrac-
tion by one co-twin toward his brother was indicated in a second set.
One member of a male pair studied by Juel-Nielsen accused his twin
brother of making sexual advances toward him.38
Twins in our second male set posed the classification conundrum that
I described in Chapter 3: one of the twins considered himself homosex-
ual, and felt sexually attracted to his brother when they met at age thirty-
five. However, his co-twin considered himself heterosexual—at the time
of assessment he was married with children. He engaged in intercourse
infrequently, yet seemed satisfied with his sexual life. But he had had an
affair between the ages of fifteen and eighteen years with an older man.
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 93

The concordance or discordance of this pair was never settled in the

minds of the investigators. The same uncertainty presented itself in the
case of one of our four female pairs, but was eventually resolved.
The female pair in question concerned forty-eight-year-old MZA twins,
one of whom was exclusively heterosexual, while the other had had a
homosexual affair that was “intense and prolonged.”39 However, she re-
garded herself as heterosexual since her marriage at age twenty-nine, but
felt attracted to both males and females nonetheless. She was classified as
bisexual by Eckert and Heston. The other three female pairs were discor-
dant for homosexuality. A review of their life histories revealed nothing
remarkable in their rearing circumstances that could explain the differ-
ence. However, some common threads seemed worth pursuing. The ho-
mosexual member of each of the three discordant female pairs had (1) a
higher body weight and height at the time of the study, (2) a “larger” body
size throughout their lifetime, (3) a later age at menarche, (4) a relative
delay in the development of secondary sexual characteristics such as breast
development, and (5) a relative delay in the onset of sexual experiences.
Genetic influence on male homosexuality was suggested by our early
data, consistent with reared-together twin studies. In contrast, environ-
mental effects on female homosexuality seemed more important. The
twins’ medical histories did not reveal illnesses or other events that were
specific to the homosexual twins.
The study concluded that “the findings are descriptive only, and should
be regarded as clues upon which to base hypotheses.”40 Eckert and Hes-
ton called for further investigation of the female findings. Subsequent re-
search on large reared-together twin samples has supported a larger gene-
tic origin of homosexuality in males, but less than was previously reported.
Genetic effects on homosexuality in females have also been found, albeit
to a lesser degree. Social and environmental explanations of female sexu-
ality are still prevalent.41
Our goal in the sexual orientation study was to demonstrate possibili-
ties. That was one of the beauties of the reared-apart twin study design.

Physical and Psychological Traits: Early Findings

Bouchard’s 1984 chapter in Sidney W. Fox’s edited volume provides an
informative capsule of the MISTRA.42 Referring to that chapter twenty-
five years later, Bouchard recalled, “I was pretty persuaded that we really
had our hands on something significant.” In 1986, he published another
overview of the project that included findings on anthropometric mea-
sures, special mental abilities, and vocational interests.43
94 B O R N TO G E T H E R — R E A R E D A PA RT

The correlations were quite similar to those from previous studies of

twins reared apart and reared together. The data continued to show gene-
tic effects on fingerprint ridge count (0.97 for both MZA males, n = 17,
and MZA females, n = 21) and height (0.92 for MZA males, n = 17; 0.81
for MZA females, n = 23). We also found that the MZA female twins were
less alike in weight (0.49, n = 23) than the MZA male twins (0.91, n = 17),
a difference also found by Shields (0.87 for MZA males, n =15; 0.37 for
MZA females, n = 29).

Cousins are not frequent subjects in behavioral genetic research, perhaps

because they are harder to locate or are less enthusiastic than twins about
participation. It was reasoned that first-cousins’ rearing situations approxi-
mate the maximum possible degree of placement similarity of separated
twins. Cousins’ home environments might be similar to the extent that
their related parents are both intelligent or athletic and provide their chil-
dren with books or basketballs. Cousin correlations would also be af-
fected by their average 12.5 percent genetic overlap.
A great deal of special mental ability data had been gathered in the Ha-
waii Family Study of Cognition, and by administering these tests we had
access to a wealth of comparative data.44 The MZA (0.41–0.66, n = 29)
correlations consistently exceeded the cousin (0.11–0.30, n = 246) correla-
tions, with the largest difference occurring for spatial ability (0.66 and
0.11, respectively). These data demonstrated genetic influence on special
mental skills, while weakening contributions from rearing factors.

Vocational interests had not been a major focus of behavioral geneticists,

one reason why the Strong-Campbell Interest Inventory was included in
the MISTRA test battery. Data from a large 1974 twin study using scales
from the Holland Vocational Preference Inventory were available for
comparison with the MISTRA data.45
The overall pattern suggested by the findings was one of genetic influ-
ence. The correlations ranged from 0.20 to 0.51 for the MZA twins
(n = 39), 0.43 to 0.55 for the MZT twins (n = 979), −0.14 to 0.35 for the
DZA twins (n = 17), and 0.20 to 0.31 for the DZT twins (n = 607). These
results were consistent with an adoption study of the Holland interest
styles that reported average correlations of 0.09 and 0.11 for unrelated
opposite-sex siblings (n = 60) and unrelated same-sex siblings (n = 23), re-
spectively.46 More comprehensive treatments of the MISTRA’s vocational
interest data would eventually appear in 1991, 1994, and 1997 articles.
 S E X U A L O R I E N TAT I O N , C O G N I T I O N , A N D M E D I C A L T R A I T S 95

1984 to 1987
The MISTRA years between 1984 and 1987 were productive in many
ways. A symposium, “The Minnesota Study of Twins Reared Apart,” was
held as part of the fourteenth annual Behavior Genetics Association meet-
ing, in Bloomington, Indiana, May 1984. The presentations included early
findings on IQ, personality, fears, physical attractiveness, and the Min-
nesota Multiphasic Personality Inventory (MMPI).
The year 1986 was the second busiest (after 1981) in terms of studying
new twin pairs—we saw a total of ten new separated twin sets plus one
set of triplets. The six principal investigators and two postdoctoral fellows
presented findings at many conferences across the United States and abroad,
including the Behavior Genetics Association and the International Soci-
ety for Twin Studies.
In 1985, Bouchard became chair of the Department of Psychology, at
which time he appointed me assistant director of the Minnesota Center
for Twin and Adoption Research. I was now more directly involved in
scheduling twins and managing the assessments. Despite his administra-
tive responsibilities, Bouchard stayed involved in every twin assessment.

Figure 4-1. The 1987 reunion of British reared-apart twins, held at the Priest
House in Derby, located in the East Midlands of England. Bouchard is in the
center, standing behind John Stroud, who is seated in the wheelchair. (Photo
courtesy Thomas J. Bouchard Jr. Unpublished newspaper photograph.)
96 B O R N TO G E T H E R — R E A R E D A PA RT

Matt McGue joined the psychology department faculty in 1985, lend-

ing his behavioral genetic expertise to many reports from the MISTRA.
Kimerly Wilcox, the postdoctoral fellow who joined the study in 1982
when I did, remained on campus working on epilepsy studies at the Dight
Institute for Human Genetics. In 1986, we added a periodontal compo-
nent to the assessment schedule, conducted by Drs. Pihlstrom, Michalo-
wicz, Herzberg, and Wolff. Every fall new graduate students began work-
ing on the project, turning their efforts into doctoral dissertations and
publications. In March 1987, our dental collaborator, Dr. JoAnn Boraas,
presented the first paper on the twins’ dental findings at the Annual Meet-
ing of the International Association for Dental Research.47 In September
1987, we began follow-up assessments of our earliest reared-apart pairs.
Several new University of Minnesota faculty members became increas-
ingly involved with the MISTRA during these years. Stephen Rich joined
the university in 1980 as a genetic epidemiologist, working on studies of
epilepsy and type 1 diabetes at the Dight Institute. He eventually devel-
oped working relationships with Bouchard, Lykken, Tellegen, and others.
Over the next several years, Rich would play a key role in analyses of our
personality and heart-rate data. Rich remained at the University of Min-
nesota until 1994 when he joined Wake Forest University’s School of
Medicine in Winston-Salem, North Carolina.48
A 1987 British twins’ reunion, held in Derby in honor of John Stroud,
was the last one Stroud attended before his death in 1989. Looking at a
photo of this event (Figure 4-1), Bouchard remarked, “[It] reminds me of
how many twin pairs we had from England. Quite amazing when you
think about it.”
In 1984, Barbara Unell, a writer and mother of opposite-sex twins,
launched and edited Twins Magazine. Twins Magazine was (and remains)
a monthly publication devoted entirely to twin development and parent-
ing issues, now under the editorship of Christa D. Reed. I held the posi-
tion of research editor from 1984 to 1998. In my first article, “The Na-
ture vs. Nurture Laboratory,” I described our work with separated twins.
I would phrase the title differently today, either as Nature-Nurture or
Nature and Nurture, because it is now widely appreciated that the two
effects work together and are separable only in a statistical sense. Think-
ing about developmental processes has clearly changed.
 CHAPTER FIVE

Pivotal Papers: Personality and IQ (1988 and 1990)

The Personality Paper

The genetic makeup of a child is a stronger influence on
personality than childrearing according to the first study to
examine identical twins reared in different families. The findings
shatter a widespread belief among experts and laymen alike in
the primacy of family influence and are sure to engender fierce
debate.1

Daniel Goleman of the New York Times published the conclusion to our
first major analysis of the personality data before it appeared in a scientific
journal. In fact, he used our conclusion as his opening lines. The Univer-
sity of Minnesota’s public affairs office provided Goleman with a draft of
the study approximately eighteen months before its 1988 publication in
the Journal of Personality and Social Psychology.2 Bouchard had told the
university’s public relations representative that the paper would be pub-
lished, but he did not give it to her. He explained to her that rules prohib-
ited the release of scientific findings in the popular press prior to their
publication in the scientific literature. “But she finagled the paper from
Lykken—he didn’t know her intentions. When the research appeared in
the New York Times she thought she had done a great thing. It really
pissed me off.”
The findings attracted considerable media attention. The paper and the
article were controversial for the way they were released as well as for
their content. A 1990 article in Science Writer by Lisa Lynch addressed
the media attention, albeit in an uninformed way: “Three years ago, by
prepublishing the results of their study, the Minnesota group made it
98 B O R N TO G E T H E R — R E A R E D A PA RT

impossible for other scientists to answer their claims.”3 Neither Lynch

(nor Goleman) really knew how our paper was released. And by the end
of our twenty-year study fifty-six DZA twin pairs had been assessed,
barely half of the nearly one hundred DZA sets Lynch had alleged.
Our finding that shared environments had little effect on most per-
sonality traits engendered new debate. Parents’ contributions to their chil-
dren’s developmental outcomes were defended in the psychological litera-
ture and questioned in the popular press.4 In fact, our paper raised the
possibility that family interaction norms might explain our social closeness
findings, a personality measure that did show significant familial effects.
Evidence of genetic components to personality was still controversial
even though other investigators had produced similar results. The inves-
tigators of the three previous reared-apart twin studies had each included
personality assessments in their programs. Their methods differed from
ours, but all three found that genetic factors explained about 50 percent
of the variation in personality. Moreover, all three described marked
personality differences within some early pairs that they associated with
the twins’ differences in physical factors, rearing practices, and social
influences.
Newman et al. and Shields also had access to MZT and DZT twin
pairs, so they made additional comparisons. Newman’s group suggested
that the nature of personality and the unreliability of the Woodworth-
Matthews Questionnaire his team administered prevented the emergence
of a discernible trend in their data.5 However, their MZA (ri = 0.58) and
MZT (ri = 0.56) correlations were nearly identical and were higher than
the corresponding DZT correlation (ri = 0.37), challenging explanations
of personality based mostly—or wholly—on shared environments.6 The
Downey Will-Temperament Test, despite its “detailed character and the
many opportunities for variation,” yielded five pairs with nearly identical
scores, ten pairs with considerably similar scores, and only four pairs
with widely different scores.7 Newman and colleagues suggested that this
test assesses relatively stable personality dimensions.
Shields found higher extraversion and neuroticism correlations for
his MZA (0.61 and 0.53, respectively) than MZT pairs (0.42 and 0.38,
respectively).8 He explained his findings with reference to the MZT twins’
intrapair dynamics—that is, differentiation due to the twin relationship,
a process that would not affect MZA twins. Still, both the MZA and MZT
twins in Shields’s study were more alike in extraversion and neuroticism
than the DZT twins (−0.17 and 0.11, respectively), which was consistent
with genetic influence. Shields’s own rating system indicated greater per-
sonality differences between the MZA than between the MZT twins,
 PERSONALITY AND IQ 99

although his most similar pairs were found in both groups. In the years
that followed, many reared-together twin studies produced corroborat-
ing evidence of genetic influence on personality traits.9
Except for Shields,10 none of the early investigators had access to DZA
twins, or to the sophisticated biometrical modeling techniques that parti-
tion genetic and environmental effects on behavior that have been devel-
oped since the 1970s. Therefore, the first significant analysis of our data
was unique in that it was the first study to (1) include all four twin groups,
organized by zygosity and rearing status (MZA, DZA, MZT, and DZT)
and (2) apply biometrical modeling techniques to the data. The reared-
together twins were identified through the Minnesota Twin Registry that
Lykken had been developing since 1983.11
Access to the four twin groups allowed us to estimate the relative
contributions of additive genetic effects, nonadditive genetic effects,
shared environmental effects, and nonshared environmental effects on
the variance in personality traits. It was of interest to see if any of the
traits showed greater MZA than MZT resemblance, as did extraversion
and neuroticism in Shields’s study. Our early findings on conservatism
(summarized in Chapter 3) showed greater similarity between MZA than
MZT twins, a difference that was tracked in later analyses.
The data for our first personality study came from Tellegen’s Multidi-
mensional Personality Questionnaire,12 which consists of 300 items pre-
sented in a true-false format. Sample items include (1) “Basically I am a
happy person” and (2) “I suffer from nervousness.” The items yield eleven
primary personality scales and three higher order factors, listed in Table 5-1
(descriptions of each are provided on the Web site). The primary scales
were developed to be relatively independent, to show no overlap among
one another.
The twins’ four (later five) questionnaire booklets were presented in the
order of their importance. The Multidimensional Personality Question-
naire was the first item in Booklet 1, closely followed by a second major
personality questionnaire, the Minnesota Multiphasic Personality Inven-
tory, which I discuss later. The personality assessment forms, variously
distributed across the questionnaire booklets, also included the 16 Per-
sonality Factor Questionnaire, and the California Psychological Inven-
tory, among others.13
Our first personality study included 44 MZA pairs, 27 DZA pairs,
217 MZT pairs, and 114 DZT pairs. The results, as seen in Table 5-1, were
straightforward: the MZ twins were more alike than the DZ twins, regard-
less of rearing status. The median correlations were 0.49 (MZA), 0.52
(MZT), 0.21 (DZA), and 0.23 (DZT), a pattern demonstrating substantial
 100 B O R N TO G E T H E R — R E A R E D A PA RT

Table 5-1. Intraclass Correlations for Scales from the Multidimensional Personality Questionnaire for Twins
Reared Apart and Together

MZA DZA MZT DZT

Scale (n = 44 pairs) (n = 27 pairs) (n = 217 pairs) (n = 114 pairs)

Primary scale
Well-being 0.48 0.18 0.58 0.23
Social potency 0.56 0.27 0.65 0.08
Achievement 0.36 0.07 0.51 0.13
Social closeness 0.29 0.30 0.57 0.24
Stress reaction 0.61 0.27 0.52 0.24
Alienation 0.48 0.18 0.55 0.38
Aggression 0.46 0.06 0.43 0.14
Control 0.50 0.03 0.41 −0.06
Harm avoidance 0.49 0.24 0.55 0.17
Traditionalism 0.53 0.39 0.50 0.47
Absorption 0.61 0.21 0.49 0.41
Higher order factors
Positive emotionality 0.34 −0.07 0.63 0.18
Negative emotionality 0.61 0.29 0.54 0.41
Constraint 0.57 0.04 0.58 0.25

Source: Adapted from Tellegen et al., “Personality Similarity in Twins Reared Apart and
Together,” Journal of Personality and Social Psychology 54 (1988): 1031–1039.

genetic influence. Interestingly, the correlations for traditionalism sug-

gested that common rearing enhances the similarity of that trait for DZT
twins more than for MZT twins who are genetically the same—the MZA
(0.53) and MZT (0.50) correlations hardly differed, but the DZT corre-
lation (0.47) was larger than the DZA correlation (0.39). Testing this
idea would require a closer look at the social-dynamics within the differ-
ent twinships.
Most revealing, the MZA twins were about as similar as the MZT twins
across the eleven different personality scales. These results would challenge
psychologists to rethink the role of parental practices and home environ-
ments in child development. The impact of this finding was profound,
no  doubt the reason why Goleman opened his article with the study’s
conclusion.
Reactions to our key result—that shared family environments have
little effect on individual personality differences—ranged from accep-
tance to dismissal. Psychologist Robert Plomin called it “the single most
important finding in behavioral genetics in the last decade.”14 Psychiatric
geneticist Elliot Gershon, referring to our study in particular and to the
behavioral genetics field in general said, “The more familiar you are with
 PERSONALITY AND IQ 101

the evidence the more persuasive it becomes.”15 In contrast, psychology

professor Leon Kamin claimed, “These are very ambiguous data that can
be interpreted any way you want. I’m not saying anyone is falsifying
facts or anything, just that we know very, very little.”16 Kamin, who raised
the possibility of fraud in Burt’s MZA twin studies, was a severe critic
of the MISTRA, continually demanding the original data to conduct the
analyses himself. Psychology professor Jerome Kagan questioned our her-
itability estimates, arguing that personality inventories may not capture
how people act in everyday life.17
Some behavioral geneticists were also skeptical. Indiana University
psychology professor Richard J. Rose, one of Lykken’s most distinguished
former graduate students, had recently reported that greater social con-
tact between twins he studied in Finland led to their greater similarity in
neuroticism.18 However, as Lykken and colleagues argued in response,
more similar twins probably sought each other out more often than less
similar twins.19 The Lykken-Rose debate continued.20 Of course, argu-
ments favoring causal links between degree of social contact and person-
ality similarity must be reconciled with research from separated twins,
something Rose and colleagues tried to do. They argued that some traits
may be more or less sensitive to contact than others, and that under-
standing the effects of contact will help to explain behavioral variation.
I wonder if the nature of contact experienced by twins growing up to-
gether differs from that experienced by twins who meet occasionally but
have never lived together. For example, twins reared together have shared
life histories and may take one another for granted as companions. In
contrast, twins reared apart may continually provide details of their lives
to one another and may view each meeting as a special occasion. Whether
these distinctions are real or relevant to twins’ personality similarity is
unknown, but MZA and MZT twins’ relations with one another are worth
investigating in this regard. Nevertheless, the MZT twins were not more
alike in personality than the MZA twins, challenging the Rose et al. study’s
conclusions.
The foregoing exchange masks the high esteem with which Rose and
Lykken regarded each other.21 According to Rose:

When David and I engaged in this friendly but public dispute, it was like a
continuation of our student-mentor interactions. Several years after our ex-
change of articles, he confided that he had enjoyed this repartee, for I “had
mounted a good defense,” although my argument and data were not com-
pelling. In retrospect, our argument, like so many in the behavioral sciences,
rested on shaky data. I had asked whether shared experience enhanced sib-
ling similarity, and sought answers in the relative similarities of twin siblings
102 B O R N TO G E T H E R — R E A R E D A PA RT

who differed in the duration of their cohabitation. More incisive evidence of

common environmental influences—it now seems obvious to me—comes
from genetic strangers (adoptive siblings) reared together.22

The Minnesota personality findings were misinterpreted by many. We did

not say that parenting does not matter—parents can help shy children feel
at ease or help rambunctious kids calm down, as Kagan had shown.23 One
implication of our work was that parents should pay close attention to each
child’s unique character traits and nurture each child’s individual interests.
Paradoxically, parental fairness seems more likely to come from treating
children differently in accordance with their individual behaviors than
from treating them alike.24 Of course, this assumes that family environ-
ments are within the normal range (i.e., free of abuse and deprivation).
The bottom line from our data was that growing up together does not
make family members alike. Instead, our findings showed that personality
similarity between relatives seems to come mostly from their shared genes.
Furthermore, environmental effects that are most important in personal-
ity development appear to be those that are experienced apart from the
family—the nonshared environmental factors; their contributions to the
different personality traits were often close to those from genetic factors.
A final point is that the Multidimensional Personality Questionnaire was
not designed to assess personality heritability; rather, it was designed to
assess personality variation. This feature of the instrument increased our
confidence in the genetic effects that emerged.
Our biometrical analyses of the personality data, shown in Table 5-2,
shed additional light on the genetic structure of personality. Estimates of
genetic influence (heritabilities) for the eleven personality scales ranged
from 0.39 to 0.58. The model included additive genetic, shared familial,
and nonshared familial components. The genetic, shared familial, and
nonshared familial components sum to 1.0. The model also included a C
parameter indicating the extent to which the genetic component was ad-
ditive (closer to 0.5) or nonadditive (closer to 0.0).
Nonadditivity was indicated for two personality scales (social potency
and control) and one higher order factor (positive emotionality).25 Thus,
some personality traits seemed to be shaped by many interactive genes,
reflecting the emergenic quality described in Chapter 3. Simple additive
models of personality, endorsed by many personality psychologists at that
time, were challenged by our data.
We had some surprises. One unexpected finding was the lack of a sys-
tematic shared environmental effect for traditionalism (endorsement of
 PERSONALITY AND IQ 103

Table 5-2. Genetic and Environmental Components Estimates for the Multidimensional Personality Questionnaire’s
Primary Scales and Higher Order Factors

Shared Nonshared
Genetic C Environment Environment

Primary scale
Well-being 0.48a (0.08) 0.29 (0.16) 0.13 (0.09) 0.40a (0.04)
Social potency 0.54a (0.07) 0.05a (0.21) 0.10 (0.08) 0.36a (0.04)
Achievement 0.39a (0.10) 0.13 (0.27) 0.11 (0.11) 0.51a (0.05)
Social closeness 0.40a (0.08) 0.19 (0.22) 0.19a (0.09) 0.41a (0.05)
Stress reaction 0.53a (0.04) 0.49 (0.17) 0.00b 0.47a (0.04)
Alienation 0.45a (0.13) 0.50b 0.11 (0.12) 0.44a (0.04)
Aggression 0.44a (0.05) 0.27 (0.19) 0.00b 0.56a (0.05)
Control 0.44a (0.05) 0.00a,b 0.00b 0.56a (0.05)
Harm avoidance 0.55a (0.04) 0.31 (0.15) 0.00b 0.45a (0.04)
Traditionalism 0.45a (0.10) 0.50b 0.12 (0.10) 0.43a (0.04)
Absorption 0.50a (0.10) 0.50b 0.03 (0.10) 0.47a (0.04)
Higher order factor
Positive emotionality 0.40a (0.08) 0.00a,b 0.22a (0.07) 0.38a (0.04)
Negative emotionality 0.55a (0.11) 0.50b 0.02 (0.11) 0.43a (0.04)
Constraint 0.58a (0.04) 0.40 (0.14) 0.00b 0.43a (0.04)

Note: Standard error in parentheses. C: additivity-epistatic component; NSE: nonshared environmental

component; SE: shared environmental component.
a. Differs significantly from the null value at P < .05.
b. Boundary solution, so no standard error computed.
Source: Adapted from Tellegen et al., “Personality Similarity in Twins Reared Apart and Together,”
Journal of Personality and Social Psychology 54 (1988): 1031–1039.

conventional family and moral values). However, in 1986, Australian

twin researchers found genetic influence but little common environmen-
tal influence on the transmission of social attitudes such as right-wing
authoritarianism. The Australian paper appeared the same year that we
submitted our personality paper for publication. Likewise, a 1970s study
showed little adoptive parent–child resemblance in authoritarianism,
relative to biological parent–child pairs. Thus, our findings were consis-
tent with those of others.
We also noted that Social Closeness (sociability; warmth and affection
toward others) was the only primary scale to show significant within-
family effects. We speculated that because people within families gener-
ally respond to affiliative or nonaffiliative behaviors in kind, a shared
family effect on social closeness was not unexpected.26 Thus, the data
showed that some personality traits differed in heritability from others,
although a number of traits showed similar heritabilities. Bouchard would
104 B O R N TO G E T H E R — R E A R E D A PA RT

revisit the Multidimensional Personality Questionnaire data in 2003

with an expanded reared-apart twin sample.

The 1989 ABC Nightline program on our personality findings was

memorable. I regard it as one of the best media portrayals of our study
because it let our participants speak for themselves. The show featured
the then twenty-eight-year-old MZA triplets Robert (Bob) Shafran, Da-
vid (Dave) Kellman, and Edward (Eddy) Galland, reunited when they
were nineteen years old. Bob plainly stated one of the key findings in our
paper: “We’re very much alike—yet our parents are very different . . .
we’re alike in ways that had nothing to do with who our parents were.”27
Bobby’s views personify active genotype-environment correlation—the
idea that people seek experiences consistent with their genetically based
tendencies. Bouchard later captured this concept in his wonderful phrase
“nature via nurture”—the idea that genetic factors are expressed “by in-
fluencing the character, selection, and impact of experiences during de-
velopment.”28 Bouchard agreed with our critics that MZA twins probably
do have more similar environments than DZA twins, but they are envi-
ronments of the twins’ making. Bouchard added that if people actively
shape their environments, then individually tailored interventions for even
highly heritable traits are possible.

The Science IQ Paper

The Science IQ paper was a significant moment in the history of the study.
The MZA twin sample was large enough to present the long-awaited IQ
data as well as findings from other physical and behavioral domains.
A master table (Lykken’s idea) compared the MZA data with available
MZT data across twenty-two physical and behavioral traits, strengthen-
ing our findings and conclusions. Many people read this report, and we
discovered new twin pairs as a result of the publication’s reach. As
Bouchard said, “The paper legitimated the study.”
The Science paper was our first publication devoted almost exclusively
to IQ, but it was the second submission on this topic. In the early 1980s,
Bouchard had submitted an IQ report to Science based on twenty-nine
MZA twin pairs and twelve DZA twin pairs, much fewer than the forty-
eight MZA twin pairs in our 1990 paper. The decision to publish was
based on pressure from one of our funding sources, which wanted to see
how the scientific community responded to the findings. The article was
rejected largely due to the small sample size, so Bouchard delayed pub-
lishing the IQ data until additional pairs were assessed. Note that the
 PERSONALITY AND IQ 105

studies by Newman et al. and Juel-Nielsen included nineteen and twelve

MZA twin pairs, respectively, and as early as 1981 the MISTRA had
studied thirty.29
The genesis of the 1990 IQ article can be traced to a 1982 conference
on the biological bases of individuality.30 Bouchard recalled that his
talk on the MISTRA was attacked “up one side and down the other” by
a participant who believed that nurture was the primary influence on
behavior. The following day, Bouchard was pulled aside by a biochem-
ist from New York who asked him for a detailed description of the
reared-apart twin design. It was Daniel E. Koshland Jr., the man who
would become Science’s editor in chief in 1985. In a roundtable discussion
after Bouchard’s presentation, Koshland called the MISTRA “a terrific
study.”31
Several years later, Koshland called Bouchard to request a paper on the
MISTRA for a special genome issue of the journal—and he needed it in
six weeks. Bouchard, who had just become the department chair, wasn’t
sure he could deliver it in that amount of time. Koshland asked, “Do you
know who you’re talking to?” To which Bouchard replied, “Yes, the edi-
tor of Science, and we’ll have a paper to you in six weeks.”
We had studied more than one hundred reared-apart twin and triplet
sets by the time our Science paper was in progress. The paper reported IQ
analyses for forty-eight MZA twin pairs for whom we had processed data;
the DZA twin sample was still modest (thirty pairs), and several recently
assessed pairs were omitted. Given the significance of this paper in the
history of the study, in Table 5-3 I have displayed the descriptive charac-
teristics for the fifty-six MZA twin pairs we had assessed at that point
in time. Both parents’ educational levels (mean: two years of high school)
were below those of volunteer adoptive parents, and their educational at-
tainment and socioeconomic status were quite variable. The twins’ aver-
age IQ score exceeded the population mean of 100, because the sample
lacked twins in the lowest ranges (below 70).
Three general intelligence measures were available for the MISTRA
twins: the Wechsler Adult Intelligence Scale (WAIS),32 the Raven/Mill-Hill
Composite,33 and the first principal component (a variable subsuming a
larger number of correlated variables) based on the special mental ability
tests in the Hawaii and Comprehensive Ability Batteries. The WAIS is an
individually administered test consisting of six verbal subtests (Informa-
tion, Digit Span, Vocabulary, Arithmetic, Comprehension, and Similari-
ties) and five nonverbal subtests (Picture Completion, Picture Arrange-
ment, Block Design, Symbol, and Object Assembly). It was the first activity
scheduled for Tuesday morning, so twins were well rested during its
Table 5-3. Age, Contact Measures, IQ Score, and Parental Education for 56 MZA Twin Pairs

Time Together prior Time Apart to

to Separation First Reunion Total Contact Father’s Years Mother’s Years
Measure Age (Years) (months) (years) Time (weeks) IQ Score (WAIS) of Education of Education

Mean 41.0 5.1 30.0 112.5 108.1 10.7 10.3

SD 12.0 8.5 14.3 230.7 10.8 4.5 3.7
Range 19.0–68.0 0–48.7 .5–64.7 1–1,233 79–133 0–20 0–19

Note: SD: standard deviation; WAIS: Wechsler Adult Intelligence Scale.

Source: Adapted from Bouchard et al., “Sources of Human Psychological Differences: The Minnesota Study of Twins Reared Apart,” Science 250
(1990): 223–228.
 PERSONALITY AND IQ 107

administration. Each twin was tested at exactly the same time by a differ-
ent professional psychometrist, who had no affiliation with the study and
who scored the protocol immediately upon completion.
The Raven is a nonverbal measure of problem-solving skill, and the
Mill-Hill is a multiple-choice test of word knowledge. These two tests are
often given together and were administered to the twins by computer as
part of Lykken’s laboratory assessment. There was no time limit for com-
pleting these tests. According to Lykken, allowing unlimited time shows
how quickly an individual thinks and how long he or she persists at a
task before giving up, factors that may work multiplicatively.34 The spe-
cial mental ability tests were administered to twins in several sessions
during the week. The rationale for testing them together is discussed in
Chapter 2.
The intraclass correlations for the general intelligence measures com-
pleted by the MISTRA twins and twins in the three previous reared-apart
twin studies are shown in Table 5-4. The consistency of the findings
across studies in the primary, secondary, and tertiary tests was striking,
suggesting that 64 to 74 percent (separate tests) and 69 to 75 percent
(mean of multiple tests) of the IQ variance is associated with genetic fac-
tors. Thus, replication both between and within studies was demonstrated.
This range was higher than the 47 to 58 percent range indicated by other
kinship studies. However, most twin and sibling studies used young sam-
ples, whereas our study and the previous reared-apart twin studies used

Table 5-4. IQ Intraclass Correlations and Standard Deviations for the Four Reared-Apart Twin Studies

Study and Test (Primary/

Secondary/Tertiary) n (pairs) Primary Test Secondary Test Tertiary Test Mean

Newman et al.
SB / Otis 19/19 0.68 ± 0.12 0.74 ± 0.10 — 0.71
Juel-Nielsen
WB / Raven 12/12 0.64 ± 0.17 0.73 ± 0.13 — 0.69
Shields
MH / Dominoes 38/37 0.74 ± 0.07 0.76 ± 0.07 — 0.75
Bouchard et al.
WAIS / RMH / FPC 48/42/43 0.69 ± 0.07 0.78 ± 0.07 0.78 ± 0.07 0.75

Note: The original data from Newman et al. (SB: 0.670, Otis: 0.727) and Juel-Nielsen (WB: 0.62)
were presented as correlations, not as intraclass correlations. Bouchard recalculated the data to
compute intraclass correlations. Juel-Nielsen did not present the correlation for the Raven. FPC:
First Principal Component; MH: Mill-Hill; RMH: Raven/Mill-Hill; SB: Stanford-Binet; WB:
Wechsler-Bellevue.
Source: Adapted from Bouchard et al., “Sources of Human Psychological Differences: The
Minnesota Study of Twins Reared Apart,” Science 250 (1990): 223–228.
108 B O R N TO G E T H E R — R E A R E D A PA RT

mostly adults. This was an important difference because IQ heritability

has been shown to increase with age.35
We examined the IQ data with reference to (1) associations between
the twins’ environmental similarities and their IQ scores, (2) contribu-
tions of prereunion and postreunion contact to IQ resemblance, and
(3) similarity of the MZA and MZT twins across other behavioral and
physical measures. These efforts were undertaken partly in anticipation
of critics’ comments, although, as I will explain later, we did not predict
one of the most common criticisms—our alleged incorrect reporting of
how much IQ variance was due to genetic factors.
Answers to questions about the effects of correlated rearing environ-
ments on IQ came partly from the twins’ responses to a “Physical Facili-
ties in the Home” checklist. This inventory asked twins to indicate the
presence of musical instruments, cars, foreign language books, and other
items in their childhood homes. They were also asked to list household
items that they recalled but were missing from our list. We obtained
information about their parents’ educations and occupations (socio-
economic indicators) as part of the Life History Interview. Each twin also
completed the Moos Family Environment Scale, a form requesting infor-
mation about their parents’ treatment and rearing of them in childhood
and adolescence. We calculated placement coefficients (indicating the
degree of co-twin similarity for socioeconomic status indicators, home
facilities, and rearing factors), correlations between IQ scores and these
environmental measures, and estimates of the degree to which similarity
in these measures contributed to the twins’ WAIS IQ similarity. The place-
ment coefficients and other results from these analyses are displayed in
Table 5-5.
The maximum contribution to the MZA trait correlations from a single
variable was 0.03 (Material Possessions). We did not detect significant
effects on the twins’ IQ scores or on their IQ resemblance from the envi-
ronmental measures. Family influences have been shown to affect sibling
similarity in cognition36 and rate of skill acquisition in young children,37
but these effects diminish in adolescence and adulthood.
Our findings did not suggest that environmental variables did not affect
the twins’ IQ scores in individual cases. Some reared-apart twins with
more favorable educational backgrounds outperformed their less advan-
taged co-twins, but the effects were not systematic. Recall that the New-
man et al. study found substantial correlations between educational ad-
vantage and IQ score (0.79 and 0.55), but also found intraclass correlations
of 0.68 and 0.74 for their two IQ measures.38 Despite some twins’ more
favorable educational upbringing, MZA co-twins reared apart were gen-
 PERSONALITY AND IQ 109

Table 5-5. Placement Coefficients for the Environmental Variables, Correlations between IQ and the
Placement Variables, and Contributions of Placement to the Twins’ IQ Resemblance

Correlation: Contribution: Placement

MZA Similarity IQ-Placement to MZA Correlation
Placement Variable (Rff) Variable (rft) (Rff × r 2ft)

SES indicators
Father’s education 0.134 0.100 0.001
Mother’s education 0.412 −0.001 0.000
Father’s SES 0.267 0.174 0.008
Physical facilities
Material possessions 0.402 0.279a 0.032
Scientific/technical 0.151 −0.090 0.001
Cultural −0.085 −0.279a −0.007
Mechanical 0.303 0.077 0.002
Relevant FES scales
Achievement 0.110 −0.103 0.001
Intellectual orientation 0.270 0.106 0.003

Note: FES: family environment scale; SES: socioeconomic status.

a. Indicates correlation that is statistically significantly different from zero at P < .01.
Source: Adapted from Bouchard et al., “Sources of Human Psychological Differences:
The Minnesota Study of Twins Reared Apart,” Science 250 (1990): 223–228.

erally more similar to one another than they were to members of other
twin pairs.
We found that the degree of social contact between co-twins was
not  associated with their IQ similarity. We also examined correlations
between the absolute within-pair IQ difference and four contact mea-
sures to see if the twins who had spent more time together were more
alike in IQ. However, these correlations were all small and not statisti-
cally significant.39 Again, these findings do not mean that environmental
variables do not affect some twins’ IQ scores. Instead, they tell us that
environmental variables work in ways that differ from how most psy-
chologists believed they did. Specifically, shared genes, rather than social
contact between family members, explained the twins’ resemblance in
measured intelligence.
Table 4 in the Science paper reported the MZA and MZT correlations
and reliabilities (consistency or repeatability of a measure) for nine classes
of variables. It is an informative compilation of data in that the MZA
and MZT twins showed remarkable similarity across measures. This was
evident for physical characteristics (e.g., systolic blood pressure, electro-
encephalogram [EEG] activity, and heart rate) and behavioral traits (e.g.,
personality, religiosity, and social attitudes). Again, these data show that
110 B O R N TO G E T H E R — R E A R E D A PA RT

shared environments have minimal effects on many traits once individu-

als reach young adulthood. An abbreviated version of that table is shown
as Table 5-6.
We drew three key conclusions from our IQ analysis:
1. Approximately 70 percent of the population variance in intelli-
gence in middle-class, industrialized societies could be linked to
genetic factors. This finding would not apply to impoverished or
underprivileged individuals for whom severe conditions may not
support normal development.40 None of our reared-apart twins
came from such adverse backgrounds. This finding also did not
imply that intelligence could not be enhanced by educational or
remedial interventions.
2. Modern Western societies’ institutions and practices did not greatly
constrain the development of individual variations in psychological
traits. The heritability of traits increases when individuals have
access to similar, trait-relevant opportunities (e.g., baseball games

Table 5-6. Intraclass Correlations (ri) for MZA and MZT Twins across Selected Measures

MZA MZT

Variables ri n (pairs) ri n (pairs) Reliability

EEG
8–12- HZ alpha activity 0.80 35 0.81 42 NA
Midfrequency alpha 0.80 35 0.82 42 NA
activity
Physiologic
Systolic blood pressure 0.64 56 0.70 34 0.70
IQ scores
WAIS-Full scale 0.69 48 0.88 40 0.90
WAIS-Verbal 0.64 48 0.88 40 0.84
WAIS-Performance 0.71 48 0.79 40 0.86
Information processing
Speed of response 0.56 40 0.73 50 NA
Psychological interests
Career and leisure 0.39 52 0.48 116 0.82
(mean of 23)
Occupational (mean 0.40 40 0.49 376 0.75
of 17)

Note: EEG: electroencephalogram; NA: not applicable; WAIS: Wechsler Adult

Intelligence Scale.
Source: Adapted from Bouchard et al., “Sources of Human Psychological Differences:
The Minnesota Study of Twins Reared Apart,” Science 250 (1990): 223–228.
 PERSONALITY AND IQ 111

are available for most boys, so baseball skills should be heritable).

Similarly, heritability decreases when environments are variable
(e.g., language characteristics are closely tied to national regions,
so accents and idioms should reflect environmental effects).
Members of Western societies display considerable variability in
intelligence, personality, interests, and attitudes, all of which show
relatively high heritabilities. In other words, individual differences
in the behaviors we measured were allowed freedom of expression
within the participants’ environments. If we had measured the
male twins’ baseball skills, it is likely we would have found genetic
effects. In fact, twin studies of athletic skill have found genetic
influence on running speed, static strength, and oxygen uptake.41
Had we compared the accents and idioms of our twins from
Georgia and Michigan, or Kentucky and Connecticut, we would
have probably found environmental effects.
3. The MZA twins’ striking behavioral resemblance was largely due
to their shared genes, which increased the probability that their
environments were alike. The immediate causes of most psycho-
logical variations are probably environmental in nature. However,
the environments of individuals are significantly fashioned by
their genotypes that selectively guide them toward certain people,
places, and experiences and away from others (i.e., active
genotype-environment correlation). Children with sports interests
will choose these activities at school and on weekends. According
to Martin and colleagues, learning experiences “augment rather
than eradicate the effects of the genotype on behavior.”42 We
believe that though the twins grew up in different homes they had
choices, and what they chose largely reflected their genetically
influenced abilities and interests. That is, it may be that the twins’
environments were similar in the sense that each took an active
role in creating them and that this tendency continued to adult-
hood. The Jim twins had workbenches in their basements, and the
fireman twins had social lives centered around their fire stations.
We concluded that environmental interventions in the areas of
intelligence, personality, and temperament would be more effective
if tailored to the specific interests and needs of individuals.

We ended the Science paper with a consideration of the relevance of

our findings to the then new discipline of evolutionary psychology. Evo-
lutionary psychology is concerned with identifying behavioral mecha-
nisms that evolved to meet the environmental challenges and demands
112 B O R N TO G E T H E R — R E A R E D A PA RT

faced by our ancestors. It attempts to understand how and why the hu-
man mind is designed the way it is, what the mind is designed to do, and
how environmental events interact with the mind to eventuate in behav-
ioral phenotypes. Two classes of explanation—ultimate and proximal—
are recognized. Ultimate explanations emphasize behavioral and cognitive
functions with reference to survival and reproduction. Proximal explana-
tions consider the immediate events giving rise to different behaviors.43
Evolutionary psychology focuses on species-typical behaviors, whereas
behavioral genetics focuses on individual differences. But both are con-
cerned with identifying the origins and functions of behavior. We sug-
gested that whatever the answer, behavioral variation is characteristic of
modern society and cannot be ignored.
Results from the Science paper appeared in hundreds of newspapers,
magazines, and broadcasts across the country and around the world.
Psychology professor Robert Plomin applauded the study, emphasizing
the rarity of explaining even 10 percent of the variation in complex traits.
Psychology professor David Rowe, while not denying the importance of
parenting, reiterated our view that parents may be less responsible for
their children’s behaviors than they thought.
In contrast, Dr. Norman Krasnegor of the National Institute of Child
Health and Human Development said that if the 70 percent IQ heritabil-
ity estimate was true “it shouldn’t matter what you do or where you go
to school.” However, heritability does not deny the importance of learn-
ing and experience. Bouchard said it best in a passage repeated in several
news sources: “We think of each pair of identical twins as one piece of
music played by two different musicians. The music can be played fantas-
tically or it may not run right. But you’ll always be able to recognize the
piece. That’s because nature writes the score. Environment is responsible
for the playing technique.”44
More interesting than the media reports were the letters we received
from Science readers. Harvard University biologist Edward O. Wilson
wrote to Bouchard directly:

The extraordinary effort and care you and your associates put into this
landmark study should remind all of the fundamental interest and impor-
tance, and reasonableness, of genetic variation in human behavior, while si-
lencing all but the diehard critics, who would go down fighting even if you
laid out the full nucleotide sequences with mathematically perfect forms of
reaction. (October 31, 1990)

There were more critical comments than congratulatory notes, but this
did not mean that most readers rejected our findings. The three individuals
 PERSONALITY AND IQ 113

who reviewed the manuscript upon submission were enthusiastic, but

they all requested modifications—even though the paper had been invited
by Science’s editor, it was not accepted unconditionally. One asserted that
“Twins reared in different homes provide the clearest and most easily
comprehended estimates of the degree to which genetic resemblance di-
rectly affects phenotypic resemblance.” Another wrote that “[The paper]
has great scientific merit, and it is also of wide general interest,” and asked
for clarification of the environmental interactions likely to affect IQ. The
third reviewer was the most revealing with respect to his or her initial
reservations and final opinion:
I have read this manuscript as a hostile reader. I have looked to find the
errors in your ways. In fact, I cannot lay my finger on any specific item that
would invalidate the paper . . . It is quite astounding to me that [general
intelligence] data collected at different times [by three reared-apart twin
studies], with different techniques, with different instruments, with different
populations, all result in very similar numerical values. The results are too
good to believe. Perhaps the evil influence of Burt [whose data proved ques-
tionable] has come into the picture . . . I would hope you would show [your
statistical colleagues] these results and see if they feel that they are a bit too
good to believe. Such an analysis might make many of your readers happier.

The expected criticisms of our Science report were familiar, having

been raised by Taylor in 1980. There were seventeen such comments in
my files, but those by MIT mathematician Richard M. Dudley and Har-
vard University microbiologist Jonathan Beckwith were emblematic of
readers’ concerns. They focused on the effects of the co-twins’ rearing
and educational similarities on their IQ scores, the pressure on the twins
to appear alike, and the inability of readers to examine the original
material.45
We responded to these challenges in a letter to Science. Within-pair
correlations between co-twins’ IQ differences and educational differences
do not weaken genetic explanations of individual differences in intelli-
gence. “It is the correlation between a measured environmental feature
and the variance in IQ not accounted for by genetic factors.”46 Some
higher-IQ twins in our study and in the others had an educational advan-
tage, but the direction of causation was unclear. It was also unlikely that
“social pressure to appear to be similar” could explain the twins’ simi-
larities in our measures of mental abilities, psychophysiology, personal-
ity, and vocational interests because the twins had not seen these proto-
cols until they arrived in Minnesota.47 Lastly, the University of Minnesota’s
institutional review board (like all university review panels) requires con-
fidentiality of participant data. This was stated in our protocol, and the
114 B O R N TO G E T H E R — R E A R E D A PA RT

twins signed a consent form acknowledging their understanding of this

agreement.
The surprise was the eight letters voicing concern that our percentage
of IQ variation associated with genetic factors was 0.49, not 0.70. In
other words, these correspondences indicated that we should have squared
0.70 to arrive at 0.49. However, while ordinary (Pearson) correlations are
squared to derive the proportion of variance in one measure explained by
the variance in the other measure, intraclass correlations already express
the proportion of variance shared by the twins.48 Other explanations of the
differences between the two types of correlations are available.49
As indicated earlier, many people have wondered whether MZ twins’
shared prenatal environments contribute to their intellectual resemblance.
Information on the reared-apart twins’ prenatal and perinatal histories
was largely unavailable. Twins, especially MZ twins, are more highly sub-
ject to adverse birth events (e.g., fetal transfusion syndrome or prematu-
rity) than DZ twins and nontwins.50 These events may result in IQ differ-
ences between co-twins when they are young, although their effects may
lessen over time.51 The late developmental psychologist Ronald Wilson
suggested that developing organisms have “buffering mechanisms” that
may protect them from early insults and that some individuals can with-
stand early trauma better than others.52
Our Science IQ paper findings have held up, having been confirmed
by reared-apart twin studies conducted elsewhere. In 1992, the Swedish
Adoption and Twin Study of Aging reported a 0.78 IQ intraclass correla-
tion for forty-five MZA twin pairs.53 These data, together with the find-
ings from the three early studies and the MISTRA, yielded a weighted
average correlation of 0.73, based on 162 MZA twin pairs.54 This finding
is extremely robust. The controversy over whether or not intelligence is
genetically mediated seems to have dissipated somewhat as scientific in-
terest and public attention have focused on factors affecting other pheno-
types, such as sexual orientation and obesity. But that does not prevent
our findings from being misused—the Science IQ article is posted on the
Web site of former Louisiana Republican State Representative and Ku
Klux Klan member David Duke under the tags, “Intelligence: Heredity Vs.
Environment” and “Racial Differences.”55 Race differences in intelligence
were never addressed in our article.

During the years of the personality and IQ papers and beyond, many
articles were published on many topics. In late 1988, Lykken’s Swiss
colleague Hans H. Stassen reviewed the EEG findings from a somewhat
larger sample of our reared-apart twins, twenty-eight MZA and twenty-
 PERSONALITY AND IQ 115

one DZA.56 Evidence of genetic effects on these measures continued—the

MZA co-twins’ EEGs were as similar to each other as the EEGs of a single
individual measured over time. Other papers reported the first reared-
apart twin analyses of job satisfaction and responses skin conductance.
None of these publications drew the same attention as did the Science ar-
ticle, but some findings strongly challenged conventional theories within
certain disciplines, industrial relations being one of them.
 CHAPTER SIX

Job Satisfaction, Cardiac Characteristics,

and More (1989–1990)

T he S cience IQ paper was a key accomplishment during the

years 1988–1990, but not the only one. Our study of job satis-
faction was the first attempt by anyone to examine genetic influ-
ence on the workplace experience. During these years, we also reported
analyses of the electrical conductance of the skin, heart rate, special men-
tal abilities, information processing, psychopathology, religiosity, intelli-
gence, and personality. The choice of behaviors to be analyzed, and when,
were affected by the amount of accumulated data, significant twin find-
ings reported by other laboratories, and personal time available to the in-
vestigators. Most of our analyses concerned intelligence and personality
because there were comparable data from the earlier reared-apart twin
studies and because these topics interested psychologists in general.

Job Satisfaction
Our collaborators on the job satisfaction paper were Dr. Richard D. Arvey
from the University of Minnesota’s Industrial Relations Center and his
graduate student Lauren M. Abraham.1 Both of them viewed studying
reared-apart twins as a novel and exciting approach to understanding
factors affecting job satisfaction.
As a Minnesota graduate student in the late 1960s, Arvey became ac-
quainted with Bouchard over discussion of a possible dissertation topic.
He returned to the University of Minnesota as a faculty member in in-
dustrial relations in 1983. The source of Arvey’s inkling that genes might
affect job-related attitudes is uncertain, but he had followed the progress
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 117

of the MISTRA since it began. “I contacted Bouchard to ask if he was

gathering job satisfaction measures on the twins and he said ‘no—why?’
We talked about the possibility that genes contributed to job satisfaction,
then he said ‘let’s do it!’ That was Tom. That was the start of our collabo-
ration, and I am grateful to him for it.” By 2010, an Economist essay
describing twin studies in management science showed how far the field
had progressed in the past twenty-five years.2
In the mid-1980s, individual differences in job satisfaction were well
known to researchers in industrial relations. In 1986, investigators found
that affective dispositions (the emotional state of the person, such as their
initiation of humor or need for reassurance) that influenced job satisfac-
tion were constant over time. But the investigators could not disentangle
the genetic and environmental sources of these dispositions, possibly be-
cause their archival data on individuals from three longitudinal growth
studies lacked genetically informative kinships, such as twins and adop-
tees.3 It seemed to us that heredity could affect the ways in which people
react to their work contexts, given that genetic factors had been shown
to affect a wide variety of other behaviors.
The reared-apart twins completed a work history questionnaire as part
of the standard life history assessment. The work history component in-
cluded the short form of the Minnesota Job Satisfaction Questionnaire
developed in 1967 by David Weiss and colleagues.4 The questionnaire
manual specifies an “intrinsic” satisfaction scale that measures workers’
personal values and interests affecting job satisfaction. The manual also
specifies an “extrinsic” scale that measures external working situations
and standards that may affect job satisfaction. A general satisfaction scale
is created by summing the scores from the intrinsic and extrinsic scales.
We also administered a one-item question measuring overall job satisfac-
tion and classified the twins’ occupations according to complexity, motor
skills, physical demands, and undesirable working conditions. Data were
available for thirty-four MZA twin pairs—twenty-five female and nine
male—who completed the questionnaire between 1983 and 1987.
The reared-apart twins showed significantly higher job satisfaction lev-
els than the normative group on all three scales (intrinsic, extrinsic, and
overall), but the differences were slight. Intraclass correlations for the
twenty individual items ranged from 0.043 (working conditions) to 0.371
(be “somebody” in the community), and five were statistically significant.
Intraclass correlations for the three satisfaction scales, the overall satisfac-
tion item, and the four job-rating scales are displayed in Table 6-1.
The intraclass correlation for the general satisfaction scale was sta-
tistically significant (ri = 0.309). Upon removing the thirteen twin pairs
118 B O R N TO G E T H E R — R E A R E D A PA RT

Table 6-1. MZA Intraclass Correlations for Job Satisfaction Measures

Variable ri

Four scales
Intrinsic satisfaction 0.315a
Extrinsic satisfaction 0.109
General satisfaction 0.309a
Overall satisfaction 0.166
Job rating scales
Complexity 0.443a
Motor skills 0.356a
Physical demands 0.338a
Unusual working conditions —

a. P < .05. Intraclass correlations less than zero were not reported.
Source: Adapted from Arvey et al., “Job Satisfaction: Environmental and Genetic
Components,” Journal of Applied Psychology 74 (1989): 187–192.

that included housewives, the correlation remained virtually unchanged

(ri = 0.304). Most interesting, the intrinsic satisfaction scale correlation
was statistically significant (ri - 0.315), but the extrinsic satisfaction scale
correlation was not (ri - 0.109). We had expected this result because in-
trinsic aspects of job satisfaction are probably tied more directly to indi-
viduals’ work experiences, such as challenge and achievement. In addi-
tion, favorable or unfavorable working conditions, like long lunches or
insufficient pay, are likely to make everyone happy or upset, whereas not
everyone gains personal gratification from being creative or productive.
Intraclass correlations for three of the four job-rating scales (complex-
ity, motor skills, and physical demands) showed that the MZA twins held
jobs that were similar on these attributes, a finding exemplifying active
genotype-environment correlation—people generally seek jobs that satisfy
their interests. Little change in the job-satisfaction scales was seen upon
controlling for the job scores, confirming that personal values are key to
satisfaction on the job.
Job interests often coincide but may be expressed differently due to
circumstances. An MZA female twin reared in the United States obtained
an advanced science degree, while her twin sister reared in China received
little education. Both twins were extremely talented in mathematics, and
the China-raised twin used her quantitative skills as a cashier. When MZA
male twins Jake and Keith met at age twenty-three, Jake was a pump me-
chanic, and Keith was a welder, jobs requiring manual dexterity and tech-
nical skill. At age fifty-five, both twins had become computer network
information technology (IT) administrators.
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 119

Even though we found significant genetic effects (approximately 30

percent) underlying intrinsic and general job satisfaction, the remaining
70 percent might be linked to factors such as drive time to work or child
care responsibilities. And, contrary to our expectation, the MZA intra-
class correlation for intrinsic satisfaction, while indicating a genetic effect,
was not significantly higher than the corresponding correlation for ex-
trinsic satisfaction; a larger sample might have yielded a statistically sig-
nificant difference between them. Lastly, our small sample was composed
largely of women, limiting the generalizability of the findings; perhaps fe-
males are less vocal about unfavorable aspects of their professions. In
later years, we would revisit the origins of job satisfaction with larger
samples and learn that our initial findings did not change.
Industrial organizational psychologists Russell Cropanzano and Keith
James challenged our findings.5 They suggested that the twin’s similarity
in job satisfaction may have been inflated by their similar socioeconomic
circumstances and that more sophisticated statistical methods should have
been applied. However, we found little co-twin resemblance in the twins’
relevant child-rearing accounts, so it seemed that they were raised in dis-
similar environments. Furthermore, there were no available studies show-
ing that early family environmental factors were substantially linked to
later job satisfaction—although this did not mean that there were none.
We agreed that job satisfaction studies would benefit from larger samples
and quantitative modeling, but the methodological approach we took was
straightforward and appropriate to the study as conducted.6
Arvey continued to collaborate with us on work-related papers until
he left Minnesota in 2006. Until 1992, Lauren Abraham (later Lauren
Keller) did as well, as a faculty member in the Department of Manage-
ment and Finance at St. Cloud University in Minnesota. Arvey recalled
being surprised by the 0.30 heritability of job satisfaction—he expected
it to be higher, given that genetic effects on behavior are pervasive. He
also said that our 1989 paper was considered “revolutionary” by several
of his colleagues because it challenged the predominantly environmental
views held by industrial relations researchers at that time.

Skin Conductance
Lykken’s laboratory in Diehl Hall, located in the mass of hospital build-
ings on the south side of the campus, was fascinating to me. His quarters
were remotely located at the end of a series of narrow hallways that were
hard to navigate at first. Staff members, myself included, delivered and
picked up one twin on Monday and the other twin on Tuesday afternoons.
120 B O R N TO G E T H E R — R E A R E D A PA RT

(Twins arrived together on Monday afternoons for photographs, hand-

prints, and fingerprints, after which one twin was escorted elsewhere for
medical and psychiatric interviews.) While the twins were at these ap-
pointments, the research staff and I reviewed their inventory booklets
and updated their progress on the participant checklist taped to the wall
of our Elliott Hall headquarters.
Lykken’s psychophysiological laboratory was packed with computers,
wires, and a booth with a soundproof door for silence while twins were
tested. His publications, such as a 1998 study of the heritability of skin
conductance, revealed a great deal about the science that went on there.
There are many meaningless stimuli in our environments such as the
sounds of washing machines cycling and the whistles of trains passing. If
we could not habituate (stop responding) to these noises, it would be nearly
impossible to focus on important tasks.
Skin conductance response refers to changes in specific electrical prop-
erties of the skin. These changes, associated with sweat gland activity, are
elicited by stimuli causing arousal, orienting (immediate response to a
change in the environment), and emotions such as fear and anger. Re-
search in this area may offer insight into why some people are more dis-
tracted by leaking faucets than others, and why some people are more
task-oriented than others.
Skin conductance response decreases in a negatively accelerating way
when individuals are presented with what Lykken called “a punctate
stimulus,” a brief tone that is presented repeatedly. However, he pointed
out that habituation (elimination of a response due to continuous expo-
sure to the stimulus that triggered the response) may be confounded by
how research subjects individually interpret the significance of such stim-
uli. For example, some individuals might start counting repeated sounds,
assuming that acquiring this information was part of the experiment.
Lykken and his colleague, William Iacono, resolved this problem in 1979
by instructing the participants to ignore any tones or other sounds they
might hear, while encouraging them to focus on an engaging task they
provided.7 They repeated this approach with our twins.
Ours was not the first laboratory to consider a genetic influence on
skin conductance response. A 1966 study of eleven MZT twin pairs and
eleven DZT twin pairs found correlations of 0.75 and 0.13, respectively,
for a measure of habituation.8 However, not all early twin studies in this
area produced consistent findings. Beginning in March 1979, the constant
stream of reared-apart pairs through Minneapolis enabled new analyses
of psychophysiological questions. Nine years later, in 1988, Lykken and
other MISTRA investigators reported skin conductance findings based
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 121

on the same four-group design used in our personality paper: MZA

(n = 43 pairs), MZT (n = 36 pairs), DZA (n = 25 pairs), and DZT (n = 17
pairs).9 The material that follows comes from the original publication
that included numerous detailed findings, charts, and graphs.
The reared-apart twins were older (mean age: 40.8 years, standard de-
viation [SD] = 12.7, range: 12–61 years) than the reared-together twins
(mean age: 21.5 years, SD = 4.8, range: 16–45 years). Three socioeconomic
indices of the separated twins’ adoptive homes were modestly but signifi-
cantly correlated (0.21–0.30, P<.05), as were their fathers’ and mothers’
highest completed school grade (0.25 and 0.40, respectively). However,
none of these variables related to the psychophysiological measures.
The measurement procedures were essentially the same for reared-
apart and reared-together twins. The only exception was the use of a dif-
ferent method of eliciting an estimate of maximal skin conductance re-
sponse for use in what is called range correction. A person’s skin
conductance varies over a particular range because of the nature of the
skin and/or the electrode area, so these factors need to be controlled.10
Twins were seated in the experimental chamber for many of the tasks
they completed in Lykken’s laboratory. Several MZA twins expressed con-
cern over being placed in a small, windowless space with the door shut,
and their co-twins usually had the same response.
In order to estimate the maximum skin conductance response, twins
used a sterile mouthpiece to blow up a small balloon suspended in a clear
cube; a pin at the top of the cube punctured the balloon when it had filled
with sufficient air. The maximum value for reared-together twins was the
peak skin conductance during the task minus the tonic (muscle tension or
contraction) skin conductance level before the task. An additional maxi-
mum estimate was obtained for reared-apart twins: the twins were told
they would hear a loud blast of noise within the next minute following the
balloon test. The maximum value was the amplitude of the skin conduc-
tance response detected within three seconds after the loud noise began.
There was also interest in how quickly the twins habituated to stimuli.
Lykken told the twins he was interested in finding out how fast they could
learn to ignore a series of loud tones. He instructed them to focus on a
story that they would hear from a speaker on the opposite wall. Elec-
trodes were attached to the index and third fingers of both twins’ hands,11
and data collection commenced three seconds after the onset of the stimu-
lus. An explanation of the biometrical model used for data analysis is
provided in the appendix of Lykken’s paper.
The MZT (0.42 to 0.72) and MZA (0.16 to 0.61) twin correlations
for the different skin conductance measures were generally substantial
122 B O R N TO G E T H E R — R E A R E D A PA RT

and exceeded the DZ (DZT and DZA combined) correlations (−0.05 to

0.35), indicating a genetic influence. The MZA twins were about as simi-
lar as the MZT twins on most measures, showing an absence of shared
environmental effects. In addition, the majority of MZT and MZA cor-
relations were larger than twice the DZ correlations, suggesting that skin
conductance characteristics are complex emergenic traits.
Lykken also looked at the MZA data organized by sex. This time, the
MZA males were the most alike on almost all the measures. This was not
the only time that greater MZA male than MZA female twin similarity
was found—the weight and sexual orientation studies also produced such
findings. The observed sex difference in skin conductance could have a
genetic basis in that MZ female twins differ in X-inactivation patterns, or
the silencing of one X chromosome in each of their cells about six to eight
days after conception.12 In fact, genes on the X chromosome can cause one
type of sweat gland anomaly.13 Repeating the biometrical analysis with
only the male twins increased the heritabilities and reduced the shared
environmental estimates to zero. This showed that some factors, possibly
genetic in origin, differentially affect male and female skin conductance
response.
These findings confirmed the 1966 study showing much higher MZ
than DZ twin resemblance in skin conductance response. In fact, about
40 percent of the total variance and most of the stable variance in the
number of trials to habituation appeared to have genetic origins. Varia-
tion in physiological processes associated with sensitivity to insignificant
environmental stimuli may underlie the individual differences Lykken
found in this twin study and in the individual behavioral differences in the
people around us.
These findings mattered because low electrodermal reactivity, al-
though rare among healthy individuals, had been found among patients
with major affective disorders such as bipolar disorder and depression.
Nonresponding also occurs independently of symptoms shown by
schizophrenic, unipolar, and bipolar patients. Based on Lykken’s findings
of genetic influence, skin conductance response might be a reliable index
of predisposition to a psychiatric disorder among people with affected
family members.
Note that the habituation stimuli (tones) consisted of alternate sounds
presented to the twins as they exhaled or inhaled. This procedure bal-
anced the possible effects of sinus arrhythmia (normal increase in heart
rate when breathing in) on one of several cardiovascular measures that
was recorded, but not analyzed in Lykken’s 1988 paper.
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 123

Cardiac Characteristics
Cardiovascular characteristics were of great interest to the medical team
headed by Dr. Naip Tuna, a Romanian-born Turkish cardiologist who
charmed the twins even though he often made them wait. Genetic factors
in heart functions were the subject of a 1989 paper by Drs. Bruce Han-
son, Tuna, Bouchard, and others.14
When the MISTRA was underway, several reared-together twin studies
had reported genetic contributions to variability in measures of the heart.15
One problem with these studies was that the heritabilities of the cardiac
measures were estimated using intraclass correlations alone. As indicated
in Chapter 2, doubling the difference between the MZT and DZT intra-
class correlations estimates a trait’s narrow heritability (additive genetic
effects), but modeling procedures allow further partitioning of genetic and
environmental variance. This situation was rectified by using a model fit-
ting approach that permitted the simultaneous analysis of the reared-
apart twin data from the MISTRA and the reared-together twin data
from studies conducted in Finland and the United States.16
I remember escorting twins to the University of Minnesota’s Variety
Club Heart Hospital for their electrocardiograms (ECGs), echocardio-
grams, and exercise stress test. An ECG traces the heart’s electrical activ-
ity during a cardiac cycle (all cardiac events occurring between two
consecutive heartbeats). An echocardiogram uses ultrasound to produce
one- and two-dimensional images of the heart. Twins especially enjoyed
the echocardiogram because they could see a visual picture of their heart’s
movement. An exercise stress test evaluates the exercise capacity and the
electrocardiographic response (rate, rhythm, and changes in ECG wave-
forms) during maximal stress.
The twins were examined separately, in succession. As usual, they
worked on their questionnaire booklets while waiting to be called, seated
far apart.
The twins in the cardiac characteristics study included twenty-nine
MZA female pairs, twenty MZA male pairs, one MZA male triplet set,
seventeen female DZA pairs, and seven DZA male pairs. The twins’ vec-
torcardiograms (graphic representation of the heart’s electrical activity)
were recorded for forty-three pairs, but only the ECG data were analyzed.
The ECG readings for the first twenty-four twin pairs were measured
manually and read by a cardiologist according to the Minnesota Code
methodology, developed in 1960 and updated in 1987. The subsequently
studied twin pairs had their data read by a cardiologist who confirmed
124 B O R N TO G E T H E R — R E A R E D A PA RT

the computer-generated data and reports. Corroboration was critical, ac-

cording to Hanson:

It truly takes an experienced eye to confirm that an abnormality called by

the computer is correct. It was also possible that a cardiologist might dis-
cover an abnormality that the computer had missed. Tuna would sit there
with his ECG calipers and confirm measurements before signing off on
ECGs from hospital patients. As I recall, Tuna looked at all the electrocar-
diograms, but he allowed his colleagues to officially read them as they were
blind to the twin status of the subjects.

Hanson also emphasized to me that normal variations in cardiac char-

acteristics are of medical interest. There was, however, very little varia-
tion among our twins, and few showed cardiac pathology. For example,
British MZA twins Barbara and Daphne had slight heart murmurs of a
nonserious nature.17 Our only exceptions included two MZA male trip-
lets who showed right bundle branch block (obstruction of conduction
of cardiac electrical impulses along the right bundle branch of the heart’s
conduction system), and one MZA female twin who showed an atrioven-
tricular block—a delay or interruption in conduction between the atria
(the two upper chambers of the heart) and the ventricles (the two lower
heart chambers). These sets were excluded from our analysis. Table 6-2
displays the intraclass correlations for the cardiac characteristics that were

Table 6-2. MZA and DZA Intraclass Correlations for Cardiac Characteristics

MZA Twins DZA Twins

a
ECG Measure ri n (pairs) ri n (pairs)

PR interval 0.78b 49 0.50b 24

QRS duration 0.45b 49 0.70b 24
QT interval 0.65b 49 0.22 24
QTc interval 0.48b 49 0.43b 24
Ventricular rate 0.49b 49 0.34 24
P duration 0.00 23 0.69b 16
ΣQRS (mV)c 0.73b 24 0.49b 18
ΣP (mV)c 0.60b 24 0.16 18
ΣT(mV)c 0.17 24 0.69b 18

a. Parameter definitions are provided in the book’s Web site. ECG: electrocardiogram.
b. P < .05.
c. Measured from the X, Y, and Z leads of the vectorcardiogram. mV: voltage
magnitude.
Source: Adapted from Hanson et al., “Genetic Factors in the Electrocardiogram and
Heart Rate of Twins Reared Apart and Together,” American Journal of Cardiology 63
(1989): 606–609.
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 125

measured. Descriptions of the measured cardiac characteristics are provided

on the Web site.
Three cardiac variables showed statistically significant MZA, but not
DZA, resemblance: the QT interval, the ventricular rate, and the EP (mV).
The P duration correlation was high and statistically significant for DZA
twins, but it was 0.00 for MZA twins, a finding for which there was no
obvious explanation other than the small sample size and/or unreliability
of the measure. The MZA and DZA correlations did not differ signifi-
cantly for any of the nine measures, so biometrical modeling analyses
were performed (to add statistical power), using the American and Finnish
samples as controls. There was no support for shared environmental ef-
fects across the two analyses. We did find that heritability estimates were
consistent across the two studies for the QT interval (US: 0.50, Finland:
0.50), while estimates for the other three measures differed to varying de-
grees (PR interval, US: 0.41, Finland: 0.60; QRS duration, US: 0.30, Fin-
land: 0.51; and ventricular rate, US: 0.52, Finland: 0.46). It was possible
that sample differences in these measures reflected relevant population-
specific genetic factors.
Overall, genetic effects on cardiac characteristics were indicated by
our analyses, but our sample was too small to detect nonadditive genetic
variance. It is also possible that the measurement and recording proce-
dures overlooked small differences in some measures that would only be
detectable with larger samples.
Tuna said the cardiac paper was read with interest and elucidated an
area that was not well known at that time. He did point out that our
findings were indirect, consisting of observational data and statistical
calculations—in contrast, studies conducted after 1989 were using more
direct DNA analyses for evaluating the heritability of ECG measures. He
noted to me, however, that a German study done as recently as 2009 has
confirmed many of our findings (albeit by way of reared-together twins)
through more modern techniques.

Special Mental Abilities

Unlike the closed-door activities that took place in Lykken’s laboratory,
I was very familiar with what happened across campus in Elliott Hall,
the building that housed our psychology laboratory. Special mental abil-
ity testing was a major component of the MISTRA, and we designated a
separate room for this purpose. Special mental ability testing included a
wide variety of verbal, spatial, perceptual, and memory tasks, seventeen
from the Hawaii Ability Battery and fourteen from the Comprehensive
Ability Battery (a description of these tests is provided on the Web site).
126 B O R N TO G E T H E R — R E A R E D A PA RT

Bouchard chose the Hawaii Battery because it had been used previously
in a large family study of cognition, providing a great deal of comparable
data. (Our MZA correlations are contrasted with the Hawaii Study’s
cousin correlations in Chapter 4.) The version we used also included
three tests from the Kit of Factor-Referenced Cognitive Tests (Identical
Pictures, Cubes, and Paper Folding). Two Hawaii Battery tests were omit-
ted (Social Perception and Progressive Matrices) because they resembled
tests already included in the assessment schedule. The Comprehensive
Ability Battery includes twenty tests, although we omitted six (Aiming,
Auditory Ability, Spontaneous Flexibility, Ideational Fluency, Originality,
and Representative Drawing) because of time limitations and because
some of these tests were less reliable and less valid than others. Looking
back, Bouchard regretted not having data from the creativity tests (Spon-
taneous Flexibility and Originality), a decision discussed further in Chap-
ter 11. The Comprehensive Ability Battery’s Esthetic Judgment test was
administered, but it is not a mental ability test and so is not discussed
here.
The sample at this time included forty-nine MZA twin pairs (forty-
seven twin sets and two triplet sets) and twenty-five DZA twin pairs,
although occasional data were missing for some individuals.18 We com-
puted intraclass correlations for all special mental ability tests and tested
the fit of two models to the data. (Additional details about these proce-
dures are available in our 1989 publication, which is discussed later in
conjunction with information processing.)19
The mean MZA special ability correlations (0.45, 0.48) exceeded the
mean DZA correlations (0.34, 0.35) for tests from both test batteries. We
also used a statistical procedure called factor analysis to reduce the sepa-
rate tests to four factors (Verbal, Spatial, Perceptual Speed and Accuracy,
and Visual Memory). The mean intraclass correlations and model fitting
results are summarized in Table 6-3.
Genetic effects were generally highest for the Spatial Ability tests and
lowest for the Visual Memory tests. This pattern repeated in our subse-
quent analyses, suggesting that memory is affected substantially by expe-
rience and training. Information that is best remembered may be either
personally meaningful or important for survival, so individual differences
are minimized. We also found significant genetic effects for some individ-
ual tests (e.g., Pedigrees, a test in which subjects answer questions about
family relationships displayed schematically).
The final section of our article presented a meta-analysis or summary
of all special mental ability twin correlations that had been reported up
to the time of our study. It was Bouchard’s idea to plot the previous twin
Table 6-3. Reared-Apart Twin Intraclass Correlations (Mean) and Genetic Model Analysis for the Special Mental Ability Tests

Test MZA ri DZA ri h2

Hawaii Ability Battery 0.45 0.34 0.47

Comprehensive Ability Battery 0.48 0.35 0.52

Parameter Estimates P values

Factor MZA ri DZA ri G E h2 General Model No Genetic Effect

Verbal 0.57 0.51 0.60 0.45 0.57 .33 .001

Spatial 0.71 0.40 0.74 0.30 0.71 .98 .001
Perceptual speed and accuracy 0.53 0.54 0.56 0.49 0.53 .18 .001
Visual memory 0.42 0.07 — — — .006 —

Note: Parameter estimates are missing when the genetic model did not fit the mean squares. The mean battery correlations and heritabilities are from
our 1990 paper and the factor correlations and genetic test data are from our 1989 chapter. E: environmental effect; G: genetic effect; h2: heritability.
Source: Adapted from Bouchard et al., “Genetic and Environmental Influences on Special Mental Abilities in a Sample of Twins Reared Apart,” Acta
Geneticae Medicae et Gemellologiae (Twin Research) 39 (1990): 193–206.
128 B O R N TO G E T H E R — R E A R E D A PA RT

data as four stem-and-leaf plots organized by the four ability factors (Ver-
bal, Spatial, Perceptual Speed and Accuracy, and Memory) and four twin
groups (MZA, MZT, DZA, and DZT). These plots allow the spread of
correlations across studies to be seen easily. The diagrams (which are not
reproduced here) let us compare our findings with findings from multi-
ple studies. Here, we calculated the heritabilities for reared-together and
reared-apart twins by doubling the difference between the MZ and DZ
twin correlations. The heritabilities of the four ability factors yielded by
the meta-analysis were very close to ours. However, except for Memory,
the estimates were slightly higher for reared-together than reared-apart
twins (Verbal 0.36, 0.28; Spatial 0.48, 0.44; Perceptual Speed and Accu-
racy 0.42, 0.16; and Memory 0.32, 0.40, respectively), suggesting that
shared environmental factors can affect special mental skills. The rela-
tively low 0.16 heritability for Perceptual Speed and Accuracy, based on
the reared-apart twins, could have reflected the high DZA correlation for
speed tests and/or sampling variation associated with the small number
of tests.
When I left the University of Minnesota for California in 1991, I took
copies of every MISTRA protocol with me. Looking through these mate-
rials years later reminded me of the mental ability instruction booklet we
used exclusively with the twins. Completion times were marked in red,
and important guidelines were underlined. We penciled in notes regarding
particular twin pairs when called for (e.g., “12-Mar-85, Sally had me re-
peat the instructions, but got the concept,” “4-Oct-1988, John coughed
repeatedly through the Spelling test, but he finished it”). This booklet was
an informative record of what actually occurred in the testing room, and
we discussed such events at staff meetings held after each assessment.

Information Processing
Besides showing us that genetic and environmental factors variously affect
the different special mental abilities, the data were informative in another
way. McGue and Bouchard used the data to explore relationships be-
tween information processing and special mental abilities, as they had
done in 1984. The MISTRA data also provided a look at the extent to
which individual differences in cognition could be linked to specific fea-
tures of the twins’ rearing homes. For example, it was possible that twins
from higher socioeconomic backgrounds outperformed twins from lower
socioeconomic backgrounds, or that twins whose parents subscribed to
books and magazines excelled on verbal tests relative to the twins whose
parents provided little reading material.
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 129

This research would also help decide which of two hypotheses—innate

neurological structure, or experience producing drive—best fit the gene-
tic processes involved in cognition. These perspectives are not necessarily
incompatible.
The Innate Neurological Structure Theory, so named by McGue and
Bouchard, posits that structural and functional brain differences underlie
individual differences in IQ.20 Thus, genetic differences in cognition would
be explained by genetic differences in neurological structures. In support
of this theory, psychologist Tony Vernon at the University of Western
Ontario in Canada showed that information processing tasks involving
speed are associated with general intelligence test scores.21 For example,
the speed with which items are recalled or object similarities and differ-
ences are recognized shows a moderate but consistent relationship with
general mental ability.
The Experience Producing Drive Theory proposes that intelligence con-
sists of learned facts and skills, such as knowing the names of capital cities
or the constellations of the northern hemisphere.22 The theory claims that
individuals seek experiences commensurate with their intellectual levels.
Inherited intellectual potentials are the tendencies to engage in activities
that facilitate learning, such as reading or observing. Thus, the inherited
potentials are not the intellectual capacities themselves.
The Experience Producing Drive Theory encapsulated the concept that
individuals actively select and create their own environments. When the
theory was conceived in 1962, it referred to multiple drives and a single
general learning capacity. In 1995, Bouchard reformulated this perspec-
tive as the Experience Producing Drive Theory–Revised, suggesting that
specialized brain structures underlie specialized mechanisms that affect
capacity and drive.23 Simply put, Bouchard wrote, “genes drive behavior
and . . . behavior determines the environments we experience.”24
The theory could potentially tell us how and why our MZA twins were
so behaviorally alike, despite their separate rearing. The MZA twins Brent
Tremblay and George Holmes unnerved Brent’s father with their focused
discussion of sports statistics when he drove them home after dinner one
evening. People with sharp mathematical minds and good memories might
stay attuned to statistical tidbits about census changes or team sports, de-
pending on their interests.
In support of Experience Producing Drive Theory–Revised, positive
associations have been reported between children’s cognitive abilities and
the features of their homes, such as their parents’ socioeconomic status.
However, genes and environments cannot be separated when biological
parents raise their biological children. Children from more advantaged
130 B O R N TO G E T H E R — R E A R E D A PA RT

families might show superior language skills relative to children from less
advantaged families, but it would be wrong to interpret parental status as
causally connected to children’s behavior. Parents transmit both genes and
environments to their children, and more socially advantaged parents are
likely to provide their children with rich linguistic opportunities and
the genetic potential for good language skills. The MISTRA disentangled
the genetic and environmental effects by examining associations between
the twins’ abilities and the characteristics of their separate rearing homes.
The participants in the information processing study were the same
forty-nine MZA and twenty-five DZA twin pairs included in our special
mental ability study. The three information processing tasks were the Pos-
ner Letter Identification, the Sternberg Memory Search, and the Shepard-
Metzler Cube Rotation (all described in Chapter 4), and the ability mea-
sures were those from the Hawaii and Comprehensive Ability Batteries.
The four ability factors (Verbal Reasoning, Spatial Ability, Perceptual
Speed and Accuracy, and Visual Memory) accounted for 55.1 percent of
the variance in ability. These results generally agreed with those from the
original 1976 Hawaii Family Study of Cognition.25
The twins’ special abilities and information processing measures were
correlated with their adoptive parents’ years of education, the adoptive
mother or fathers’ highest occupational status, the twins’ number of resi-
dences between ages one and nineteen years, and the availability of forty-
one different physical facilities in the home. The forty-one home facilities
were organized into four independent groups by factor analysis: Material
Possessions (e.g., tape recorder, power tools), Scientific/Technical (e.g., stop-
watch, telescope), Cultural (e.g., atlas, classical records), and Mechanical
(e.g., fishing equipment, leather-working tools). The physical facilities
checklist, mentioned in Chapter 5 in conjunction with the IQ data, is de-
scribed more fully here.
The fifty-seven correlations computed between the special abilities and
rearing-home characteristics (parental education, parental occupation, and
number of residences) ranged between −0.19 and 0.16. Only four correla-
tions were statistically significant, but three of the four appeared in a
counterintuitive direction. For example, one of the three correlations sug-
gested that lower cognitive ability was associated with a higher rearing
social status. Several significant correlations between the cognitive ability
and physical facilities measures also exhibited a counterintuitive direc-
tion. McGue and Bouchard suggested that “whatever the relevant envi-
ronmental contributors to a child’s cognitive development may be, these
factors are unrelated to the educational and socioeconomic status of that
child’s parents.”26 In other words, environmental factors clearly affect
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 131

cognitive skills, but they are not the parental factors people generally as-
sociate with child development. Psychologists need to look more widely
for these influences, perhaps in children’s peer groups, community cen-
ters, and extracurricular activities.
Significant twin resemblance was found for parental education, mate-
rial possessions, and mechanical possessions, indicating that the twins’
rearing homes were not chosen entirely at random. However, McGue and
Bouchard did not attach considerable meaning to this finding because the
environmental measures were only weakly linked with the cognitive mea-
sures, and because the degree of environmental similarity was the same
for the MZA and DZA twins. Twin similarity in rearing factors did not
appear to explain twin similarity in special mental abilities or informa-
tion processing.
The three information processing components (Speed of Response,
Acquisition Speed, and Speed of Spatial Processing) were the same as those
we reported for the smaller 1984 and 1986 twin samples.27 The relation-
ships found between these components and the special abilities were also
similar to those found earlier, showing overlap between these two types
of measures.
The genetic analysis of the special ability data showed that, with two
exceptions, the general model fit the data, and all genetic components
were statistically significant. About 50 percent of the variance in Verbal
Reasoning, Spatial Ability, and Perceptual Speed and Accuracy was asso-
ciated with genetic factors. The two exceptions were the Hawaii Ability’s
Word Beginning and Endings test (generating words beginning and end-
ing in specified letters) and the Immediate Visual Memory factor (mem-
orizing an object list and immediately identifying those objects in a new
list). The general model fit most of the information processing measures
except for two Sternberg parameters (percentage correct, and secondary
or long-term memory).
McGue and Bouchard noted (somewhat surprisingly) that the amount
of time required for twins to decide whether two letters are the same or
different correlated as highly with verbal abilities (mean: −0.39) as the
verbal abilities correlated among themselves (mean: 0.43). The reasons
behind such results are speculative. For example, the correlation between
the speed factor and the specific abilities could be due to a general speed
factor common to all the speed measures. Alternatively, several indepen-
dent processes with specific associations to the cognitive skills, combined
into the composite factor, could be the cause as well. The MISTRA data
did not allow a decision as to which of these two alternatives was the
better interpretation.
132 B O R N TO G E T H E R — R E A R E D A PA RT

The component and composite speed measures differed in terms of the

pattern and magnitude of genetic effects. McGue and Bouchard suggested
that the composite measures might have reflected inherited features of
neurological structure that are important to a wide range of tasks, consis-
tent with the Innate Neurological Structure Theory. However, the compo-
nent speed measures could have reflected acquired processes important to
problem-solving in specific cognitive domains, consistent with Experience
Producing Drive Theory.
The MZA twins’ similar abilities and information processing skills
may come from their having the same genetic blueprint that guides them
toward interesting opportunities and the mastery of new skills. This can
help us understand why Mark and Jerry became firefighters, why Bar-
bara and Daphne found humor in most things (Figure 6-1), and why
Oskar and Jack preferred reading books from back to front.

1989 to 1990
The genetic effects found for job satisfaction, heart rate, mental abilities,
and information processing were generally what I expected. The striking

Figure 6-1. The “Giggle Twins,” Daphne Goodship (left) and Barbara Herbert.
The twins had been close friends for thirty years; Daphne’s passing in 2011 in
her early seventies was a great loss for her sister. (Photo credit Nancy L. Segal.)
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 133

finding from so many of our analyses was the negligible effect of the
rearing-home environment on behavior. Recall, also, the finding that envi-
ronmental effects outside the home were important, and often explained
more of the variance than the genetic ones. For example, cardiac character-
istics were affected by nonshared environments, but not by shared envi-
ronments, and intrinsic job satisfaction was only 30 percent heritable. Both
findings surprised people who subscribed to the more conventional view
that families significantly influence individual differences in their children’s
behavior—especially people who had never worked with twins, adoptees, or
children.
I have found that parents of dizygotic multiple birth children and par-
ents of adopted children are very wise regarding behavioral origins. They
know without being taught that their sons and daughters have distinctive
cognitive skills, temperaments, and talents and—most significantly—that
they differ considerably from their siblings no matter how much their par-
ents try to raise them the same way. In February 2010, I had the rare op-
portunity to attend an awards reception at the White House where I met a
celebrity mother of female dizygotic twin infants. “They are so different!”
she told me. “Is that normal?” Similarly, the father of a virtual twin pair
in one of my studies (same-age unrelated siblings who provide direct esti-
mates of shared environmental influence on behavior)28 had expected his
daughters to be somewhat different, “but not like night and day.”29
The reared-apart twins kept us occupied in many ways during 1989
and 1990. Bouchard and postdoctoral fellow Kimerly Wilcox published
an encyclopedia article on new developments in behavior genetics.30 Find-
ings from the MISTRA were included in that article (and would also ap-
pear in the 1992 and 1997 updates). Bouchard and I published a com-
mentary in response to a target article on sex differences in mathematical
reasoning ability.31 Bouchard also contributed to an adoption bibliogra-
phy published by the American Adoption Congress, something he could
not have done had it not been for the MISTRA.32 Our Science IQ article
was reprinted in Le Journal International de Médecine, an international
European journal,33 and Lykken published a full description of the Min-
nesota Twin Registry.34
A curious twist in the history of the MISTRA is that while federal fund-
ing was mostly denied, our study laid the groundwork for the success of
the Minnesota Twin Family Study, which has been federally funded for more
than twenty years since its launch in 1989. Lykken’s Minnesota Twin Regis-
try, started in 1983 and completed in 1990, also received federal funding.
The MISTRA team continued to share its findings with colleagues.
Bouchard and I delivered twenty-five formal presentations, many at
134 B O R N TO G E T H E R — R E A R E D A PA RT

scientific meetings, and appeared on approximately twenty television or

radio programs. In addition to studying six new sets of twins in 1989 and
five in 1990, we studied four follow-up sets, two adoptive siblings of the
reared-apart twins, and several of their biological siblings. Eighteen arti-
cles about the study appeared in newspapers, magazines, and other
sources, including the New York Times Magazine, Science News, and the
1990 World Book Year Book.
I didn’t see all these articles when they first appeared, but Bouchard col-
lected them in annual reports that he sent to the study’s funding sources.
It was worth reading them years later because they reminded me of why
people appreciated the twins’ stories so much. “Keith Quilter’s happiness
will be complete when the twin brother he grew up not knowing joins
him for his wedding day,” wrote British newspaper columnist Allison
Kelly.35 “When we first laid eyes on each other it was just a shock,” said
Jake Hellbach of his reared-apart twin, Keith Heitzmann. “And it seemed
like right off, we had this bond, right from the second we met each other.”36
Mark Newman observed that meeting his reared-apart twin Jerry Levey

Figure 6-2. Professor Bouchard and his “identical twin.” (Photo credit Rich
Ryan Photography.)
 J O B S AT I S FAC T I O N , C A R D I AC C H A R AC T E R I ST I C S , A N D M O R E 135

was “like returning from a vacation”—when they were reunited, their

rapport was instant, and they only had to fill in the details.
It is understandable that news of twins finding one another at any age
elicits happiness from the twins, their families, and the public. The fact that
twin reunions are news shows what people value. Bouchard was not a twin,
but one photographer pretended that he was, as shown in Figure 6-2.
As we approached the 1990s, there was evidence that findings from
the MISTRA were affecting the ways that some scientists thought about
behavior and the value of behavioral genetic studies. According to an-
thropologist Melvin Konner, the substantial contribution of genetic ef-
fects and the negligible impact of shared environments were seen to “fly
in the face of our modern beliefs about how we have come to be who we
are.” He added that “it is perhaps the crowning irony of current research
that we need behavior genetics to help us find out.”37
 CHAPTER SEVEN

Psychopathology and Religiosity (1990)

Tourette Syndrome
I learned about Tourette syndrome when I was a graduate student assis-
tant at the Illinois State Psychiatric Institute in Chicago between 1974
and 1982. I noticed someone there who showed continual grunting and
clearing of the throat, and periodic twirling of a lock of hair with the
forefinger of one hand. I asked one of my supervisors about these behav-
iors, and he said, “Tourette.”
Tourette syndrome was first described in 1885 by the French neurolo-
gist Georges Gilles de la Tourette.1 The symptoms are distinctive—they
include involuntary motor and phonic tics that appear by age seven, but
they can appear as late as age eighteen. Some patients also display copro-
lalia, the use of loud and obscene language. So, in 1985, when Bouchard
and I were visited by one female member of an affected MZA female/
DZA male Swedish triplet set, I knew what I was seeing.
This was my first meeting with Elsa, who had participated in the MISTRA
in 1981 at age fifty, the year before I arrived. She took part in the study with
her MZA twin sister Frida and co-triplet brother Erik, both of whom
showed the signs and symptoms of Tourette. These siblings were separated
at two months of age and reunited at forty-eight years. Elsa didn’t know she
was adopted until she was twelve, when she asked her mother for birth
documents required by her school. When her mother brought the papers
home she momentarily hid the word adopted with her hand before telling
her daughter that she was an adopted triplet. When she finally revealed the
truth, Elsa cried and insisted to her mother, “I am yours.”
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 137

Growing up, Elsa felt different from her family, partly because she be-
lieved her parents were much smarter than she was, but also because of
her tics. She told me she developed frequent eye blinking at age three, and
in childhood began counting things obsessively and turning her body con-
tinuously in a clockwise direction as she walked down stairs. Some of her
relatives who knew she was adopted believed that her adoption experi-
ence had provoked these symptoms.
Over the years, Elsa was both curious and worried that her biological
triplet siblings would reject her. Then, when Elsa was in her forties, she
became a headmistress at a daycare center in a Swedish community away
from her home. One day, she was “drawn into a certain shop” to purchase
some dress material. The shopkeeper greeted Elsa warmly, not realizing
that she didn’t really know her—when Elsa said that she had a twin, the
shopkeeper explained that she had made a wedding dress for Frida and
had confused the two. At this point, Elsa asked the woman to deliver a
message to her sister. When Frida called her at work, Elsa replied, “I know
your name.” This was the first contact between these sisters in over forty
years. They met their brother several weeks later after finding him in
Sweden through hospital and church records.
The triplets presented a new medical case history. No one had formally
diagnosed Tourette syndrome in reared-apart twins or triplets, although
some characteristic behaviors had been described in two early cases.2
Tourette-like symptoms were also noted in a pair of our thirty-five-year-
old MZA female twins. Both twins were compulsive counters of just about
anything, keeping track of the wheels on trucks and counting themselves
to sleep. Such repetitive, ritualistic behaviors are part of the suite of
symptoms characterizing Tourette, but they do not alone warrant a diag-
nosis of the syndrome.
Several studies of Tourette syndrome in twins reared together had
been published by the time Elsa returned to Minneapolis in 1985. A study
published that same year3 and several subsequent clinical studies demon-
strated genetic contributions to the disorder.4 Thirty years later, in 2005,
Yale University researchers would discover that chromosome 13 inversion
(rearrangement of genetic material within a chromosome) was linked to
some cases of Tourette, providing new evidence that this disorder has a
genetic basis.5
In September 1983, psychiatrist Maurice W. Dysken, whom I had known
at the Illinois State Psychiatric Institute, had relocated to the Minneapolis
Veterans Administration Medical Center. Dysken arranged to have Elsa
observed in an overnight sleep laboratory at the nearby Hennepin County
Medical Center. We also reviewed all three triplets’ medical records and
138 B O R N TO G E T H E R — R E A R E D A PA RT

interviewed Elsa more extensively about the nature and timing of her
symptoms. We arranged for all three triplets to complete a survey, based
on questionnaires created by Yale University psychologists David L. Pauls
and Kenneth K. Kidd. These surveys were translated into Swedish and ad-
ministered to the triplets in 1987 and 1988 by nursing staff at the Karolin-
ska Institute in Stockholm. They included questions about each triplet’s
symptoms and those of any biological family members with whom they
were familiar. The triplets had half-brothers and half-sisters on both sides
of their biological family and had had meetings and telephone contact
with some of them.
All three triplets met Diagnostic Statistical Manual (DSM)-III criteria
for Tourette syndrome, demonstrating genetic influence.6 We discovered
that the triplets’ biological father and uncle were both affected with the
disorder.7 The fact that the triplets were not raised with these relatives
strengthened a genetic explanation of this disorder. Adding to this view
was the fact that the son and daughter of Elsa’s MZA female co-triplet
had both been affected. Most importantly, all three triplets had been
reared apart since early infancy, and each had developed his or her symp-
toms independently. The MZA female co-twins did not show identical
symptom profiles, a finding that we linked to unknown environmental
causes. This outcome had been reported before—the famous identical
Genain quadruplets were concordant for schizophrenia, but varied mark-
edly in the nature and severity of their symptoms.8
This Swedish triplet set with Tourette disorder was the only MISTRA
case study we ever reported in the scientific literature; we preferred to
publish statistical findings on as many separated pairs as possible. It was
also important to honor our informed consent agreements with the twins
by keeping their life history data confidential.9

Psychopathology
Our first quantitative analysis of psychiatric symptoms appeared in 1990
in the journal Biological Psychiatry.10 The lead author on this effort was
clinical psychologist Will Grove. Grove earned his doctoral degree from
the University of Minnesota’s clinical psychology program in 1983,
working under Paul Meehl; he then spent a year in the University of
Iowa’s psychiatry department. He returned to the University of Minne-
sota as a faculty member in psychiatry in 1985, and also joined the psy-
chology department in 1990. His research specialties include the etiologi-
cal and behavioral genetic aspects of psychopathology. Grove, along with
Elke Eckert, administered the Diagnostic Interview Schedule to many of
our twins beginning in 1990.
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 139

The Diagnostic Interview Schedule is a structured protocol used with

nonclinical populations (i.e., healthy individuals without psychiatric diag-
noses). It covers a number of DSM-III diagnoses, but only three—alcohol
abuse/dependence, drug abuse/dependence, and antisocial personality
disorder—were examined with regard to our reared-apart twin sample.
Investigations in the 1960s and 1970s had shown that psychiatric dis-
orders and symptoms in children and adults (e.g., alcoholism, drug abuse,
and antisocial personality traits) run in families. Environmental explana-
tions were often attached to these findings, although evidence of genetic
effects was accumulating from twin and adoption studies. In fact, gene-
tic effects on psychiatric disorders had been suggested in the previous
MZA studies. Depending upon the criteria for separation, these cases
included twelve or fourteen pairs with schizophrenia (seven or nine con-
cordant).11 Co-twins in one other pair were concordant for bipolar disor-
der, and several sets were concordant for depressive disorders. These early
studies also indicated genetic effects on alcohol consumption. Seven sets
were concordant for nondrinking, three sets were concordant for heavy
drinking, and six sets were concordant for moderate or light drinking;
three discordant sets included one light drinker and one nondrinker.12
Our first formal analysis included sixty-five members of thirty-two
MZA twin and triplets sets who had participated in the study between
1979 and 1988. These twins had completed the psychiatric interview sepa-
rately and with different examiners who were blinded to the twin’s zygos-
ity and other information about them. The Diagnostic Interview Sched-
ule was not introduced into the MISTRA until 1982. Prior to that time,
psychiatric interviews for the first forty-two pairs were unstructured in
format and nonblind in administration. In order to increase the replica-
bility of the findings and to control for possible examiner bias, all new
pairs and all follow-up pairs studied in 1982 and later completed the in-
terview under the new conditions.
The twins were normal volunteers, meaning that very few met the cri-
teria for Diagnostic Statistical Manual-III Axis I psychiatric disorders or
antisocial personality.13 Grove, therefore, decided to use signs and symp-
toms that counted toward DSM-III, Research Diagnostic Criteria, or
St.  Louis group criteria diagnoses of alcohol abuse/dependence, drug
abuse/dependence, childhood aspects of antisocial personality disorder,
and adult aspects of antisocial personality disorder.14 In other words,
some individuals might show several symptoms of drug abuse (e.g., took
drugs five times or more) or antisocial personality (e.g., was fired from
two or more jobs), but not enough to warrant a diagnosis of these condi-
tions. Four quasi-continuous scores were obtained for these symptom
categories.
140 B O R N TO G E T H E R — R E A R E D A PA RT

Twelve individual twins satisfied the criteria for alcohol abuse and/or
dependence (18 percent), nine for drug abuse and/or dependence (14
percent), and seven for antisocial personality (11 percent). The proband-
wise concordance rates (probability that a twin is affected if his or her
co-twin has a given condition) for these conditions were 33 percent,
36 percent, and 29 percent, respectively.15 These concordance rates were
low, but they did not mean that genes were unimportant—rather, we sus-
pected that using diagnoses was inappropriate because our twin sample
was generally nonclinical. Grove explained that if two twins are heavy
drinkers, one twin might fulfill the criteria for alcohol abuse, but his or
her co-twin, while a heavy drinker as well, might miss the benchmark
for abuse. Such twins would be discordant according to formal criteria
for alcohol abuse, but they would both be alcohol abusers. He thus ana-
lyzed the data using symptom counts.
The heritabilities found for drug abuse/dependence (0.45), childhood
antisocial personality (0.41), and adult antisocial personality (0.28) were
significant at the 0.10 level. The heritability for alcoholism (0.11) was
lower than the 0.67 probandwise concordance value that had recently
been reported for reared-apart twins in Finland, although it was similar to
the 0.14 intraclass correlation reported for reared-apart twins in Sweden.
We also computed heritabilities, phenotypic correlations (extent of re-
semblance between two observed measures), and genetic correlations
(extent to which two traits are affected by common genes).
We found a high genetic correlation between alcohol and drug abuse/
dependence (0.78), suggesting that these disorders are affected by shared
genetic factors. However, the phenotypic correlation between alcohol and
drug abuse/dependence was low (0.26). When one trait shows low herita-
bility (e.g., alcoholism) but shows a high genetic correlation with another
trait (e.g., drug abuse), then there are probably only a few genes that mod-
estly influence both traits. The genetic correlations for the other behaviors
ranged from 0.53 to 0.87, suggesting genetic overlap among them.
There were several limitations to this study. First, the data were not
gathered on more than one occasion to determine whether the twins’ re-
sponses remained constant. Second, some younger twins may not have
passed through the risk periods for symptom development, resulting in
“pseudo discordance.” Third, despite the power of the MZA twin design
to separate genetic and environmental effects, the sample size was mod-
est, running the risk of small sample fluctuation. Fourth, more complex
genetic models, encompassing genotype × environment interactions and
environmental effects, could not be applied because of the small sample.
Even while the twins were nonclinical volunteers, the rates of alcohol
abuse/dependence, drug abuse/dependence, and antisocial personality
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 141

slightly exceeded those reported in the general population: 11 percent,

5  percent, and 2 percent, respectively. Therefore, our sample was not
“super-normal,” as some of our colleagues had suspected. Of course,
our 1990 analysis included only a subset of the reared apart twins (31
MZA pairs and one triplet set). It was possible that sampling variation
due to the small sample size contributed to the high rate of antisocial
personality.
In August 1990, I attended the World Psychiatric Association’s Re-
gional Meeting in Oslo, Norway. I presented a sampling of the findings
from our studies of Tourette disorder, substance abuse and antisocial be-
havior, and personality. I also presented an updated analysis of our stud-
ies on sexual orientation that built upon our 1986 article.16
When the 1986 article appeared, we had studied fifty-five pairs and had
observed concordance for homosexuality in one MZA male pair, ques-
tionable concordance in another MZA male pair, and discordance in four
MZA female pairs. Four years later, we had studied ninety-five separated
twin sets, sixty-two MZA and thirty-three DZA. We identified eight MZA
twin pairs (three male and five female) and two DZA pairs (one male and
one female) in which at least one member displayed homosexual or bi-
sexual preferences. Twins in our most recently studied MZA male pair at
that time were discordant for homosexuality. Twins in our more recently
studied MZA female pair and DZA pairs were also discordant. As in our
1986 study, the homosexual female twins had experienced delayed pu-
berty and a larger adult maximum body weight relative to their hetero-
sexual co-twins.
It still seemed to us (at the time) that male homosexuality might involve
a complex genotype × environment interaction, whereas female homosex-
uality was more closely tied to environmental influences because none of
the MZA female pairs were concordant. The lack of concordance in our
two DZA pairs was consistent with a genetic contribution to sexual ori-
entation, although the small number of twins was limiting. This updated
analysis was published in the 1990 proceedings from the conference in
Norway.17

Religiosity
Behavioral Genetics is the most complete compendium of behavioral
genetic research around.18 This book gets larger with each reprinting, a
change that reflects the accumulation of new methods, findings, and per-
spectives. Some sections undergo considerable revision, making it in-
structive to compare older and newer editions as a way of tracking prog-
ress in the field. It has been revised four times since its first publication in
142 B O R N TO G E T H E R — R E A R E D A PA RT

1980.19 The 1990 version included the following passage on religious

beliefs and commitment:

Although some attitudes and beliefs, such as conservatism, appear to be in-

fluenced by heredity, others are, as we might expect, due primarily to shared
environmental influences. In Loehlin and Nichols’s (1976) large study of
high school twins, identical and fraternal twin correlations were, respec-
tively, 0.56 and 0.67 for belief in God and 0.60 and 0.58 for involvement in
religious affairs.20

That same year the MISTRA published a genetic analysis of religious

interests, attitudes, and values.21 Compelling evidence of genetic influence
on religiosity was observed, a finding that was replicated by others over
the next few years. This body of work explains the changes that were ap-
parent in subsequent editions of Behavioral Genetics—the relevant sec-
tion in the 2008 volume, most recent to the time of this writing, reads:

A twin-family analysis confirmed heritabilities of about 50% for traditional-

ism as well as showing high heritabilities for sexual and religious attitudes . . .
Religious attitudes were the focus of a special issue of the journal Twin Re-
search . . . A recent study suggests that the heritability of religiousness in-
creases from adolescence to adulthood.22

The wave of behavioral genetic analyses of attitudes in the 1990s was

probably inspired by the 1986 Australian study showing genetic effects (50
percent) on conservatism and traditionalism.23 People were variously ex-
cited by and skeptical of these findings. Two of the more excited individu-
als were University of Minnesota psychology graduate students, Niels
Waller and Brian Kojetin. Waller and Kojetin asked Bouchard if they
could analyze the MISTRA religious interest data, and he agreed.
The results were striking—about 50 percent of the observed variance
in five measures of religiosity was tied to genetic factors. Waller wanted
to turn this work into his doctoral dissertation, but Bouchard talked him
out of it. Genetic influence on religious attitudes and values was contro-
versial at that time, and Bouchard worried that Waller wouldn’t get a job
upon graduating. Waller and Kojetin still analyzed the religiosity data,
publishing their article in 1990 in Psychological Science.24 The study in-
cluded five measures of religious interests, attitudes, and values that were
embedded within the standard inventories we administered:

• The Religious Fundamentalism Scale. This scale included twelve

items from the Minnesota Multiphasic Personality Inventory that
assessed fundamental religious beliefs.25
• The Religious Occupational Interests Scale. This was a four-item
scale that measured interest in religious occupations.26
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 143

• The Strong-Campbell Scale. This scale measured interest in reli-

gious occupations, school subjects, and activities. It was modified
slightly from the Religious Activities Basic Interest Scale of the
Strong-Campbell Interest Inventory.27
• The Religious Leisure Time Interests Scale. This was a five-item
scale that measured interest in religious leisure time activities such
as attending services.28
• The Allport-Vernon-Lindzey Religious Values Scale. This was one
of six scales that assessed the salience of a religious values system.29

The reared-apart twin sample included fifty-three MZA twin pairs (twenty
male and thirty-three female) and thirty-one DZA twin pairs (nine male
and twenty-two female). The reared-together twin sample from Lykken’s
registry included 458 MZT twin pairs (156 male and 302 female) and
363 DZT twin pairs (123 male and 240 female). The reared-apart twins
completed all five scales, while the reared-together twins completed the
Religious Leisure Time Interests and Religious Occupational Interests
Scales only. Intraclass correlations for the different scales are summarized
in Table 7-1.
The MZA and MZT correlations were uniformly higher than the cor-
responding DZA and DZT correlations, consistent with genetic effects. We
also noted that the Leisure Time scale correlations were higher for reared-
together twins than for reared-apart twins, although the differences were
not statistically significant because of the small reared-apart twin samples.

Table 7-1. Intraclass Correlations for Five Religious Interest Scales for Twins Reared Apart and Together

MZT DZT
Scale (458 pairs) (363 pairs)

Leisure time 0.60 0.30

Occupational interests 0.41 0.19

MZA DZA
(31–52 pairs) (21–31 pairs)

Leisure time 0.39 0.04

Occupational interests 0.59 0.20
Leisure time 0.55 −0.22
Occupational interests 0.49 0.15
Leisure time 0.55 −0.08

Source: Adapted from Waller et al., “Genetic and Environmental Influences on

Religious Interests, Attitudes, and Values: A Study of Twins Reared Apart and Together,”
Psychological Science 1 (1990): 138–142.
144 B O R N TO G E T H E R — R E A R E D A PA RT

Model fitting was performed twice, once using all four twin groups who
had completed two scales, and once using the MZA and DZA twins
who had also completed the other three scales. The genetic analyses let
Waller and Kojetin partition the environment into shared and nonshared
components. Neither a purely genetic nor purely environmental model
could explain the data. Instead, genetic influence explained approximately
50 percent of the variance in all five measures, as did nonshared environ-
mental factors.30 The shared environment contributed little to the vari-
ance. The twins’ Verbal, Performance, and Full-Scale IQ scores could not
account for the genetic component underlying religiosity.31
The 50 percent genetic influence on religiosity and the negligible shared
environmental effect were fairly new and controversial findings. These find-
ings meant that parents had less influence than they thought over their
children’s religious activities and interests as they approached adolescence
and adulthood. Instead, the study showed that religious pursuits were af-
fected more by individuals’ behavioral predispositions in conjunction with
their unique experiences, such as attending a workshop, taking a class, or
reading a book.
Bouchard considered the findings striking. Years later, he told Time
magazine that the data “completely contradicted my expectations.”32
When Bouchard presented the religiosity results and others at the 1990
American Association for the Advancement of Science meeting, biologist
Garland Allen objected: “Behavioral genetic studies, such as these, tend to
lead the public to blame behavior, such as alcoholism and criminality,
on genetics.”33 However, behavioral geneticists believed that the findings
could help the public understand these behaviors and work more effec-
tively toward their control.
Subsequent twin and family studies (including our own) continued to
support the original results. Debate over the degree to which genes influ-
ence religious behaviors has continued, although most researchers agree
that religious affiliation (e.g., Catholicism or Judaism) is decided by one’s
rearing family.
The MISTRA investigators believed that complex behaviors such as reli-
gious interest are probably influenced by many genes, each having a tiny
effect on the outcome. It made sense that the pathways from genes at the
molecular level to behavior at the observable level might operate through
personality traits such as traditionalism, absorption (emotional responsivity
to engaging sights and sounds), or harm avoidance (tendencies that might
encourage individuals to seek religious activities or community support).
Several MZA twin pairs were memorable because of their religious at-
titudes and activities. Debbie Mehlman and Sharon Poset, who met at age
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 145

forty-five, are the best example of twins whose affiliations differed but
whose religious involvements were closely aligned. The twins were sepa-
rated as infants and adopted by different New Jersey families—living just
forty-five minutes apart. Debbie was raised Jewish, while Sharon was raised
Catholic before becoming a Protestant.
Debbie learned at age forty-five that she had been adopted and that
she had a twin. One weekend after Debbie’s father had died, her mother
decided that the time was right to tell her daughter about her true back-
ground; Debbie thought her mother’s visit was for the “money talk”—to
acquaint Debbie with the family’s bank accounts “in case anything hap-
pened.” “Boy, was I in for a surprise!” she recalled. Debbie hired a private
investigator who located Sharon in Nicholasville, Kentucky, about 650
miles from Debbie’s West Hartford, Connecticut, home. Sharon knew she
had been adopted, but neither she nor her family knew she had a twin.34
The private investigator first contacted Sharon’s parents in New Jersey,
and they decided to travel to Kentucky to tell Sharon the news. Sharon
also assumed her parents were coming for the “money talk.”
Prior to being found by her sister, Sharon had prayed that someone spe-
cial would come into her life after her son had relocated to Colorado. She
said Debbie was the answer to her prayers. The twins’ televised reunion
shows them falling into each other’s arms at Bradley International Airport
in Windsor Locks, Connecticut.35 A photo taken a month later shows them
wearing t-shirts imprinted with the words “I’m With My Twin Sister.”36
After learning she was adopted, Debbie worried that she was not really
Jewish because Jewish law states that Jewish children are born to Jewish
mothers. Consequently, Debbie completed a conversion ritual along with
her daughter, and remarried her husband of twenty-three years. Her rabbi
advised Debbie to do this in order to have formal proof of her Judaism.
It eventually turned out that the twins’ biological father was Jewish, but
their biological mother was Protestant.37
Sharon had stopped going to church on Sundays because her husband
worked on those days and she did not want to go alone. But when her son
was born she wanted to “be more grounded and know how to raise him
religiously,” so she returned to the Catholic Church. However, she was
disillusioned to find that practices such as fasting had been “watered
down.” Sharon eventually chose a nondenominational Evangelical Chris-
tian Church that does not have the strict rules her Catholic church once had
but does offer the “right blend of faith, morality, forgiveness, and grace.” She
feels she should have been pleased by the Catholic Church’s more relaxed
atmosphere, but she saw the change as possibly reflecting a lack of sub-
stance or meaning. Sharon wished that her life was as organized around
146 B O R N TO G E T H E R — R E A R E D A PA RT

her faith as Debbie’s—both twins liked the rituals and formality of reli-
gious services and holidays. But Sharon was dissatisfied following “rules
that made no sense,” though she still missed “calendar things” and liked
the idea of holidays that do not fall on the same day each year, as occurs
in Judaism. Some aspects of Judaism were familiar to Sharon while she
was growing up because of her father’s Jewish clients.
Debbie and Sharon are both strong followers of their respective reli-
gions, something they acknowledge in themselves and each other. They
discuss religious issues with each other more easily than with most of
their adoptive siblings who do not have the same interest or involvement.
Each twin participates in the other twin’s religious activities when they
visit. Their different religious worlds are mutually enriching.
Soon after meeting, the twins discussed how they would feel if one con-
vinced the other to convert to the other’s religion. But that never happened,
and they say they have more respect for each other for staying with their
respective faiths. Debbie and Sharon also exemplify the MISTRA’s religi-
osity studies in their belief that each would have embraced their co-twin’s
religion had their rearing families been reversed. Sharon said, “I think
I am programmed to be religious . . . whatever religion I was exposed to
as a child would have been very important to me through my life.”
The practices of other MZA twin pairs also agreed with our findings.
Mark and Jerry, the volunteer firefighters who met by chance at age thirty-
one, had been raised by adoptive Jewish families. Neither twin showed
any interest in religion as an adult, but both consider themselves Jewish
when asked.
Two of the more interesting religious life histories concerned British
twins Elaine Alin and Mary Holmes. As Mary wrote in her 2008 book Be-
ing You, both twins were raised Catholic, but religion was a major pres-
ence in Mary’s home and a moderate presence in her sister’s. Over the
years, Mary and Elaine have tried to reconcile their belief in genetic influ-
ence and their belief in God, but these conversations with each other end
“ambiguously.” Mary eventually turned away from her faith, wondering
if this was because “it strangled my early years.” Mary also asked, “Does
Elaine’s belief in a God stem from her freedom of choice?”38 She may be
right to associate their current religious differences with the religious dif-
ferences in their rearing homes.
Mary and Elaine knew as children that their biological mother was
Jewish, information that fascinated Mary but meant nothing to her twin
at the time. In her book, Mary described a visit to Israel with her husband
Tony whose idea it was to take that trip. When Mary stood at the top of
Masada (Israel’s historic site where, in a.d. 73, the Jewish people com-
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 147

mitted suicide rather than surrender to a Roman legion) and saw the
landscape and surrounding Dead Sea, she believed she had been there
before. Perhaps her knowledge of being half-Jewish made her feel con-
nected to that historic site, reviving a sense of spirituality.
Mary and Elaine’s different outlooks never interfered with how well
they got along together, but not all twins’ religious lives synchronized
so well. Religious differences were a source of contention between MZA
female twins Betty and Alice. Betty rejected most religious beliefs, while
Alice was deeply involved in her spiritual life. These tendencies were not
apparent during the twin’s assessment in February 1983, soon after their
1982 reunion, but they developed years later. Over the years, Betty and
Alice sought counseling to hold their relationship together and are still
friendly. Their efforts along these lines underscore the significance of the
twin relationship—it is fair to assume that most people would not tolerate
companions with such opposing views but would tolerate their twin. This
situation was also exemplified by MZA twins Jack and Oskar who were
raised Jewish and Catholic, respectively (see Chapter 2). These twins al-
ternately liked and angered each other until Oskar’s death in 1997.39
Data on the twins’ religious interests and behavior continued to be gath-
ered, although another article on this topic was not published until 1999.40

Personality and Environment

By the late 1980s, behavioral geneticists and some personality researchers
were satisfied with the evidence showing that genes affected personality.
The new focus was no longer on whether or not genes affected personal-
ity traits, but instead on how genetic and environmental factors combined
or interacted to produce personality differences among people. A 1987
review in Behavioral and Brain Sciences by behavioral geneticists Robert
Plomin and Denise Daniels had summarized extant evidence showing the
limited role of shared environmental factors in personality development.41
Their two lines of evidence were (1) biological relatives reared apart and
together do not differ in their degree of personality similarity, and (2) non-
biological relatives reared together show little personality resemblance,
patterns researchers (including the MISTRA investigators) were finding
in their data. But the idea that shared family environments affect person-
ality persisted because most developmental psychologists studied intact
biological families in which genes and environments cannot be disentan-
gled. It seemed intuitively correct to explain parent-child and sibling simi-
larities according to common upbringing and shared experience. But
these conclusions did not fit the behavioral-genetic findings.
148 B O R N TO G E T H E R — R E A R E D A PA RT

Some 1980s psychologists, such as Theodore Wachs, pointed out that

twin and family studies had not paid enough attention to specific fea-
tures of the environment, such as home facilities, parental involvement,
and cultural factors that might influence personality development. Wachs
studies the role of physical and social environmental influences on be-
havior and has interests in behavioral genetics.42 He raised a legitimate
concern with which Bouchard and McGue concurred. In particular, they
noted that behavioral geneticists often relied on indirect environmental
measures (i.e., the general lack of personality resemblance of adopted in-
dividuals) as indices of shared environmental effects. Bouchard and Mc-
Gue corrected this problem in their 1990 study by examining the reared-
apart twins’ personality traits with reference to specific features of their
rearing homes.43
The reared-apart twins completed the California Psychological Inven-
tory, developed by Bouchard’s Berkeley mentor Harrison Gough. This in-
ventory measures “folk concepts” or “aspects and attributes of interper-
sonal behavior that are found in all cultures and societies, and that possess
a direct and integral relationship to all forms of social interaction.”44 The
specific form that was used included self-report true-false items orga-
nized into eighteen scales such as Dominance and Tolerance. Sample items
are “I like to be the center of attention” and “I get pretty discouraged
sometimes.” The twins also completed the Moos’ Family Environment
Scale, a form assessing ten dimensions of the rearing family, such as Con-
flict and Independence.45 This scale presents a series of true-false state-
ments about the rearing family such as “There was a feeling of together-
ness in our family” and “We fought a lot in our family.” Genetic analyses
of both scales were conducted, as were analyses of how the rearing family
affected the twins’ personalities. The participants were forty-five MZA twin
pairs and twenty-six DZA twin pairs. The DZA twin sample included
three opposite-sex twin pairs.
Both MZA and DZA twins scored lower than the inventory’s reference
sample on seven Family Environment Scale items, especially Expressiveness,
Independence, Intellectual and Cultural Orientation, and Active Recre-
ational Orientation. Both twin groups also scored higher than the refer-
ence sample on Moral-Religious Orientation, Organization, and Control.
The twins also showed greater variability in some scores than did the refer-
ence group. This seemed surprising because the twins’ scores did not differ
from those of their spouses who were not adopted.
There were two possible explanations for these sample differences.
First, the reference sample seemed to include some children, whereas the
MISTRA sample included only adults. Second, it was possible that the
high percentage of British twins (40 percent) affected the findings.
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 149

The increased variability we observed for the scores on five of the Fam-
ily Environment Scale categories (Cohesion, Expressiveness, Conflict,
Independence, and Achievement Orientation) increased our chances of
finding meaningful correlations between the personality traits and the
home environment scales.46 Restriction in the range of adoptees’ rearing
environments in some studies has been criticized because it reduces the
chance of finding associations among variables, but this would not be a
factor in ours. The fact that the twins’ scores did not differ from those of
their nonadopted spouses suggested that our twins came from a wide
range of family backgrounds.
The personality data were organized into five factors: Extraversion,
Emotional Stability, Flexibility, Consensuality, and Femininity. Gough had
reported four factors similar to our first four, but our fifth one (Feminin-
ity) had been reported in some previous studies. The Family Environment
Scale was organized into three factors, Cohesion-Conflict, Positive Con-
straint, and Encouragement of Individual Growth.
Co-twin correlations for the ten family environment scales and three
family environment factors were positive but not substantial. The MZA
correlations ranged from 0.09 (Control and Expressiveness scales) to 0.38
(Independence scale), while the DZA correlations ranged from −0.23
(Moral-Religious Orientation) to 0.42 (Conflict). It appeared that, on
average, the MZA twins’ homes were more alike than the DZA twins’
homes, but the difference was not statistically significant. Combining the
MZA and DZA data let us estimate the maximum amount of selective
placement (extent to which co-twins had been placed in similar homes).
The mean of the pooled primary scale correlations was 0.17, and the mean
of the pooled factor correlations was 0.19, showing that selective place-
ment was modest. However, the important question was whether similar-
ity in the twins’ rearing homes was associated with their personality
similarity. First, we had to see how similar their personalities were.
The mean MZA twin personality correlation was 0.45 (range: 0.10 to
0.64), and the mean DZA correlation was 0.18 (range: −0.28 to 0.53). The
intraclass correlations were modest for some MZA scales, but they were
larger for the more reliable factor scores. The average heritability was 0.49;
in fact, the heritabilities for the separate scales were significantly greater
than zero for almost every personality scale. This analysis concurred with
our 1988 personality study showing that personality is about 50 percent
heritable.
Correlations were computed between the personality and family envi-
ronment scales and factors, although only the factor correlations are dis-
played in Table 7-2. Factors have greater reliability than separate scales
because they are based on multiple outcomes.
150 B O R N TO G E T H E R — R E A R E D A PA RT

Table 7-2. Correlations between the California Psychological Inventory and Family Environment Scale
Factor Scores

Family Environment Factors

Positive Encouragement of
Personality Factors Cohesion-Conflict Constraint Individual Growth

Extraversion −0.14a −0.03 0.18a

Emotional stability 0.16a 0.07 0.10
Flexibility 0.09 −0.09 −0.07
Consensuality 0.49a 0.16a 0.19a
Femininity −0.11 −0.02 −0.04

a. P < .05.
Source: Adapted from Bouchard and McGue, “Genetic and Rearing Environmental
Influences on Adult Personality: An Analysis of Adopted Twins Reared Apart,” Journal of
Personality 58 (1990): 263–292.

Six of the fifteen correlations were significant, but only the one involv-
ing Consensuality and Cohesion-Conflict was substantial (0.49). This
meant that the more one sees the world as others see it, the less conflict
there was in the childhood home. However, these correlations must be
squared to obtain the proportion of variance in the personality measure
associated with variance in the environmental measure. Doing so indi-
cated that about 25 percent of the variance in Consensuality could be
explained by the variance in Cohesion-Conflict. The amount of explained
variance in the other five significant correlations ranged from only 2.0
percent to 3.6 percent. This indicated that, overall, the rearing home envi-
ronment has little affect on adult personality. Few significant correlations
between co-twin differences on the personality and family environment
measures emerged, demonstrating that whether or not the twins’ rearing
homes had similar or different features was unrelated to their personality
similarity as adults.47
The most striking finding from this analysis was that the personality
heritability of 45 to 50 percent agreed with those from studies of twins
reared together, both children and adults. The California Psychological
Inventory is also very different from the Multidimensional Personality
Questionnaire we analyzed in 1988, but we obtained comparable genetic
estimates across the two studies.
None of the personality analyses suggested that twins reared apart
were more alike than twins reared together, as Shields had shown in 1962.
The data are the data, but some observations can be meaningful. When I
meet twins, I often find that MZT co-twins behave more similarly when
 P SYC H O PAT H O LO GY A N D R E L I G I O S I T Y 151

they are separated, making it hard to know which twin is which. But when
twins meet me together, their tendencies toward extraversion-introversion
or dominance-submission start to show. In contrast, I rarely observed
these behavioral changes among the MZA pairs, probably because they
lacked extensive interactional histories. For example, whether I spoke in-
dividually or jointly to the MZA fireman twins Mark and Jerry, they both
seemed to take over the conversation with their loud speech and playful
style. When I was with either or both of the MZA Scottish twins Caroline
and Margaret, their gentle manner and respectful demeanor (and strong
accent) made them indistinguishable (and often incomprehensible). Per-
haps current self-report personality inventories are insensitive to subtle
changes in MZT twins’ behaviors because these inventories reflect be-
havior averaged over time. Future MZA and MZT twin studies might ask
twins to compare their personality traits to those of their co-twin, as well
as provide the usual self-report.

1990
The years leading up to 1990 brought several twin pairs to Minneapolis
for their ten-year follow-up study. This was an opportunity for me to meet
some twins who had been assessed before I had arrived in Minnesota.
I saw the Jim twins again after not having seen them since 1979, on the
set of Chicago’s WLS-TV program Friday Night. They seemed as similar
as ever. I also met the MZA male twins whose sexual orientation had
been so difficult to classify—and it still was.
Some of the new twin pairs we studied in 1990 were also memorable.
The twins I recall most clearly were elegant blonde Australian women,
Judy and Cara, who had met through mistaken identity. In the late 1980s
and 1990s, Australia had only two department store chains, David Jones
(DJs) and Myer. Judy was a fashion buyer for DJs, and Cara was a fashion
buyer for Myer, in their respective Brisbane stores. The twins found each
other when fashion representatives visiting each store had asked, “Why
are you two-timing, working for rival chains?” After this had happened
several times, the representatives realized the women were twins and ar-
ranged for their reunion. According to Nick Martin, who brought them
to our attention,
it was a beautiful story of [the] genotype channeling them toward the same
very specific niche . . . I believe that Bouchard and I met them for dinner
within a year of their being reunited and were amazed (at least I was) at
their similarity of dress, hairstyle, postures, and tastes. They wore the same
sort of jewelry, smoked the same brand of mentholated cigarettes, and even
152 B O R N TO G E T H E R — R E A R E D A PA RT

spoke with the same voice, although one had an English accent because of
her upbringing. I took them with me on a national TV program on twins,
and as a result we found about six other pairs.

Judy described the bouquet of flowers she had given to her sister in
honor of their reunion. There were white flowers mixed with flowers of the
“palest pink possible.” That shade of pink was their favorite color, even
before they met.
By the end of 1990, 107 reared-apart twin pairs had completed the
MISTRA’s assessment. And in 1990 alone, reports about the MISTRA
had appeared in twenty-six publications, including World Book Encyclo-
pedia,48 Der Spiegel magazine,49 and the Philadelphia Inquirer.50 Scientific
papers in a variety of medical and behavioral areas were being published.
We also started a new and unexpected collaboration with the periodontal
faculty.
 CHAPTER EIGHT

Dental Traits, Allergies, and

Vocational Interests (1991–1992)

This study provides new evidence of a marked genetic compo-

nent to dentate status and dental caries experience and confirms
previous reports of acknowledged inherited contributions to
tooth size, malalignment, occlusion, and morphology.
—JoAnn C. Boraas, L. B. Messer, and M. J. Till

The Dental Assessment

Pediatric dentist Michael J. Till’s career began at the Naval Air Academy
station in southwest Texas where he transported patients’ files to the den-
tal office. Till, retired from the University of Minnesota at the time of this
writing, deserves full credit for the dental component added to the
MISTRA in 1980. Till had read about the MISTRA in the campus news-
paper and saw the potential for dental research. When I met him in 2009
at the University Marriott Hotel, he recalled the “fun days” when twins
came to his clinic.
When the MISTRA dental study began, several reared-together twin
studies had reported genetic influence on number of teeth, dental caries,
tooth eruption, and dental spacing.1 Farber’s review of the early MZA
twin studies revealed a strong resemblance in the twins’ dental struc-
tures such as tooth size and shape.2 But according to Till, dentists in
the 1980s thought that dental health was most closely tied to profes-
sional treatment and personal care. Fortunately for the study, most of
the reared-apart twins predated the use of fluoride (a substance that
prevents and even reverses tooth decay) and sealants (thin plastics that
bond to the tooth’s surface and prevent erosion). In the absence of den-
tal interventions, the reared-apart twins were ideal subjects for examin-
ing genetic and environmental influences on caries and restorations.
“The patterns we saw leaned toward genetic effects,” Till said. He re-
called MZA co-twins who followed different diets but showed few den-
tal differences.
154 B O R N TO G E T H E R — R E A R E D A PA RT

The MISTRA’S first dental paper appeared in 1988, and is discussed in

this chapter along with 1991 findings from the periodontal (gum) exami-
nation we added to the twins’ assessment schedule in 1986. A 1993 den-
tal paper, an extension of the first, is also summarized here.
Dental exams were scheduled on Wednesday mornings and took about
forty-five minutes to complete. Till’s waiting room and examination cu-
bicles in the pediatric clinic were decorated with pictures of animals
and toys, intended to calm children’s fears. I am sure that this scenery
calmed some of our adult twins as well while they completed dental his-
tory questionnaires.
The research reported in the first dental article was a master’s thesis
project completed by Dr. JoAnn Boraas. The study included ninety-seven
participants from forty-four reared-apart twin pairs and three reared-
apart triplet sets. Some twins had missing data because of severe gagging,
lack of teeth, full banding (metal parts used to secure braces to the teeth),
or lack of consent, so the number of pairs varied across measures.
This exploratory study focused on dental morphology (form and struc-
ture), abnormal alignment, lesions, malocclusions,3 and restorations rather
than on rare dental conditions such as accessory cusps or shovel-shaped
incisors. Soft- and hard-tissue examinations were also conducted. The in-
terrater reliability for 35 percent of the radiograph diagnoses of cavities on
adjacent teeth was 99.7 percent. Full-mouth study models of teeth were
also made for each twin.
Most twins were each examined by two dentists, either faculty mem-
bers or dental residents who are licensed dentists. These examinations
were performed initially by Till and Dr. Louise Messer, and later by Bo-
raas and Dr. John Conry. The dental examiners did not know the twins’
zygosity, but Till said that if the twins looked a lot alike and one presented
a dental feature of interest they “might have looked harder at the other
twin” to see if there was a match.
Till was surprised by the first set of findings, and remarked, “I did not
know they would be as alike as they were.” Fifteen of the seventeen dental
variables correlated significantly for the MZA twins, whereas only three
correlated significantly for the DZA twins. Furthermore, three MZA cor-
relations significantly exceeded the corresponding DZA twin correlations.
The distribution of the variables was unaffected by age and sex, although
age was the best predictor of number of teeth present.
The MZA twins showed significant resemblance in many dental mea-
sures despite any differences in diet and treatment. The twins’ mean age
of 40.6 years would have allowed these rearing factors to show maximal
difference effects, but they did not. Genetic influence on measures such as
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 155

salivary factors, oral flora, tooth eruption time, and tooth eruption se-
quence may have explained the MZA twins’ similar history of cavities.
No explanation for the lack of genetic influence on overbite and overjet
was available; in fact, this finding conflicted with previous twin study
findings. Till thought this finding was the biggest surprise.4
Dental terms are exotic, such as “Carabelli’s trait” and “premolar
grooves.” Carabelli’s trait is an extra cusp or pointed structure on a first
molar; premolar grooves are flat elevations on the premolar teeth. Both
were examined for twin resemblance in antimeres, homologues, and het-
erologues. Antimere teeth are the matching teeth appearing on a person’s
left and right sides of the mouth (e.g., two first permanent molars, one on
the left and one on the right). Homologue teeth are teeth with similar
positions, structures, functions, or characteristics (e.g., incisors). Hetero-
logue teeth are teeth with different structures, positions, functions, or
characteristics (e.g., molars differ from incisors in structure, position,
and function).
Resemblance in antimere symmetry concordance (resemblance) was the
same for MZA and DZA twin pairs for both Carabelli’s trait and premolar
grooves, suggesting an absence of genetic influence. However, MZA ho-
mologous concordance was approximately twice that of DZA concor-
dance for these two traits, this time indicating the contribution of genetic
factors. Heterologous concordance was also higher for MZA than DZA
twins, indicating the strength of environmental influences. Lack of mirror
imaging (trait reversal) was indicated by the similar values for homolo-
gous and heterologous concordance within the two twin groups. Reared-
together twin studies have also failed to find evidence of mirror-imaging
in these dental measures.
Just for fun, our dental colleagues put one MZA twin’s upper jaw (max-
illary) plaster model over the co-twin’s lower jaw (mandibular) model to
check the fit. It worked almost perfectly, showing that the twins’ teeth had
evolved and rotated in similar ways. Till said that the DZA twins’ models
were “all over the place.”
Boraas’s 1988 paper was the first to address dental findings in reared-
apart twins. She confirmed what so many parents had told me about their
identical twin children’s matching teeth. Graduate dental programs at
Minnesota and elsewhere referred new students to the MISTRA studies,
and they still do. Boraas’s work earned her the 1988 American Academy
of Pediatric Dentistry Graduate Student Research Prize.
In a larger updated analysis including forty-six MZA twin pairs and
twenty-two DZA twin pairs, every measure showed greater MZA than
DZA twin resemblance. However, the Surfaces Restored Index was the
Table 8-1. Initial and Updated Comparisons of Dental Parameters in MZA and DZA Twin Pairs

MZA DZA

Parameters n (pairs) ri n (pairs) ri

Dentate status
Teeth presenta 28 0.62d 10 −0.19
46 0.45c 22 0.04
Teeth presenta (excluding molars) 28 0.69d 10 0.06
46 0.49c 22 0.11
Treatment status
Teeth restored 28 0.66d 10 0.09
46 0.57c 22 0.30
Teeth restored index 28 0.65d 10 0.51b
46 0.61c 22 0.31
Teeth restored index (adjusted) 28 0.74d 10 0.47
Surfaces restoreda 28 0.71d 10 −0.04
46 0.46c 22 0.20
Surfaces restored index 28 0.72d 10 0.33
46 0.67 22 0.17
Surfaces restored index (adjusted) 28 0.77d 10 0.46
Treatment/caries status
Surfaces restored or carious index 28 0.79d 10 0.34
(adjusted)
44 0.58c 22 0.26
Tooth size
Maxillary incisors 13 0.94d 2 0.06
Mandibular incisors 19 0.63d 6 0.33
Malalignment
Rotation/displacement index 26 0.43c 7 −0.27
Rotation/displacement index 26 0.43c 7 −0.09
(adjusted)
Occlusion variables
Overbite 23 0.03 6 −0.15
Overjet 23 0.06 5 0.31
Arch width
canine-canine 24 0.81d 4 0.92c
molar-molar 14 0.69d 4 0.76b

a. MZA > DZAs.

b. P < .05.
c. P < .01.
d. P < .001.
Source: Adapted from Boraas et al., “A Genetic Contribution to Dental Caries,
Occlusion, and Morphology as Demonstrated by Twins Reared Apart,” Journal of Dental
Research 67 (1988): 1150–1155; and Conry et al., “Dental Caries and Treatment
Characteristics in Human Twins Reared Apart,” Archives of Oral Biology 38 (1993):
937–943.
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 157

only measure in which the MZA and DZA twins differed significantly,
possibly because the larger sample was still modest in size. Every trait
correlation was significant for the MZA twins, while none were signifi-
cant for the DZA twins. The initial and updated analyses are summarized
in Table 8-1.
The updated dental trait heritabilities ranged between 0.52 and 0.82.
The Surfaces Restored Index’s heritability equal to 1.00 possibly re-
flected sampling error.5 Interestingly, all the MZA correlations fell some-
what since the 1988 analysis, and most of the DZA correlations in-
creased. It is possible that the MZA correlations were overestimated and
the DZA correlations were underestimated in the previous analysis that
included smaller numbers of twins. But the big picture did not change—
this second study underscored the importance of examining older twins
for whom dental problems were more likely present, allowing for more
sensitive tests of genetic effects.
Conry, who completed the updated analysis, speculated that the genetic
transmission of oral structure may be key to the MZA twins’ similarity
in caries experience.6 People with shorter upper jaw (maxillary) arches
are at greater risk for cavities. The reared-apart twins were not examined
on a site-by-site (location) or quadrant-by-quadrant (upper or lower right
or left area) basis, which would have helped identify the role of tooth
morphology in dental decay.
The twins’ behaviors might also illuminate the genetics of dental dis-
ease. Genetically based intellectual factors related to understanding health-
related issues, and personality factors related to vigilance in dental care
could be important. Unrelated individuals might achieve different suc-
cess levels in preventing dental disease despite following similar dental
regimens.
Meanwhile, in the mid-1980s we added measures of periodontal disease
to the study. Periodontitis is a common cause of adult tooth loss, and
about three-fourths of Americans experience some periodontal symptoms.7
We began bringing twins to Moos Tower’s seventh floor immediately after
their dental examinations.

The Periodontal Clinic

In August 1986, periodontal graduate student Bryan Michalowicz was
about to go on vacation when he got a call from his advisor, Dr. Bruce
Pihlstrom. Pihlstrom said that they needed to submit a research proposal
to the National Institute of Dental Research8 for studying periodontal
traits in reared-apart twins. It rarely happens, but the proposal (an RO1
158 B O R N TO G E T H E R — R E A R E D A PA RT

grant, which supports specific projects based on investigators’ interests

and competencies) was funded upon the first review.
Michalowicz hadn’t heard of the MISTRA until Pihlstrom’s call, but
it came at a good time, just as he was searching for a thesis project. Few
periodontal studies of reared-together twins had been done by 1986. A
1958 German study of six- to twenty-nine-year-old twins had reported
genetic influence on calculus (hardened dental plaque or tartar).9 How-
ever, a 1969 study of twelve- to seventeen-year-old twins failed to find
genetic influence on several periodontal measures, including calculus.10 A
problem with these studies was their use of relatively young twins who
may not have passed through the risk period for periodontal disease.
Once Pihlstrom and Michalowicz had decided to study adult twins reared
apart, they recruited adult twins reared together via David Lykken’s twin
registry. This work would be a first.
When the study began every reared-apart twin was examined individu-
ally by the same two periodontists. After about the sixth set, Michalowicz
told Pihlstrom that because some measures, such as inflammation, were
judged subjectively, the possibility of bias from impression of twin type
was a concern. That is, suspecting that twins were MZA (or DZA) could
cause examiners to unintentionally reduce (or widen) co-twins’ recorded
differences. (This was not a problem for evaluating radiographs or
plaque composition, which can be done blindly.) According to Michalo-
wicz, Pihlstrom was a “meticulous” investigator and took a preemptive
approach to potential criticism. His solution for avoiding bias was to
cover the first twin using bibs stapled together to create a mask. “This
was always a fun thing because you could hear a voice, but not see a
face.” Once Michalowicz was trained and interrater reliability was high,
each twin was seen by either Michalowicz, Pihlstrom, or their colleague
Jim Hinrichs, so masking was discontinued.
The first periodontal article appeared in 1991 in the Journal of Peri-
odontology.11 The sample included sixty-three MZT twin pairs, thirty-
three same-sex DZT twin pairs, and fourteen MZA twin pairs. Attach-
ment loss, probing depth, and gingival and plaque indices were obtained
from the surfaces12 of a representative set of six teeth, known as the
Ramfjord teeth. Sigurd Ramfjord, a Norwegian investigator and Pihl-
strom’s mentor at the University of Michigan, had devised this system
which reduced assessment time from forty-five minutes to ten. The six
teeth reflect the mean periodontal status of the individual. Twins with
fewer than five Ramfjord teeth present were omitted from the study.
The genetic analyses of attachment loss were performed on mouth
means. Dorothy Aeppli, a co-investigator and biostatistician in the Biom-
etry Division of the University of Minnesota’s School of Public Health,
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 159

used bootstrap sampling (the drawing of multiple samples from a popula-

tion to estimate the replicability of the results)13 to derive confidence in-
tervals (estimates of the range of values of an unknown population mea-
sure) for these values because of their sensitivity to non-normality. Seven
hundred samples were drawn from the MZT and DZT groups and com-
bined randomly. The correlations and heritabilities were then computed
for each sample. The same procedure was used for the MZA twin pairs.
Pihlstrom recalled, “I have very vivid memories of our first analysis. I
almost fell off my chair; I could not believe it. Bryan [Michalowicz] was
with me. The heritabilities were about 50 percent. I was very happy that
we could make a contribution to the field. Our paper changed the way
people thought about periodontal disease.” The findings, presented in
Table 8-2, were later confirmed by a Virginia Commonwealth University
reared-together twin study in which Michalowicz collaborated.14 Both
the MZA and MZT correlations were consistently higher than the DZT
correlations, demonstrating genetic effects.
The heritabilities for attachment loss, probing depth, gingivitis, and
plaque ranged between 0.38 and 0.82 (twins reared apart) and 0.38 and
0.51 (twins reared together). These findings explained the inconsistencies
among extant genetic studies of periodontitis. That is, previous work had
focused on the microbiota (microscopic organisms in a particular envi-
ronment) of plaque, whereas the MISTRA results highlighted the effects
of hosts’ (i.e., twins’) genetic influences on disease. The MZT and MZA
twins’ oral bacteria were not more alike than that of the DZ twins, but
the extent and severity of their periodontal symptoms (response to the bac-
teria) were more alike. These findings brought unexpected media atten-
tion to our periodontal team.
The second periodontal paper appeared later in 1991.15 The study com-
bined sixty-two MZT, twenty-five DZT, and thirty-three MZA twin pairs
in an analysis of alveolar bone height. Alveolar bones are ridges contain-
ing sockets that are located on the bones that hold the teeth. Genetic
effects on this measure had never been examined.
A device holding each twin’s head in place was used to obtain panoramic
radiographs.16 Bootstrap sampling was used to estimate the between-
pair and within-pair variances, intraclass correlations, heritabilities, and
confidence intervals as in the previous periodontal analysis. These values
are summarized in Table 8-3 (the actual values of these statistics were not
provided in the article). The study showed that alveolar bone height has
a significant genetic component.
An important point from this study was that the genetic mechanisms
associated with alveolar bone height might be tied to periodontal pa-
thology (bone loss) or to anatomical development (bone thickness). This
Table 8-2. Twins’ Mean Values, Variances and Intraclass Correlations, and Heritability Estimates for the Periodontal Measures

Parameter n (pairs) Xˉ (SD)a ri (90% CI)b H Together c H Apartd

Attachment loss (mm) 0.48 (0.21–0.71) 0.38 (0.01–0.79)

MZT 43 −1.0 (0.57) 0.49 (0.28–0.70)
DZT 20 −0.9 (0.47) 0.08 (0.00–0.30)
MZA 12 −0.5 (0.57) 0.38 (0.01–0.79)
Probing depth (mm) 0.51 (0.12–0.68) 0.69 (0.48–0.86)
MZT 63 2.9 (0.34) 0.54(0.34–0.68)
DZT 33 3.0 (0.33) 0.16 (0.00–0.40)
MZA 14 3.0 (0.50) 0.69 (0.48–0.86)
Gingival index 0.40 (−0.17–0.66) 0.82 (0.69–0.89)
MZT 63 1.1 (0.26) 0.58 (0.44–0.67)
DZT 33 1.2 (0.29) 0.37 (0.05–0.62)
MZA 14 1.4 (0.54) 0.82 (0.69–0.89)
Plaque index 0.38 (0.02–0.56) 0.46 (0.13–0.69)
MZT 63 −0.1 (0.39) 0.39 (0.18–0.57)
DZT 33 −0.1 (0.37) 0.06 (0.00–0.30)
MZA 14 0.3 (0.43) 0.46 (0.13–0.46)

Note: H = Heritability.
a. MZT and DZT twins did not show mean differences on any of the four measures (P < .10).
b. The ri and confidence intervals (CI) were obtained from the bootstrap distribution (ri = median).
c. Heritability (2 − [riMZ − riDZ]) was estimated by the median of the bootstrap samples.
d. Heritability (ri MZA) was estimated by the median of the bootstrap samples.
Source: Adapted from Michalowicz et al., “Periodontal Findings in Adult Twins,” Journal of Periodontology 62 (1991): 293–299.
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 161

Table 8-3. Intraclass Correlations, Heritabilities, and Confidence Intervals for Alveolar Bone Height of
MZT, DZT, and MZA Twin Pairs
a
n ria 90% CIb h2 90% CI of h2

MZT 62 0.70* 0.53–0.80 0.36 −0.15–0.77

DZT 25 0.52* 0.25–0.70
MZA 33 0.55* 0.35–0.74 0.55 0.35–0.74

a. Median values of the bootstrap samples.

b. 5th and 95th percentiles of the bootstrap distribution, P < .05. CI: confidence
interval.
Source: Adapted from Michalowicz et al., “A Twin Study of Genetic Variation in
Proportional Radiographic Alveolar Bone Height,” Journal of Dental Research 70 (1991):
1431–1435.

conclusion makes sense in relation to previous findings of genetic influ-

ence on skeletal dimensions17 and on periodontal pathology, as shown
in the initial 1991 article. This second periodontal article had less of an
impact on the field than the first, but it was important work in another
sense—if the alveolar bone height had not shown significant genetic in-
fluence, it would have detracted from the initial study.
Findings from the first 1991 MISTRA article also led several groups
to study genetic susceptibility to cytokine (immunological agent) produc-
tion. The now-editor of Peridontology, Ken Korman, referenced the study
as a basis for developing cohorts to investigate host susceptibility. Accord-
ing to Michalowicz, a seminal 1997 article of Korman’s linking a specific
gene to inflammation was inspired by the MISTRA research.18 As a result,
Pihlstrom and Michalowicz were asked to author a genetics chapter for
the most prestigious periodontal textbook Carranza’s Clinical Periodon-
tology.19 Pihlstrom also was asked to present the twin study data at a
symposium, “Genetics of Periodontal Disease,” at the 2001 meeting of the
International Association of Dental Research, in Chiba, Japan.
Michalowicz misses the scientific excitement and interactions with the
twins now that the study has ended—this was a common theme among
the MISTRA colleagues. Therefore, I was surprised that Pihlstrom felt
differently. He explained that the team went as far as it could go—posing
questions, providing answers, and getting confirmation from other inves-
tigators. “While one always misses the excitement of scientific discovery
it’s rare that a research project has a definite beginning and end like this
one. In that sense, I’m glad that we were able to bring it to a definite con-
clusion,” he said. The Minnesota periodontal group would publish sev-
eral more articles in 1999 and 2000 and would win additional awards.
In  fact, the first 1991 article was cited under the section “Advances in
162 B O R N TO G E T H E R — R E A R E D A PA RT

Pathogenesis of Periodontal Disease” in a special 2009 supplement to the

Journal of the American Dental Association, “150 Years of Progress.”20
Like his dental counterpart Boraas, Michalowicz won the American
Academy of Periodontology’s 1987 national competition for outstand-
ing graduate student research. He received the Balint Orban Award for
that work, which by then included thirty reared-apart twin pairs. “I was
dumbstruck—some real giants in the field said that we quantitated find-
ings that they had suspected for years,” Michalowicz recalled.
Regardless of the accomplishments, there were also regrets. Both
Michalowicz and Pihlstrom acknowledged their inability to assay the
plaque without destroying the sample for other analyses. Having serum
would have allowed them to look at the twins’ reactions to antibodies,
testing their theory that response to bacteria is genetically mediated.
This, and identifying genes linked to periodontitis, would have been the
next logical steps in their research program. But this would have required
the expertise of genetic epidemiologists interested in periodontal diseases,
and few people are qualified to take on such tasks. The search for genes
tied to life-threatening conditions like cancer and diabetes are given greater
priority. One can live with periodontitis. Nevertheless, one of the benefits
of revisiting the MISTRA investigators has been rekindling their excite-
ment in reared-apart twin data that have yet to be analyzed. Some planned
analyses in periodontology and other areas will be described in the final
chapter.
Pihlstrom may stay involved in future analyses of the reared-apart twins,
but at a distance. He relocated to the National Institute for Dental and
Craniofacial Research, in Bethesda, Maryland, in 2002, and was acting
director of the Extramural Division of Clinical Research, a position he
held until 2007. He is now an independent oral health research consul-
tant. One of his missions is promoting research activities among young
scientists, and he encourages them to apply for RO1s.

Allergies and Antibodies

Allergies are common medical conditions involving abnormal reactions
to otherwise harmless substances, such as lobster and house dust.21 Blu-
menthal had always been interested in genetic and environmental factors
affecting atopic disease (allergy predisposition). Before the MISTRA got
started, he had discussed twin research with Lykken, although they had
never collaborated. Understandably, he saw the MISTRA as a unique op-
portunity to explore factors affecting asthma, rhinitis (nasal inflammation),
skin-test response, total serum levels of immunoglobulin E (IgE, proteins
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 163

that include antibodies elicited by allergens), and specific levels of IgE

as measured by the radioallergosorbent test (RAST, an allergy test done
using a blood sample).
Previous reared-apart twin studies did not systematically study response
to allergens, but allergic reactions had been observed among twenty-nine
of the early MZA twin pairs.22 The conditions included eczema, urticaria
(raised itchy skin that usually indicates an allergy), heat spots (eruption of
small vesicles in warm weather), asthma, hay fever, bronchitis (inflamma-
tion of the main air passages to the lungs), nasal polyps, sinusitis, frequent
colds, and unspecified allergic reactions. Approximately half the cases were
concordant, including six of eight pairs with two or more allergies, suggest-
ing genetic effects.23 Reared-together twin and family studies had also indi-
cated genetic effects on allergic conditions.24
In 1985, the MISTRA had reported genetic effects on IgE, as dis-
cussed in Chapter 4. That work included twenty-six MZA and ten
DZA twin pairs.25 Six years later, Blumenthal had access to fifty-three
MZA twin pairs (twenty-three male and thirty female) and twenty-one
DZA twin pairs (five male and sixteen female).26 The analyses were done
in combination with data from thirty-four MZT and twenty-nine DZT
twin pairs recruited from the Minnesota Twin Registry, and seventy-six
MZT and eighty-two DZT twin pairs from the Finnish National Public
Health Institute. The mean ages of the reared-apart Minnesota twins and
reared-together Finnish twins were similar (MZA: 38.91 ± 12.89, DZA:
43 ± 10.22, MZT: 35 ± 13.00, DZT: 38 ± 13.00), but the reared-together
Minnesota twins were younger (MZT: 29 ± 11.00, DZT: 30 ± 13.00). The
prevalence rates for asthma and seasonal rhinitis in both Minnesota twin
groups were similar to those reported in population studies.27
Data sources included the twins’ medical examinations, blood samples,
and skin testing. The IgE levels of most twins were determined by double-
antibody radioimmunoassay, a sensitive test for measuring antigens in the
blood, although the paper radioimmunosorbent test (PRIST) was used
to measure total IgE response for twins in the Finnish sample; the two
methods are comparable, so the data were pooled. Specific serum IgE lev-
els were measured by RAST. Concordance rates for seasonal rhinitis and
asthma for reared-apart and reared-together twins from the Minnesota
studies are presented in Table 8-4.
The MZA-DZA concordance difference for asthma was the only value to
reach statistical significance. Had the samples been larger, they might have
provided the power for detecting other meaningful differences. The values
were also lower than those cited in Blumenthal and Bonini’s 1990 review,28
possibly due to methodological differences. The intraclass correlations
164 B O R N TO G E T H E R — R E A R E D A PA RT

were statistically significant for all twin groups, with the exception of the
DZT twins from Minnesota, as shown in Table 8-5.29
Biometrical analyses of the serum IgE data did not show differences
between reared-together twins from Minnesota and Finland, so the data
were combined. The genetic effect was estimated as 0.564, and the non-

Table 8-4. Probandwise Concordance Rates for Seasonal Rhinitis and Asthma for the Minnesota Twin
Samples

Seasonal Rhinitis

MZA DZA MZT DZT

Number concordant 9 2 8 3
Number discordant 12 5 10 8
Percentage concordant 60 44 62 43

Asthma

MZA DZA MZT DZT

Number concordant 4 0 2 0
Number discordant 1 2 1 2
Percentage concordant 89a 0 80 0

Note: Concordance was only studied in pairs with one or more affected twins.
a. P < .05 concordance.
Source: Adapted from Hanson et al., “Atopic Disease and Immunoglobulin E in Twins
Reared Apart and Together,” American Journal of Human Genetics 48 (1991): 873–879.

Table 8-5. Intraclass Correlations for Immunoglobulin E Level

Minnesota
n (pairs) ri 95% CI

MZA (49) 0.640a 0.442–0.779

DZA (21) 0.486a 0.087–0.751
MZT (34) 0.422a 0.106–0.661
DZT (27) 0.258 −0.123–0.573

Finland

MZT (76) 0.560a 0.384–0.696

DZT (82) 0.365a 0.152–0.530

a. P < .05.
Source: Adapted from Hanson et al., “Atopic Disease and Immunoglobulin E in Twins
Reared Apart and Together,” American Journal of Human Genetics 48 (1991): 873–879.
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 165

shared environmental effect was estimated as 0.436. A shared environ-

mental effect was not indicated because the twins reared apart were more
alike than the twins reared together. In addition, a trend toward greater
skin-test and RAST concordance in MZA (55 percent, 50 percent) and
MZT (70 percent, 50 percent) twin pairs than DZA (50 percent, 0 percent)
and DZT (28 percent, 33 percent) twin pairs was observed. The small
sample sizes possibly prevented the differences from reaching statistical
significance, but it may be that some antigen sensitivities are influenced
by environmental factors.
The first study of IgE levels in twins reared apart showed a genetic ef-
fect for total serum IgE level with little effect from the shared family envi-
ronment. This was evidenced, in part, by the similar total IgE levels for the
MZA and MZT twins. However, examining the RAST and skin-test data
by specific antigen did not indicate genetic effects. More might have been
said about genetic effects on these antigens in particular, and on allergy
predisposition in general, had the twins been selected for atopic disease.
Such a sample would have shown a greater variety of symptoms and
symptom severity.

Vocational Interests
There were many medical “firsts” from the MISTRA, as shown by the
studies in dentistry, periodontology, and immunology. There were also
psychological “firsts” occurring simultaneously. One was our study of
vocational interests.
Psychology graduate student Dan Moloney worked on the MISTRA
between 1985 and 1988, administering tests and processing data. He
came to the University of Minnesota because Bouchard “gave a great talk
on the twins” at St. John’s University, and Moloney was captivated. He
often went to the airport to pick up the new pairs and recalled how easily
and naturally the reunited twins got along.
An important question was whether the heritability of vocational in-
terests would match the 50 percent we had found for personality traits.
Moloney predicted that the heritability of vocational interests would be
lower—and he thought that family influences would play a part. His big-
gest contribution to the MISTRA was his analysis of the vocational inter-
est data, work that earned him his PhD degree.30
Moloney’s study built upon Bouchard’s preliminary 1986 analysis of
vocational interests. Bouchard had shown genetic influence on the Strong-
Campbell’s six General Occupational Themes. Five years later, data were
available for fifty-two MZA twin pairs and twenty-seven DZA twin
166 B O R N TO G E T H E R — R E A R E D A PA RT

pairs. The larger sample also allowed us to look at whether family rear-
ing factors and home facilities (e.g., artwork or gardening equipment)
affected vocational interests. This would be the first vocational interest
study to use twins reared apart and one of the few to use adult twins. Pre-
vious studies had included mostly younger subjects, which was problem-
atic since vocational interests do not stabilize until about age eighteen.31
Moloney’s would also be the first such study to apply biometrical model-
ing, and to examine the effects of family factors on vocational interests.
In keeping with the MISTRA’s goal of within-study replication, voca-
tional interests were assessed by two measures: the Strong Vocational
Interest Blank–Strong-Campbell Interest Inventory (SVIB-SCII) and the
Jackson Vocational Interest Survey.32 In the first section of the Strong,
twins indicated “Like,” “Indifferent,” or “Dislike” for 281 items concern-
ing occupations, school subjects, activities, leisure activities, and types of
people. In the second section, twins indicated their preference for one ac-
tivity over another in each of 30 pairs. In the third section, twins responded
“Yes,” “No,” or “Undecided” to 14 personality items that could be self-
descriptive. Their answers yielded scores on 6 General Occupational
Themes, 23 Basic Interests Scales, and 207 Occupational Scales. Moloney
analyzed only the Basic Interests and General Occupations in his article.
The Jackson survey included 289 paired-comparison statements con-
cerning work activities. Twins chose the more personally interesting ac-
tivity in each pair, providing answers that yielded 34 scales with 17 items
each. The other two measures included in the study were the Family En-
vironment Scale and Physical Facilities Checklist, described in Chapters
5 and 7.
The vocational interest data went through numerous statistical ma-
nipulations. Three factor analyses were performed to see if the variables
formed a smaller number of independent and coherent subsets. The first
factor analysis was applied to the combined Basic Interest Scales and
Jackson scores; the second was applied to the ten Family Environment
scales; and the third was applied to the forty-one Checklist items.
Mean scores on the General Occupations and Basic Interests Scales
were similar for MZA and DZA twin pairs and slightly lower than those
reported in the inventory manual. Mean scores on the Jackson were also
similar for the two types of twins and similar to the Jackson norms. The
only exception was the significantly lower Adventure scale score for the
DZA twins. The Adventure scale taps people’s enjoyment of novel situa-
tions and the seeking out of unusual or dangerous situations. This differ-
ence may have reflected the twins’ older age, relative to the Jackson
sample that included high school and college students.
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 167

The interest scales were organized into ten factors: Enterprising, Aca-
demic Orientation, Artistic, Investigative, Work Style, Realistic, Social,
Adventure, Medical, and Conventional. All six occupational scales were
included, as were others identified in earlier work. The Family Environ-
ment Scale yielded three factors: Cohesion versus Conflict, Positive Re-
straint, and Encouragement of Individual Growth. This was the same so-
lution reported by Bouchard and McGue in their 1990 personality study,
based on a slightly smaller sample, described in Chapter 7.33 Moloney’s
analysis also reduced the home facilities to four factors—Material Posses-
sions, Cultural, Scientific/Technical, and Mechanical/Outdoor—the same
ones obtained in the information processing study.
The series of analyses indicated considerable genetic influence (ap-
proximately 40–50 percent) on vocational interests. The mean MZA
intraclass correlations were 0.38 (General Occupations) and 0.39 (Basic
Interests), with scales ranging from 0.12 (Sales) to 0.57 (Science). The
mean DZA intraclass correlations were 0.01 (General Occupations) and
0.05 (Basic Interests), with scales ranging from −0.27 (Domestic Arts) to
0.33 (Merchandising). The mean heritabilities for Occupations and Basic
Interests were 0.35 and 0.37, respectively. In most cases, a general model
(one that included genetic and environmental effects) fit the data structure
better than a model that included only environmental effects. The general
model was rejected only for Domestic Arts and Office Practices due to
their large negative correlations.
The Jackson showed a pattern similar to that of the Strong, increasing
our confidence in the findings. The mean MZA intraclass correlation
was 0.43, with scales ranging from 0.20 (Independence) to 0.69 (Creative
Arts). The mean DZA intraclass correlation was 0.11, with scales ranging
from −55 (Physical Science) to 0.67 (Professional Advising). The mean
estimated heritability for the Jackson was 0.44. The general model did not
fit the Physical Science and Professional Advising scales because the DZA
correlations were not equal to half the MZA correlations. The same sets
of analyses were performed on the ten interest factors.
The mean MZA and DZA intraclass correlations for vocational interests
were 0.50 and 0.07, respectively, and the mean heritability was 0.50. The
general model failed only for Academic Orientation because of the large
correlational difference between the MZA and DZA twin pairs.34 Overall,
the analysis showed that occupational interests were substantially guided
by genetic potentials. But it was our final set of analyses that were most
provocative and corroborated the genetic findings, shown in Table 8-6.
We found little evidence that rearing impacted vocational interests. Only
three correlations between the three family factors (Cohesion-Conflict,
168 B O R N TO G E T H E R — R E A R E D A PA RT

Table 8-6. Intraclass Correlations, Heritabilities, and Genetic Tests for the Ten Vocational Interest
Factors

χ 2 for test of:

No Genetic
MZA DZA Effect General Model
Scale ri ri Heritability ± SE (3 df) (2 df)

Enterprising 0.41 0.30 0.50 ± 0.12 10.04 2.21

Academic 0.73 0.19 0.82 ± 0.06 17.04 6.58
orientation
Artistic 0.52 −0.07 0.50 ± 0.11 13.03 2.12
Investigative 0.68 0.02 0.66 ± 0.09 18.00 1.92
Work style 0.25 −0.03 0.22 ± 0.14 2.76 0.48
Realistic 0.44 −0.09 0.41 ± 0.12 9.85 1.90
Social 0.54 0.02 0.52 ± 0.11 12.50 1.27
Adventure 0.54 0.04 0.53 ± 0.13 10.44 2.43
Medical 0.47 0.14 0.49 ± 0.14 9.10 2.19
Conventional 0.38 0.17 0.38 ± 0.13 6.28 0.29

Note: No Genetic Effect Model: χ2 > 7.82 is significant at .05; General Model: χ2 > 5.99
is significant at .05. Statistically significant χ2 values are shown in bold; statistically
nonsignificant χ2 values that differ significantly are in italic. df: degrees of freedom; SE:
standard error of the mean.
Source: Adapted from Moloney et al., “A Genetic and Environmental Analysis of the
Vocational Interests of Monozygotic and Dizygotic Twins Reared Apart,” Journal of
Vocational Psychology 39 (1991): 76–109.

Positive Constraint, and Encouragement of Individual Growth) and ten

interest factors were statistically significant. The strongest relationship was
between Artistic Interests and Encouragement of Individual Growth—but
only 5 percent of the individual differences in the former were explained
by the latter! Moreover, the direction of the relationship was implausi-
ble—it was unlikely that greater Encouragement of Individual Growth
reduced Artistic Interests.
The same theme of negligible family influence on vocational interests
emerged when we looked at links between the home facilities factors (Ma-
terial Possessions, Cultural, Scientific/Technical, and Mechanical/Outdoor)
and interest factors. Few significant correlations emerged, and in only
one case did a facility factor (Cultural) explain more than 10 percent of
the individual differences in an interest factor (Artistic). Furthermore,
prestudy contact between the twins had little effect on their vocational
interest resemblance—correlations between the ten interest factors and
contact measures ranged between −0.03 (total contact time) and 0.11
(time together before separation).
 D E N TA L T R A I T S , A L L E RG I E S , A N D V O C AT I O N A L I N T E R E ST S 169

Moloney remembers staring at the computer with Bouchard, studying

the statistics that reflected the negligible family effects on vocational inter-
ests. “We went over and over it, but our analyses were correct. Bouchard
finally said that we needed to let the data tell us what was there—that
was the idea behind the MISTRA.”
Moloney’s study had both theoretical and practical implications. Con-
sistent with previous studies, genetic influence was found to substantially
shape vocational interests. Studies showing vocational interest resem-
blance between parents and children from intact biological families could
not truly claim that children’s interests were based on what their parents
did or how they raised them. Of course, family environments may change
over time with consequences for children’s behaviors, and such effects
have been analyzed using intact biological families.35
We did find that the DZA twins were less alike in vocational interests
than the DZT twins described in the twin literature. This finding recalled
Tellegen’s suggestion that common rearing may induce behavioral simi-
larity between siblings when they are not genetically identical. The na-
ture of social relationships between DZ twins, in general, have been ne-
glected relative to those of MZ twins. But if we are to understand how
genes and environments work together, it will be necessary to give DZT
and DZA twin relationships higher research priority.
Our analysis left about 50 percent of the variance in vocational inter-
ests unexplained by genes and shared environments, so we looked to-
ward nonshared environmental factors. This area is murky, and discover-
ing what these factors are has been a goal of behavioral geneticists for
some time. In 2000, University of Virginia psychologist Eric Turkheimer
discussed this “gloomy prospect,” underlining the difficulty of finding trait-
relevant nonshared environmental events because unique experiences af-
fect behavior in unsystematic ways.36 The guidance of parents, teachers,
and other mentors can significantly affect the career paths that children
eventually choose. Reading a great book or going to the theater can also
inspire interest that may turn into a profession. How these influences oper-
ate, and why they vary across individuals poses a formidable challenge in
behavioral genetic investigations.

1991 to 1992
The years 1991–1992 added eleven new reared-apart twin pairs to the
study (two MZA and nine DZA) for a total of 118 pairs. We also con-
ducted follow-up studies, bringing back seven pairs who had been as-
sessed previously. I was thrilled to meet Barbara and Judy, the twins I
170 B O R N TO G E T H E R — R E A R E D A PA RT

reunited as a graduate student who had visited Minnesota two years be-
fore I arrived. They showed the same closeness and familiarity of sisters
who had grown up together. It was just as Bouchard and DiLeonardi had
observed at their first meeting at the Minneapolis airport in 1980.
In September 1991, I accepted a faculty position at California State
University at Fullerton. I had been with the MISTRA for nine years, the
first three as a postdoctoral fellow and the last six as a research associate.
Bouchard had stepped down as department chair, freeing him to run the
study more directly. It was time for me to establish an academic niche of
my own. The thought of leaving the MISTRA saddened me, although I had
no time to dwell on such feelings then. Knowing I would soon be gone,
Bouchard had me schedule four twin pairs back to back during my last
four weeks in Minneapolis. We had never assessed so many twins spaced
so closely together. It was an exhausting but fitting farewell.
 CHAPTER NINE

Creativity, Work Values, and Evolution (1992–1993)

D avid T. Lykken was brilliant and funny, frank and contro-

versial. He enjoyed hearing other people’s ideas and had many
of his own, some hotly contested. He deprecated lie detection1
and advocated parental licensure, the issuing of permits to couples who
met government standards for reproducing.2 Lykken was also a skilled
writer and storyteller. He supplemented his scientific concepts and conclu-
sions with real-life examples, many drawn from his experiences with the
reared-apart twins.3
Lykken was also a clever methodologist. He showed how statistical
power increased by 20 percent by studying twins reared apart versus
twins reared together.4 In twin research, he was famous for introducing
emergenesis, a concept he detailed in a 1992 article in American Psycholo-
gist that remains controversial to this day.5

Emergenesis: Lykken’s Legacy

Emergenesis refers to genetically influenced traits that do not run in
families. Emergenic traits are thought to emerge out of complex configu-
rations of polymorphic genes that come together by chance in an indi-
vidual.6 According to Lykken, emergenesis explained extraordinary
musical talent, mathematical ability, and athletic prowess in families in
which musicians, mathematicians, and athletes were absent or ordinary.
Individuals born in later generations would lack these talents because
genetic segregation would dissolve the gene combinations underlying the
unusual trait. Lykken reasoned that such processes might explain Triple
172 B O R N TO G E T H E R — R E A R E D A PA RT

Crown champion Secretariat’s running speed and Indian mathematician

Srinivasa Ramanujan’s numerical genius. Secretariat’s speed surpassed
that of his distinguished lineage, and Ramanujan owned just two math
books, both in a foreign language. The skills of such horses and people
may differ qualitatively, not just quantitatively, from those of others.
Emergenic traits include any complex behavior that appears rarely
in families, not just unusual talents or skills. The unique behaviors of
several of our MZA twin pairs are illustrative of emergenic traits. There
were female co-twins who giggled incessantly, female co-twins with af-
finities for dogs and horses, and male co-twins with outstanding mechani-
cal skills; these male twins each separately diagnosed a faulty wheel bear-
ing on Bouchard’s car. Few DZA twins showed such similarities. Lykken
noted that these observations were only “suggestive in the ‘context of
discovery’ ”7—as I indicated, the important information was the marked
MZA-DZA difference in the size of the intraclass correlation for a given
behavior. That is, a high MZA (or MZT) correlation, coupled with a low
DZA (or DZT) correlation, flagged a possibly emergenic trait. Lykken
concluded that emergenic traits could only be detected with twin studies
because MZ twins share all their genes (and, consequently, their complex
gene configurations), whereas DZ twins share only half their genes on
average.8 It would be rare (though possible) for nonidentical twins to
inherit the same multigene combination.
The MISTRA analyses suggested an emergenic quality to several of the
twins’ measured traits. These traits included the alpha midfrequency (an
electroencephalogram spectrum parameter), the well-being scale (from the
Multidimensional Personality Questionnaire), and the rate of electroder-
mal habituation (physiological response to loud sounds). Lykken’s studies
of twins reared together added arts and crafts interests to the list of pos-
sible emergenic traits, and data from the Swedish reared-apart twin stud-
ies added extraversion.
In his 1992 paper, Lykken included a section called, “Emergenesis ver-
sus Epistasis,”9 probably anticipating objections from colleagues. Epista-
sis, a standard concept in genetics, refers to interactions among several
nonallelic genes, genes located on different chromosomes or on different
places within the same chromosome. In contrast, emergenesis goes beyond
gene interaction to include a large number of genes and to describe ob-
servable features “at the molar level, as configurations of partly genetic
traits.”10 Lykken reasoned that most complex traits are quite a bit re-
moved from specific genetic events, so emergenic characteristics can be
thought of as composites of lower level dispositions, working in concert
with environments. For example, athletic talent may require a certain
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 173

balance of speed, musculature, and motivation plus opportunities and

training. One may think of emergenesis as epistasis writ large. Of course,
major genetics advances have been made since 1992, especially with the
unraveling of the human genome in 2001. Scientists are now searching for
genes predisposing people to complex traits such as schizophrenia, autism,
and cancer, with mixed success. Finding specific genes linked to these dis-
orders is difficult because genes interact with each person’s unique genetic
background and environment—and there are many genes from which to
choose.
Lykken speculated that chance combinations of genes underlying emer-
genic traits could be qualitative steps in the evolutionary process. Such
“emergenic evolution” would require long time spans and would sup-
plement, not replace, evolutionary changes linked to accumulating mi-
cromutations (small changes in DNA that may be harmful, beneficial, or
neutral to the organism and population).11 Evolutionary perspectives on
human behavior were gaining attention in the 1990s, and it was exciting
to think that twins reared apart could contribute to this awareness. It
turned out that MZA and DZA twins were valuable resources for testing
other evolutionary-based hypotheses, as I will describe in this and subse-
quent chapters. Thus, emergenesis was an interesting and creative con-
cept that generated new ideas about behavior and ways to test them.
Earlier in this book, I suggested that comparing the different editions
of Plomin’s Behavioral Genetics could track progressive changes in the
field. The 1990 edition did not reference emergenesis, but the three edi-
tions from 1997 and after did, calling it “an extreme version of epistat-
sis.”12 Emergenesis is Lykken’s legacy, but epistasis remains a vital concept
in genetic research. For example, models assessing the capacity of viruses
to replicate in the absence and presence of drugs had greater predictive
power if they included multiple epistatic effects (i.e., interactions among
mutated genes).13 Where multiple epistatic effects end and emergenesis
begins is a key question.

Reassessing Creativity
Emergenesis showed up in a 1993 creativity paper from the project.14 This
study was led by Niels Waller, the first author on the religiosity paper
discussed in Chapter 7. Waller had been invited to comment on British
psychologist Hans Eysenck’s target article on creativity, published in
Psychological Inquiry in 1993.15 Eysenck viewed creativity as reflecting
personality and cognitive factors working together interactively, rather
than additively. At that time, twin studies had found low heritabilities for
174 B O R N TO G E T H E R — R E A R E D A PA RT

creative thinking (devising something new or original)16 and ideational

fluency (producing ideas that meet certain requirements), but none had
examined creative personality (displays of energy, flexibility, independence,
sensitivity, and openness).17 In fact, creativity in any of its manifestations
had never been studied in reared-apart twins.
The Creative Personality Scale embedded within Harrison G. Gough’s
300-word Adjective Check List was completed by the reared-apart twins.
The scale is a thirty-item measure of trait creativity based on responses
from over 1,700 individuals, many from fields requiring creative activity.
Twins endorsed items they believed were self-descriptive (e.g., reflective
and unconventional).18
Creativity data were available for forty-five MZA twin pairs, one MZA
triplet set, and thirty-two DZA twin pairs. The intraclass correlations
were 0.54 (MZA twins) and −0.06 (DZA twins)—the marked twin group
difference in the size of the correlations was consistent with emergenesis.
Waller concluded that creativity was genetically influenced but did not
run in families. Waller also challenged Eysenck’s view of psychoticism as
a creativity component because Eysenck’s twin correlations did not fit
the pattern expected for emergenic traits.

Interests and Talents

The continued development of the Minnesota Twin Registry and MISTRA
enabled additional analyses combining twins reared apart and together. In
1993, Lykken published the first reared-apart/reared-together twin study
of interests and talents.19
A mostly environmental view of the origins of interests and talents was
prevalent in the early 1990s. A 1991 paper by research psychologist Ita-
mar Gati indicated that genetic effects explained less than 5 percent of the
variance in vocational interests.20 Our vocational interest paper by Dan
Moloney, published the same year, moved the figure closer to 50 percent.21
Many other questions were addressed by Lykken in the 1993 paper on the
heritability of recreational interests,22 such as:
1. Do vocational interests, leisure-time interests, and talents inter-
correlate, and do they produce interest factors?
2. Do MZT twin intraclass correlations provide good heritability
estimates of interests and talents?
3. Are interests and talents emergenic?
The reared-apart twin sample used in the interest analysis included
fifty-three MZA twin pairs (nineteen male and thirty-four female), thirty-
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 175

six DZA twin pairs (eleven male and twenty-five female), and one MZA
triplet set (male). The interest inventory was introduced into the study
when it was already under way, so only about half the twins completed
the forms in the laboratory; twins who had been to Minnesota com-
pleted them at home and returned them by mail. The reared-together
twin sample included 924 twin pairs, in which 618 pairs were female and
524 pairs were MZ. These twins received an inventory packet by mail.
A group of 198 pairs (forty-nine MZ male, fifty-three MZ female, forty-
four DZ male, and fifty-two DZ female) completed the forms again three
years later.
The composition of the reared-together twin sample exemplified another
of Lykken’s contributions, known as the Rule of Two-Thirds.23 In 1987,
Lykken showed that volunteer twin samples in many studies are generally
composed of two-thirds MZ twins and two-thirds female twins.24 MZ
twins may be more invested than DZ twins in being twins and in partici-
pating in twin-related activities. It is also well known that females are
generally more willing research volunteers than males.
The twins completed four different measures created by Lykken:25

1. The Occupational Interest Inventory included 100 item compos-

ites that tapped broad aspects of occupational interest (e.g., skilled
trades such as appliance repair, auto mechanics, and carpentry).
Twins were told to assume equal pay and equal status across all
occupational groups. They indicated “Like,” “Dislike,” or “Indif-
ferent” for each one, with regard to how much they would enjoy the
type of work and that field’s working conditions.
2. The Minnesota Leisure-Time Interest Test included 120 items
describing a range of recreational activities (e.g., fishing, enjoying
nightlife, or going on safari). Assuming no time, money, age, or
health limitations, twins indicated the frequency with which they
would engage in such activities (1 = never to 5 = as often as
possible).
3. The Minnesota Talent Survey was a forty-item inventory of
familiar talents. An example was “Carpentry: (1) You can make
furniture, cabinets, etc., (2) You can hang a door, use power saws,
build a doghouse, etc., (3) You can do minor repairs using hand
tools, or (4) You are unskilled with hammer or saw.” Twins chose
the alternative that best fit their self-perception.
4. The Self-Rating Inventory consisted of thirty-one items concern-
ing qualities not directly assessed by the first three measures. An
example is abstract intelligence: the ability to solve intellectual
176 B O R N TO G E T H E R — R E A R E D A PA RT

problems, to understand complicated issues, to figure things out.

Twins compared their traits and abilities to those of others on a
five-point scale (1 = lowest 5 percent to 5 = highest 5 percent).
In addition to these, the Multidimensional Personality Questionnaire
was available for all reared-apart twins. This questionnaire was also in-
cluded in the packet sent to reared-together twins.
The ranges and mean values of the intraclass correlations for the inter-
est factor scores did not differ between reared-apart male and female twin
pairs, so the data were combined. We found that genetic influence explained
over 30 percent of the variance in occupational and leisure interests,
whether we looked at the separate inventory items or at the factors com-
posed of related clusters of these items. Organizing the data into larger
scales and superfactors yielded average heritabilities of 48 percent and 53
percent, respectively.
The average intraclass correlations for the MZA and MZT twins were
nearly identical at the item level—0.34 and 0.32, respectively. At the fac-
tor level the MZT twins showed somewhat greater resemblance and less
variability than the MZA twins—the average correlations were 0.49 and
0.42, respectively. A difference in the factor structure (derived from the
reared-together twin data) may have been responsible for this finding—
many MZA twins were raised outside the United States and may have had
different interest opportunities. Overall, our results contrasted sharply
with the less than 5 percent genetic contribution reported by Gati two
years earlier. Our results were also consistent with those from our 1991
vocational interest study. Lykken’s interests analysis was another success-
ful effort at replication within the MISTRA.
Each of the three questions posed at the beginning was answered:
1. Factor analysis organized the data into thirty-nine factors and
eleven superfactors. The superfactors and their reliabilities (inter-
relations among the test items) are listed in Table 9-1.
2. The MZT correlation appeared to be a more reliable and conser-
vative estimate of heritability than the value obtained by doubling
the difference between the MZT and DZT correlations. That is
because the formula overestimates narrow heritability due to MZ
twins’ shared nonadditive genes.
3. Emergenesis characterized 67 percent of the 291 items and 62
percent of the fifty interest-talent factors. Arts and Crafts was
exemplary—the MZT correlation was high, while the DZT
correlation approached zero. Such traits would fit the definition
of emergenesis.
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 177

Table 9-1. Eleven Superfactors Based on the 39 Interest-Talent Factors

Alpha
Superfactor (reliability)

Intellectual and educated 0.90

Breadth of interest 0.84
Self-esteem 0.71
Adventurous versus harm-avoidant 0.88
Solidarity 0.65
Artificer versus athlete 0.82
Religious orientation versus sensual indulgence 0.89
Personal attractiveness and charm 0.55
Agrarian activities 0.86
Male physician 0.69
Female physician 0.73

Source: Adapted from Lykken et al., “The Heritability of Interests: A Twin Study,”
Journal of Applied Psychology 78 (1993): 649–661.

Our answer to question 3 was especially interesting because many MZA

twins’ jobs and activities were similar but not exactly the same. The
reared-apart twin firemen, Mark and Jerry (Figure 9-1), both installed elec-
trical equipment, but one worked with burglar alarms and the other with
fire-suppression systems. Both twins loved music, but Mark preferred rock
and roll, and Jerry preferred country and western. Both twins were inter-
ested in forestry, but Marked had worked in the field, while Jerry had
studied forestry at school. The twins were also football fans, although
Mark favored the Dallas Cowboys and Jerry favored the Washington Red-
skins. The twins lived about sixty miles apart in New Jersey when they
met, so their team preferences did not reflect their hometowns. The twins
said that despite these differences they “liked each others’ styles.”
No one would claim that there is a gene predisposing individuals
to favor burglar alarms over fire alarms, or the Dallas Cowboys over the
Washington Redskins. Most important is that just by being themselves
Mark and Jerry showed exactly what Lykken was talking about in his
paper—the idea that particular interests seem less heritable than the dis-
positions underlying them.26 What persuaded Mark to take on forestry
in an applied way and what made Jerry pursue it in the classroom prob-
ably reflected their nonshared experiences. A huge challenge for behavioral
geneticists is to identify those nonshared experiences. Mark and Jerry also
show that how twin researchers define similarities and differences, and
whether they focus on specific traits or broad tendencies, affect heritability
estimates.
178 B O R N TO G E T H E R — R E A R E D A PA RT

Lykken also talked about a “cafeteria of experience.” Choice of experi-

ences and reactions to them are partly determined by one’s physique, abil-
ity, and temperament. But opportunities must also be available, and some
opportunities may be more appealing at one time than another. Given the
complex pathways from genes to behavior, it still amazes me that MZA
twins can be so similar in what they do and what they like. But it is not
their similarities alone that are impressive, it is their similarities relative to
those of other genetically and environmentally related family members. It
would be impossible to explain MZA twins’ similarity and adopted sib-
lings’ lack of similarity without reference to genetic factors.
Interests and talents have certain qualities that make them important
and meaningful to us. Our data also suggested that these same qualities
extend to the work we choose to do. Our research exploring the origins of
work values gave us insight into how and why we select some professions
over others.

Figure 9-1. Firemen twins Jerry Levey (left) and Mark Newman were reunited
in 1985 at age thirty-one. A year later, Jerry is enjoying a Budweiser, the twins’
favorite brand of beer. Note the pinky finger underneath the glass—both twins
held a glass this way. (Photo credit Nancy L. Segal.)
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 179

Workplace Values
In 1992, theories about what people valued in their jobs focused on envi-
ronmental events, such as family experiences and occupational socializa-
tion. Available theories at that time did not consider genetic influence,
reflected by the empirical studies reporting environmental links between
work values and gender, and education and experience. Together with our
colleagues Richard Arvey and Lauren Keller, with whom we had collabo-
rated previously, we analyzed the twins’ work values.27
The sample included twenty-three MZA twin pairs and twenty DZA
twin pairs. This sample was small relative to those included in other
analyses we conducted at this time because the inventory we used—the
Minnesota Importance Questionnaire—was a late addition to the study.28
The questionnaire included 190 paired-comparison items and 20 items
requiring absolute judgments of the importance of each work outcome.
Twins answered the paired-comparisons with reference to the base ques-
tion: “Which is more important in my ideal job?” A sample item is: “(a) I
could be busy all the time, or (b) The job would provide an opportunity
for advancement.” Twins also answered “Yes” or “No” to absolute judg-
ments with reference to the root phrase: “On my ideal job, it is important
that . . .” A sample item to which they applied that phrase is “I could
make decisions on my own.”
Twins also provided their current job so that we could assess the simi-
larity of their occupations. Five of the thirty-two twin pairs who pro-
vided this information held very similar jobs to their co-twin, and all
were MZA pairs. We also coded the twins’ jobs for occupational level or
category, such as blue collar versus white collar, and other features. Blue-
collar (e.g., flight attendants) and white-collar (e.g., attorney) jobs differ,
but they may be similar in complexity such as in their hierarchies of
authority and range of responsibilities. Similarly, two flight attendants or
two attorneys may hold the same type of job, but their jobs may differ in
terms of their number of bosses and assigned tasks.
The twins’ job levels were generally not the same—both twins in four
out of nineteen MZA pairs and two out of thirteen DZA pairs held blue-
collar positions. These data were not complete, but it seemed unlikely
that similarity in work values would be associated with similarity in job
level. Previous studies had suggested that parents’ socioeconomic status
affected their children’s work values. However, in their 1989 chapter on
information processing, McGue and Bouchard showed that the co-twins’
rearing parents’ socioeconomic levels were not very similar.29 The present
sample included a subset of the larger sample, so it was unlikely that this
180 B O R N TO G E T H E R — R E A R E D A PA RT

measure would influence twin resemblance in work values. However,

within-pair differences in parental socioeconomic status could lead to
differences in work values—a twin from a high-status family might value
creativity more than a co-twin from a lower status family who might in-
vest more in sick leave.
The Minnesota Importance Questionnaire items were organized into
twenty individual scales and six factor scales. The score variances did not
differ between the MZA and DZA twins, but the mean Autonomy score
was significantly higher for the DZA than MZA twins. The DZA twins
also scored higher than the MZA twins on the individual scales of activ-
ity, compensation, creativity, independence, and responsibility. If these
were DZT twins, I might have explained this finding with reference to
DZ twins’ generally lower investment in their twin relationship and pos-
sibly greater desire for individual achievement. There is no obvious ex-
planation for these mean score differences in reared-apart twins.
The correlations for five of the six factor scales, shown in Table 9-2,
were significant for the MZA twins, but only one (Achievement) was sig-
nificant for the DZA twins. Correlations for ten of the twenty individual
scales were significant for the MZA twins, but only two were significant
for the DZA twins. More striking were the median work value correla-
tions of 0.42 for the MZA twins and 0.16 for the DZA twins, and the
median individual scale correlations of 0.34 for the MZA twins and 0.20
for the DZA twins (not shown in the table). These results were our first
indication that what we value at work is influenced partly by our genes.
It was unfortunate that the Minnesota Importance Questionnaire was
not available for a larger twin sample—the lack of statistical power may
have explained why only two individual scale correlations (company

Table 9-2. MZA and DZA Intraclass Correlations for the Minnesota Importance Questionnaire (MIQ) Work
Value Factor Scales

Factor Scale MZA DZA

Achievement value 0.43a 0.41a

Comfort value 0.43a 0.17
Status value 0.42a 0.24
Altruism value 0.14 0.15
Safety value 0.40a 0.07
Autonomy value 0.41a 0.13

a. P < .05.
Source: Adapted from Keller et al., “Work Values: Genetic and Environmental
Influences,” Journal of Applied Psychology 77 (1992): 79–88.
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 181

policies and human relations aspect of supervision) showed a significant

MZA-DZA difference.
The heritability estimates (based on model fitting and corrected for at-
tenuation) ranged from 0.37 (Altruism) to 0.68 (Achievement), with a
median estimate of 0.46. Our relatively low heritability for Altruism was
at odds with the 0.53–0.56 estimate reported in the literature.30 Perhaps
our measure captured a different aspect of altruism or was less reliable.
Our findings showed that approximately 60 percent of the variance in
work values was tied to environmental factors and measurement error.
People are not born with work values; these values must be acquired from
work experiences and other life events. It may be that genetic factors
underlying intellect, personality, and temperament predispose us to favor
certain work outcomes and circumstances. It is also likely that job satis-
faction (analyzed by members of our research team in 1989) is affected
by how much work values are actualized while on the job, and whether
opportunities are available to pursue one’s ideal job.
Our findings made sense considering the accumulating evidence of gene-
tic influence on social and religious attitudes we saw in the literature as
well as in our own studies. We felt that our results could help organizations
and companies understand why improved work conditions and other perks
may not be embraced with the same enthusiasm across employees. In June
2010, I asked our collaborator, Arvey, to reflect on these findings when I
saw him at the International Congress of Twin Studies in Seoul.
Like most researchers, Arvey described additional measures he wishes
we had gathered from the twins. He was especially interested in details
about the twins’ work histories, such as those of the reared-apart twin
volunteer firemen Mark and Jerry. Calling the two men a “rare base rate,”
Arvey wondered about the processes that independently led each twin to
his job. Jerry, who had always wanted to be a fireman, had chased fire
trucks as a child. Mark became interested in firefighting when his friends
thought it would be fun to try, but he became committed to the service
after saving a woman’s life.31

Evolutionary Perspectives
The new discipline of evolutionary psychology emerged out of sociobiol-
ogy in the late 1980s and early 1990s. The key concept of sociobiology
was that of natural selection acting through differential reproductive suc-
cess. As such, sociobiology focused on fitness maximization, or behavior
based largely on survival and gene transmission by reproduction. In con-
trast, as I explained in Chapter 5, evolutionary psychology is concerned
182 B O R N TO G E T H E R — R E A R E D A PA RT

with finding mechanisms that evolved to meet the challenges and de-
mands of human existence, such as telling friends from foes and distin-
guishing safety from threat. Thus, it focuses mostly on explanations that
consider behavioral and cognitive functions with reference to survival and
reproduction (i.e., ultimate explanations). However, explanations of be-
havior based on immediate causal events—proximal events that activate
psychological mechanisms—also interest evolutionary psychologists.32
Evolutionary psychology and behavioral genetics have some common
goals, yet their different foci have kept them largely apart.33 As described
briefly in Chapter 5, evolutionary psychology is concerned mostly with
species uniformities, whereas behavioral genetics is concerned mostly with
individual differences. However, behavior geneticists have shown that
virtually all measured traits display genetic variation, causing evolution-
ary psychologists to ask why meaningful genetic variation exists. Various
explanations for trait variability have been proposed, such as mutations
(gene changes) and nonadditivity (genetic variance resulting from interac-
tion among genes that is not eliminated by selection), and continue to be
debated. Evolutionary psychologist and University of Minnesota graduate
Steven Gangestad believes that life history theory, the view that humans
and nonhumans direct environmental resources toward activities that will
benefit their growth, survival, and reproduction, can offer answers.34 Be-
haviors that are optimal at one time may change as a function of environ-
mental situations (nutritional availability), personal circumstances (an
individual’s age), and other factors. Human intellectual, personality, and
hormonal systems have evolved such that certain behaviors are more fa-
vorable under some circumstances than others.
There have been efforts to bring behavioral genetics and evolutionary
psychology closer together. In 2001, Bouchard and University of Texas
psychologist John Loehlin attempted to reconcile evolutionary and be-
havioral genetic perspectives on personality.35 They suggested that basic
emotions and motivations such as anger, fear, nurturance, and curiosity
vary across a broad range because different levels of these traits would
not be detrimental to fitness, except at the extremes where they would be
selected against. Genetic variation in such traits could reflect the different
environmental conditions that Gangestad discussed.
Given the intellectual divide, few researchers attend meetings of both
the Human Behavior and Evolution Society and the Behavior Genetics
Association. This is unfortunate because (1) behavioral genetics provides
informative methods for testing evolutionary-based hypotheses, and
(2) evolutionary psychology offers behavioral genetics another theoreti-
cal perspective for interpreting findings.
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 183

I was introduced to sociobiology in the late 1970s as a graduate student.

My mentor at the University of Chicago, the late Daniel G. Freedman, was
writing Human Sociobiology: A Holistic Approach at that time.36 Many
of his students, including me, attended his informal weekly seminar on
the subject. Freedman’s book, published in 1979, appeared four years
after E. O. Wilson’s controversial book Sociobiology: The New Synthesis,
which had given a name and set of concepts to a large body of research.37
Sociobiology was a comprehensive review of findings on sociality, and
made a strong case for a biological basis to social behavior. Wilson’s last
chapter on the biological basis of human social behavior became quite
controversial.
W. D. Hamilton’s 1964 inclusive fitness theory of altruism (altruism will
be selected for when the behaviors benefiting a recipient exceed the cost to
an actor) may have had the greatest impact on sociobiological (and evolu-
tionary psychological) research of any scientific work in this area.38 Ham-
ilton asserted that altruism should be directed more often toward close kin
(i.e., individuals likely to carry common alleles) than distant kin as a way
of indirectly transmitting one’s genes to future generations. In general, one
should be predisposed to help a sister over an aunt, and an aunt over a
cousin—sisters share half their genes on average, by descent, whereas nieces
and aunts share one-quarter and cousins share one-eighth.
Hamilton defined inclusive fitness as an individual’s reproductive suc-
cess, enhanced by the individual’s effects on relatives other than offspring.
Of course, such behaviors would not occur consciously. No one helps
anyone else for the purpose of preserving his or her own genes, but people
act as though they do—helping others feels good in an immediate sense,
one reason why such behavior may be maintained. Identification of imme-
diate events affecting the display and continuity of social closeness, coop-
eration, and other beneficial behaviors between relatives are also impor-
tant. It seemed that using inclusive fitness theory to understand human
social relationships would not detract from the significance of everyday
influences on social behavior, but would provide another way of interpret-
ing that behavior.
The implications of Hamilton’s reasoning for MZ and DZ twin rela-
tionships were striking. Perhaps MZ twins’ unusual closeness could be
understood in an ultimate sense—that is, it would pay to act more kindly
toward one’s MZ twin than one’s DZ twin because that would be like
“helping” one’s own genes. It would also add a “why” to the twin’s be-
havior. My 1982 doctoral dissertation—a study of cooperation, competi-
tion, and altruism between twins—was undertaken with these themes
in mind.
184 B O R N TO G E T H E R — R E A R E D A PA RT

I always believed that twins reared apart would offer unique tests of
evolutionary-based hypotheses.39 Canadian psychologists Chuck Craw-
ford and Judith Anderson agreed, as shown in their 1989 article on the
topic.40 They suggested grouping MZA twins according to hypotheses
about specific behaviors or rearing conditions and comparing their devel-
opmental pathways. For example, some evolutionary psychologists have
predicted that female sexual maturity should occur earlier in father-absent
homes than in father-present homes.41 They have reasoned that it would
be a better life history strategy for girls from father-absent homes to
leave their disorganized environments to start families of their own. The
idea of relating different behavioral outcomes to different life experiences
is exactly what the MISTRA was all about—our goal (as stated in the In-
troduction) was to identify associations between differences in the twins’
life histories and the twins’ behavioral differences. There were some wonder-
ful opportunities to use the MZA and DZA twins in evolutionary-based
analyses.
The late evolutionary psychologist Linda Mealey claimed that for
evolutionary psychologists “kinship, via the effect of inclusive fitness,
constitutes a core construct of relevance to all social interaction.”42 Twin
studies exemplify this view, especially studies of social relationships be-
tween MZA and DZA co-twins, a topic I review in Chapter 12.
Several years after arriving in Minnesota, I met Linda Mealey, a psy-
chology professor at St. John’s University in St. Cloud, Minnesota. She
and I became close friends and colleagues and continued our association
after I left Minnesota for California. Linda was interested in how gene-
tic and environmental variables affected reproduction-related variables,
such as age at marriage and number of children. She asked Bouchard for
access to the reared-apart twin data, and he agreed—Bouchard was care-
ful about releasing his files outside the department, but he was generous
and encouraging when he trusted colleagues and believed they were seri-
ous about their work. In 1993, Linda and I published the first paper ana-
lyzing reared-apart twin data in the context of evolutionary theory.43 We
showed that while some behaviors related to reproduction may be gene-
tically influenced, they do not affect the number of children.
Our sample included fifty-five MZA twin pairs and twenty-seven DZA
same-sex twin pairs. The mean ages were 38.2 years (standard deviation
[SD] = 14.5) for the males and 42.2 years (SD = 10.7) for the females, so
not all twins had completed their reproductive years. The twins had also
been raised in modern societies that had transitioned from high fertility
and mortality to low fertility and mortality, probably reducing the vari-
ance of some measures. Small, statistically significant relationships may
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 185

have been larger without these constraints. Linda and I drew our data
from the Life History Interview and the Briggs Life History Form.44 The
intraclass correlations for our seven outcome measures are displayed in
Table 9-3.
These values suggested that genetic factors influenced age at first
marriage. It turned out that place of birth, a factor that (naturally!) is con-
stant for co-twins, had a significant effect on age at first marriage. Gene-
tic effects were also suggested for age at first date, paralleling what we
found for age at first marriage and what others have found for age at first
intercourse. Among males, age at first marriage was also associated with
parents’ religious affiliation. The high DZA correlations for timing of chil-
dren suggested an environmental effect outside the family but shared by
the twins—perhaps the prevailing cultural expectation was that people
should start their families by a certain age. The lack of genetic influence
on number of children surprised me. When Linda and I began our work,
there were no twin studies on family size. In later years, twin studies esti-
mated the heritabilities for number of children to be 0.32 for males and
0.34 for females.45
Our goal, mostly exploratory, was to see if personality and health char-
acteristics (e.g., extraversion and teenage health) were related to repro-
ductive events (e.g., age at first child and number of children). The indi-
vidual findings from that study were less meaningful than the big picture.
Our data suggested that males’ reproduction-related behaviors were more
affected by genetically influenced measures of personality and health than
were females.’ For example, in males later age at first marriage was associ-
ated with low activity level (−0.35) and poor childhood health (−0.44).

Table 9-3. Intraclass Correlations for Reproduction Outcome Measures

MZA DZA
Measure (n = 25–47) (n = 15–26)

Age at marriage 0.76 0.53

Age at first child 0.19 0.64
Age at second child 0.38 0.55
Number of children 0.06 0.10
Age at first date 0.33 −0.02
Dating frequency 0.33 0.40
Not want children 0.33 0.15

Note: n = pairs.
Source: Adapted from Mealey and Segal, “Heritable and Environmental Variables
Affect Reproduction-Related Behaviors, but Not Ultimate Reproductive Success,”
Personality and Individual Differences 14 (1993): 783–794.
186 B O R N TO G E T H E R — R E A R E D A PA RT

Females’ reproductive outcomes were influenced more by family dynam-

ics than were the males’. For example, females with multiple marriages had
experienced parent-child conflict and family instability. The only impor-
tant reproduction variable—reproductive success—was directly related
to childhood health and only for males.
The outcomes of our study conformed to evolutionary-based predic-
tions. That is, males are more likely to have their reproductive capaci-
ties reduced under adverse conditions such as poor health than are fe-
males. Males with less favorable sexually dimorphic characteristics are
also less likely to attract potential mates than males with favorable
characteristics. Thus, we concluded that while proximal or immediate
developmental mechanisms may have evolved in accordance with evolu-
tionary expectations, they might not be relevant in modern societies.
The MISTRA researchers completed other evolutionary-based analyses
of the reared-apart twin data during the study’s later years and beyond. It
was important to incorporate this perspective into our work because of
rapidly growing interest in this area. Bouchard tells people that he started
studying evolutionary psychology because I encouraged him to. He was
only partly right—he also knew that it was becoming an important per-
spective in the field, one that would add meaning to our findings. In fact,
the word evolution eventually appeared in the titles of seven of our publi-
cations. Evolutionary considerations were also discussed in other papers,
such as Lykken’s emergenesis paper reviewed previously46 and our 1990
IQ paper published in Science.47
My friend Linda Mealey passed away in 2002 at the age of forty-seven.
Her condition was diagnosed as cancer, and it spread quickly. I still resist
the urge to call her when interesting topics arise, especially related to twins.

Personality and Cognition

Invitations to write book chapters increase when laboratories are produc-
tive. The growing number of empirical papers from the MISTRA proba-
bly led to the 1993 publication of several chapters and a book, Twins as
a Tool of Behavioral Genetics, coedited by Bouchard and German gene-
ticist Peter Propping.48 Because Bouchard’s chapters on personality and
intelligence published in other collected works surveyed many studies in
the field, not just the MISTRA, they will not be reviewed here in full. In-
stead, I have summarized them briefly, highlighting material on the reared-
apart twins. Some of the MISTRA findings reported in the chapters were
new because they were preliminary analyses on their way to becoming
full-length papers. Sometimes older data were used in new ways.
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 187

Bouchard authored an overview of genetic and environmental influence

on adult personality in a volume edited by Joop Hettema and Ian Deary.49
The first part of his two-part treatise summarized the major model fitting
results from twin, adoption, and extended family studies on the Big Five
personality traits. The Big Five emerged as a dominant personality para-
digm in the early 1990s. Big Five traits are easy to remember if you think
of OCEAN: Openness (imaginative, inventive versus narrow, simple),
Conscientiousness (organized, efficient versus disorderly, careless), Extra-
version (sociable, dominant versus retiring, withdrawn), Agreeableness
(cooperative, sympathetic versus quarrelsome, cold), and Neuroticism
(calm, stable versus distraught, unstable). The second part of his chapter
critiqued material that has been discussed throughout this book, so I will
focus on the first part of his chapter.
The first of Bouchard’s two tables displays model fitting results from a
meta-analysis of reared-apart twin data from Minnesota and Sweden,
an analysis originally done by John Loehlin (Table 9-4).50 Shared family
environments make modest contributions in both models. The data also
show the difficulty in choosing between models requiring special twin
and sibling environment parameters and nonadditive genetic effects. Spe-
cial twin and sibling effects occur when the behavior of one twin or sib-
ling affects the behavior of a co-twin or co-sibling. The late behavioral
geneticist David Rowe also observed “sibling mutual influence”—pairs
of same-sex nontwin brothers and sisters who liked each other or shared
friends were more alike in delinquency than those who disliked each
other or did not share friends.51
Bouchard’s second table showed the MZA, MZT, DZA, and DZT Big
Five correlations based on personality data from a Swedish twin study and
various samples from the Minnesota reared-apart and reared-together

Table 9-4. Alternative Models for Twin Data on the Big Five Personality Traits

Additive MZ Sibling Additive Equal

Big Five Trait Genes Environment Environment Genes Epistasis Environment

Extraversion 0.36 0.15 0.00 0.32 0.17 0.02

Neuroticism 0.31 0.17 0.05 0.27 0.14 0.07
Conscientiousness 0.28 0.17 0.04 0.22 0.16 0.07
Agreeableness 0.28 0.19 0.09 0.24 0.11 0.11
Openness 0.46 0.05 0.05 0.43 0.02 0.06
Source: Adapted from Bouchard, “Genetic and Environmental Influences on Adult Personality:
Evaluating the Evidence,” in Foundations of Personality, ed. P. Joop Hettema and Ian Deary, 15–44
(Dordecht: Kluwer, 1993).
188 B O R N TO G E T H E R — R E A R E D A PA RT

twin studies. Table 9-5 shows only the weighted mean correlations based
on these samples. The MZA data suggested an average heritability of
about 0.50, and the lack of difference between the MZA and MZT twins
suggested a shared environmental effect of zero. However, the 0.50 herita-
bility left plenty of room for environmental influences unique to individu-
als. Nonadditive genetic effects were indicated for extraversion and neu-
roticism because the DZA and DZT correlations were less than half the
MZA and MZT correlations, suggesting emergenic effects on these traits.
Bouchard often said that the absence of shared environmental effects
on personality was “counter-intuitive, but true.” It was certainly one of
the most provocative and controversial findings from our study. The fact
that other investigators were finding the same thing using other instru-
ments and samples increased our confidence in the findings. I accept
these results, based as they are on the twins’ self-reports. We also asked
the twins’ spouses and companions to complete the Adjective Check List
to describe the twins’ personality, information that is not yet analyzed. It
is likely that these data would have shown somewhat less MZA co-twin
similarity than that found from the twins’ self-reports. A previous reared-
together twin study reported that peer ratings of the twins’ personalities
indicated genetic influence across the different traits, although less than
that derived from the twins’ self-ratings.52
If the twins had rated themselves relative to their co-twins, I wonder
if we would have captured what Shields suggested—that MZA twins are
more alike than MZT twins on some personality traits because MZA
twins did not develop social roles within their twinship. As I’ve written

Table 9-5. Weighted Mean Correlations for the Big Five Personality Traits for Minnesota Twins Reared
Apart and Together

Big Five Trait MZA MZT DZA DZT

Extraversion 0.51 0.53 −0.03 0.17

Neuroticism 0.54 0.47 0.27 0.15
Conscientiousness 0.50 0.57 0.09 0.35
Agreeableness 0.51 0.42 0.10 0.18
Openness 0.60 0.43 0.31 0.18
Mean 0.53 0.48 0.15 0.20

Note: The combined samples included the following pairs: MZA (n = 61), DZA (42),
MZT (n = 99), and DZT (n = 99) who completed the California Psychological Inventory);
and MZA (n = 52), DZA (n = 33), MZT (n = 553), and DZT (n = 459) who completed the
Multidimensional Personality Questionnaire.
Source: Adapted from Bouchard, “Genetic and Environmental Influences on Adult
Personality: Evaluating the Evidence,” in Foundations of Personality, ed. P. Joop Hettema
and Ian Deary, 15–44 (Dordecht: Kluwer, 1993).
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 189

elsewhere in this book, people interacting with twins usually find that
twins behave more similarly when they are apart than when they are to-
gether. However, we did not detect this effect in the MISTRA self-report
personality data.
Plomin and McClearn’s Nature, Nurture, and Psychology summarized
major behavior genetics advances that had occurred during the previous
decade.53 At its 1992 meeting, the American Psychological Association
had named genetics as one of the themes that best reflected present and
future psychological research.54 The MISTRA probably had a lot to do
with that—the book’s editors wrote that McGue’s chapter fell within
what they considered “the front lines of genetics research.”55
A striking feature is the series of graphs reproduced in Figure 9-2 that
show IQ intraclass correlations from past and present reared-apart twin
studies.56 The correlations ranged between 0.64 and 0.74 for the three
previous studies, and 0.69 and 0.78 for the more recent MISTRA and
Swedish studies. The weighted average correlation was 0.73, using a to-
tal of 162 MZA twin pairs. The similar finding based on the three earlier

1.0
n = 45
n = 19 n = 12 n = 38
n = 48 TOTAL =
.8 162
.78
.74 .73
MZA Correlation

.68 .69
.6 .64

.4

.2

0
Newman Juel-Nielsen Shields Bouchard Pedersen Weighted
et al. (1937) (1965) (1962) et al. (1990) et al. (1992) Average

Study

Figure 9-2. MZA Twin IQ correlations from five studies. Adapted from
McGue et al., “Behavioral Genetics of Cognitive Ability: A Life-Span Perspec-
tive,” in Nature, Nurture, and Psychology, ed. Robert Plomin and Gerald E.
McClearn, 59–76 (Washington, D.C.: American Psychological Association, 1993).
Note: Bars at the top indicate standard errors.
190 B O R N TO G E T H E R — R E A R E D A PA RT

studies and the MISTRA was robust, as I indicated in Chapter 5, but it

was more so with the addition of the forty-five Swedish pairs.
The MZA twin correlation directly estimates heritability. Thus, Mc-
Gue noted that IQ heritability was about 75 percent, considerably higher
than the 50 percent figure provided by other kin comparisons. Because
most MZA twins were adults, and because the results were consistent
across studies, he concluded that the large MZA IQ correlation probably
reflected increased genetic influence over time, rather than biased sam-
pling or assessment.
McGue’s reasoning fits nicely with my own work on virtual twins, the
same-age unrelated pairs I have been studying since the early 1990s. The
virtual twin IQ correlation, which directly estimates shared environmen-
tal effects on behavior, is 0.28 (n = 140), leaving the remaining 0.72 ac-
counted for by genetic factors, nonshared environmental factors, and
measurement error.57 These results appear to mirror those from the MZA
twins for the following reason: recall that a subgroup of the virtual twins
(n = 43 pairs), who were very young children when I first studied them,
showed a drop in their IQ correlation from 0.30 to 0.11 at their second
testing in later childhood.58 This change suggested a reduction in shared
environmental effects on IQ over time, with both genetic and unique en-
vironmental factors becoming more important. It is likely that when vir-
tual twins are assessed as adolescents and adults their IQ correlation will
approach zero, as reported in other adoption studies.59 If this were the
case, it would correspond to the MZA IQ findings of about 0.75 genetic
and 0.25 nonshared environmental variance plus error. However, based on
accumulated studies of twins and other kinships, a more accurate parti-
tioning of adult IQ variance may be 60 percent genetic, 35 percent non-
shared environmental, and 5 percent measurement error.60
Bouchard’s 1993 chapter “The Genetic Architecture of Human Intelli-
gence” did not include new MISTRA twin data.61 Instead, it provided evi-
dence supporting a hierarchical structure of approximately ten primary
abilities with three major peaks: General Visualization, Fluid Intelligence,
and Crystallized Intelligence, with General Intelligence at the top. Gen-
eral Visualization refers to mental manipulation of figural information,
Fluid Intelligence refers to genetically based abilities, and Crystallized
Intelligence refers to skills acquired through education and experience.
Memory span and vocabulary are examples of the ten primary mental
abilities that cluster under their relevant factors (i.e., Fluid Intelligence
and Crystallized Intelligence, respectively). Bouchard also called for
meta-analytic studies of intelligence and an end to ad hoc criticism of
individual studies. In this context, he revisited some of the objections to
 C R E AT I V I T Y, W O R K V A L U E S , A N D E V O L U T I O N 191

his reared-apart twin studies that I detailed in earlier chapters, most no-
tably Chapter 3.
Bouchard and Propping’s book, Twins as a Tool of Behavioral Gene-
tics, focused on what twin studies had revealed about general intelligence
and personality traits. It resulted from a 1992 Dahlem workshop held in
Berlin on the subject: “What Are the Mechanisms Mediating the Genetic
and Environmental Determinants of Behavior?” Dahlem conferences are
unique in that background papers, written around discussion topics, serve
as the basis for small group discussions. MISTRA data were not pre-
sented, but some interesting historical notes on twins were embedded
within Bouchard and Propping’s introduction.
They wrote that the idea of studying twins went back many years,
to before Galton’s time. Saint Augustine, in The City of God, used twins
to refute astrological claims. But, as discussed earlier, it was Sir Francis
Galton who is famous for developing the twin method for formal scien-
tific inquiry in 1875.62

1992 to 1993
In August 1993, we began our collaboration with colleagues at the Jean
Mayer U.S. Department of Agriculture Human Nutrition Research Cen-
ter on Aging (Energy Metabolism Laboratory) at Tufts University. Now
twins would fly from Minneapolis to Boston to complete a five-day analy-
sis of body size measures, food intake, and energy expenditure. This five-
year study, funded for one million dollars by the U.S. Department of
Health and Human Services, assured us that separated twins would con-
tinue to be identified and assessed. Finding financial support was always
our biggest difficulty (as is described in Chapter 14), and we never knew
if the study would continue for two years or one year—or at all.
We added five new twin pairs to the study in 1993, two MZA and
three DZA opposite sex sets, as well as four spouses and partners. We also
studied seven follow-up twin pairs (four MZA and three DZA), as well
as four spouses. We presented seventeen conference papers and invited
lectures. In California, I began working on twin studies of odor identi-
fication and sensitivity, using the University of Pennsylvania Smell
Identification Test.63 I sent forms to Minnesota for the reared-apart twins
to complete.
The first reared-apart female twin pair I had studied in 1982 was one
of the four MZA twin pairs who returned to Minnesota in 1993. The
twins sent me their photographs, and it was striking but not surprising
to see how alike they still looked and how similarly they had aged. The
192 B O R N TO G E T H E R — R E A R E D A PA RT

twins reminded me of a photo Bouchard had hung in all of his offices

showing a pair of beautiful elderly MZ female twins (Figure 9-3). The
photo had been a gift from identical twin photographers, Kathryn M.
Abbe and Frances M. Gill, and had appeared in their stunning volume,
Twins on Twins.64 Better than any study, the picture captures essential
elements of the aging process—the graying hair and matching creases in
their skin suggest that physical changes are largely influenced by genetic
factors. A small inset of the twins looking so similar at a younger age
underlines the idea of genetically based physical development, similar to
what I noticed about the Minnesota pair. Bouchard loved this photo-
graph, and he relished photographing reared-apart twins beside it.
Abbe and Gill also gave Bouchard a unique set of slides showing them-
selves together at many different ages. The images of these twins as infants,

Figure 9-3. Bouchard’s favorite photo: identical twins Genevieve and Eloise
Reed at age ninety-four. (Photo courtesy Kathryn M. Abbe.)
Figure 9-4. MZA twins Keith (left) and Jake as infants, at age eleven or
twelve (Keith) and ten (Jake), at age twenty-four on their first assessment day in
Minnesota, and at age fifty-five during a taped interview for Keith’s son’s school
project. Keith is right-handed, and Jake is left-handed. Note the opposite
positioning of the handles of the twins’ coffee cups. (Photos [infant, childhood,
and adult years] courtesy of Keith and Jake. Photo credit [Minnesota study]
Thomas J. Bouchard Jr.)
194 B O R N TO G E T H E R — R E A R E D A PA RT

children, young women, and beyond suggested a rare time-lapse photog-

raphy in which the individual ages before viewers’ eyes. In this case, it was
two individuals, and they were growing up exactly the same way. Bouchard
often opened his twin lectures with these slides, which regularly capti-
vated his audiences. When we positioned the MZA twins’ separately shot
childhood and adult photos next to each other, we saw the same thing.
The genetically influenced unfolding of physical growth and development
is well known, but MZA twins make this process especially striking, as
shown in Figure 9-4.
 CHAPTER TEN

Family Environments, Happiness, Sensation Seeking,

and the MMPI (1994–1997)

B etween the years 1994 and 1997, the MISTRA published

papers on a number of new research topics. The twins showed
that current perceptions of the past are genetically influenced,
people have happiness set-points, and practice does not always lead to
perfect performance. They also showed that there was genetic influence
on the personality correlates of psychopathology on the Minnesota Mul-
tiphasic Personality Inventory (MMPI), and that the same genes affected
impulsivity and sensation seeking. By the start of 1994, nearly 130 pairs
had been studied.

All in the Family

Graduate student Yoon-Mi Hur (now a faculty member at Mokpo
National University in South Korea) joined the MISTRA in 1990 after a
two-year stint in the University of Minnesota’s psychometrics program.
She was charming, funny, and exceedingly smart. After graduating from
Minnesota in 1993, she returned to South Korea to teach classes, con-
duct research, and develop the South Korean Twin Registry. “The strik-
ing behavioral similarities I witnessed during the assessment and analyses
convinced me of the importance of genes in human behavior,” Hur re-
called. Her first MISTRA paper out of seven, co-authored with Bouchard,
examined twins’ perceptions of their childhood environments.1
The genetics of the family environment drew considerable research
attention in the late 1980s and early 1990s. It sounded contradictory, but
evidence supported genetic influence on family environment measures and
196 B O R N TO G E T H E R — R E A R E D A PA RT

on how we construct and perceive our environments. Parents who enjoy

reading and have books and magazines at home are likely to have intelli-
gent children—but these parents also transmit genetic factors associated
with high intelligence to their sons and daughters. Some children will
thrive in educationally stimulating environments and seek them out,
while others may find them somewhat overwhelming and prefer to learn
in quieter spaces.
Research on the genetics of family influences began with the late psy-
chologist David Rowe’s 1981 and 1983 studies showing that adolescent
MZT twins held more similar perceptions of parental support (i.e., accep-
tance or rejection), but not parental control (i.e., permissiveness or restric-
tiveness), than DZT twins.2 His results were replicated in Sweden using the
retrospective reports of adult twins reared apart and together.3 In 1995,
the MISTRA published the first analysis of reared-apart twins’ perceptions
of their family environments using multiple indicators. One measure was
the Block Environmental Questionnaire, which assesses parental charac-
teristics.4 The other measure was the Family Environment Scale described
in Chapter 7. Hur used the same parenting dimensions described in the
previous studies in order to test the generality of these findings.
Data were available for over one hundred reared-apart twin pairs,
fifty-eight MZA and forty-six DZA. The standard version of the Block
included ninety-two items describing parental characteristics and the family
context. Six scale scores were generated (Maternal Acceptance-Rejection,
Paternal Acceptance-Rejection, Cohesion, Maternal Intellectual-Cultural,
Paternal Intellectual-Cultural, and Organization) based on the responses
of 276 twins and their spouses. The data were also reduced to two major
parent-perception factors: Support (parental support and acceptance) and
Organization and Cultural Orientation (parental restrictiveness and con-
trol). Factor analysis of the ten Family Environment Scales (Cohesion,
Expressiveness, Conflict, Independence, Active Recreational, Achievement
Orientation, Intellectual-Cultural, Moral Religious Emphasis, Organiza-
tion, and Control) produced the same two major factors that we found
for the Block.
The support-related scales in both the Block and Family Environment
Scale showed relatively high MZA correlations (e.g., Cohesion: 0.42; In-
dependence: 0.39), while the control-related scales showed relatively low
MZA correlations (e.g., Organization: −0.07; Moral Religious Emphasis:
0.18). The pattern of correlations in Table 10-1 demonstrated that we
replicated Rowe’s previous findings of genetic influence on support and
the absence of genetic influence on organization; thus, the reared-apart
twin sample was not atypical in terms of family perceptions. Within-study
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 197

Table 10-1. MZA and DZA Intraclass Correlations for the Block Environmental Questionnaire and Family
Environment Scale

Block Environmental Questionnaire

MZA DZA

n (pairs) 39–44 31–37

Factor
Support 0.51a −0.24
Organization and cultural 0.12 0.31b

Family Environment Scale

MZA DZA

n (pairs) 58 46
Factor
Support 0.31a 0.17
Organization and cultural −0.03 −0.19

Note: The Block Environmental Questionnaire was added one year after the study
began, so it was available for fewer twins than the Family Environment Scale.
a. P < .01.
b. P < .05.
Source: Adapted from Hur and Bouchard, “Genetic Influences on Perceptions of
Childhood Family Environment: A Reared Apart Twin Study,” Child Development 66
(1995): 330–345.

confirmation was provided by comparable results from the two question-

naires. Modeling showed that, for adults, Support during childhood was
affected by additive genetic and nonshared environmental factors, whereas
Organization was mostly affected by the nonshared environment; that is,
genetic factors played a modest role in how adults viewed parental con-
trol during childhood. A model fit to the combined data showed that
genetic factors explained 44 percent of the variance in Support and 28
percent of the variance in Organization.
Hur speculated that genetic effects on Support reflected reactive
genotype-environment correlation, that is, that MZA twins probably
elicited matched responses to their behaviors from their respective rear-
ing family members. However, it was also possible that the MZA twins
processed, interpreted, and/or recalled information the same way, so even if
their recollections were distorted (as most people’s are to some degree)
they were distorted in the same manner. As Hur pointed out, retrospective
bias may be partly genetically based.
Hur also suggested that the common rules and standards of Western
families could explain the reduced heritability for Organization. Parental
198 B O R N TO G E T H E R — R E A R E D A PA RT

rule-setting is universal, but most of the twins came from Western cul-
tures. Of course, some parents are more permissive than others. Another
possibility is that some children and adolescents reject parental authority
in most forms, so even indulgent parents seem restrictive in their chil-
dren’s eyes.

The Roots of Happiness

David Lykken was enormously interested in what makes people happy—
lykken means “the happiness” in Norwegian. Lykken’s twin analyses
showing genetic influence on happiness drew considerable attention.5
One of his most striking findings was that one MZA twin’s happiness at
time 2 was closely associated with his or her co-twin’s happiness at time
1, showing a heritable, stable component to that trait. However, Lykken
also believed happiness (including his own) could be raised, at least mo-
mentarily, by doing things one liked to do.
Studies on the roots of happiness had been ongoing for years when
Lykken began this research. Everyone wants to be happy, but not every-
one is, so scientists had tried finding the life events associated with hap-
piness. A 1995 study by David Myers and Edward Diener showed that
higher social status and higher wages, factors expected to raise happi-
ness, did not.6 Instead, it seemed that people are able to cope reasonably
well with their misfortunes and still maintain a characteristic degree of
happiness. At the same time, not everyone was equally happy. Myers and
Diener suggested that happiness was affected by social relationships, reli-
gious faith, cultural factors, and personal goals.
Lykken used twins reared apart and together to explore genetic influ-
ence on human happiness. His book on the subject was an outgrowth of
his 1996 twin study conducted with Auke Tellegen.7 They first examined
the happiness levels of 2,310 reared-together twins using a contentment
measure from Lykken’s Self-Rating Inventory. Twins were asked, “Taking
the good with the bad, how happy and contented are you on the average
now, compared with other people?” Responses ranged from 1 (lowest 5
percent of the population) to 5 (the highest 5 percent of the population).
Lykken found that 86 percent of the twins placed themselves in the top 35
percent. Lykken suggested, albeit tentatively, that natural selection might
have favored happy people who were likely to mate, have children, and
become our ancestors.8 This could be true, but only to an extent because
natural selection would not have favored habitually happy people who
might have overlooked threats and physical dangers.9 It may also be that
many twins (especially MZ twins) are happier and more secure than others
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 199

because of their close companionship. A Danish study found a lower

suicide risk among twins than nontwins, which the investigators explained
by twins’ close social ties.10 It is unfortunate that Lykken’s analysis was
not applied to nontwin individuals.
In another analysis, Lykken replaced his single-item measure with the
well-being scale from Tellegen’s Multidimensional Personality Question-
naire. Educational attainment, social status, income, and marital status
explained just trivial amounts of the happiness variance. Even Tellegen’s
traditionalism scale that correlated moderately with religious commit-
ment correlated only 0.05 with well-being, meaning that religious people
are not happier or unhappier than nonreligious people. This finding chal-
lenged the view that embracing religious beliefs predisposed people toward
happiness. But Lykken’s findings on the heritability of well-being drew
more attention—he was the first researcher to show that genes partly ex-
plained individual differences in happiness.
The Multidimensional Personality Questionnaire had been taken twice
by seventy-nine MZT twin pairs and forty-eight DZT twin pairs at the
mean ages of twenty and thirty years. The correlation between their well-
being scores obtained at the two different ages was 0.50. There is another
informative calculation that can be done with twin data called a cross-
twin, cross-time correlation. Here, twin A’s score at one age is correlated
with twin B’s score at another age, and vice versa. If this statistic is higher
for MZ than DZ twins, it means that the stable part of the trait is gene-
tically influenced.
Lykken found that the MZT twins’ cross-twin, cross-time correlation
of 0.40 was equal to 80 percent of the 0.50 retest correlation (0.40/0.50).
However, the same correlation for the DZT twins was about zero. These
results suggested that genes affected what Lykken called “the happiness
set-point” or stable part of happiness. The happiness set-point is each per-
son’s characteristic happiness level that might fluctuate depending on daily
events and experiences. Subsequent research has supported the idea of a
genetically influenced happiness set-point, although individual differences
in returning to one’s baseline level have been observed.11 The near-zero
DZT and 0.44 MZT happiness correlations, based on the first question-
naire administration, suggested that happiness might be emergenic.
The Multidimensional Personality Questionnaire was also available
for forty-four MZA twin pairs and twenty-seven DZA twin pairs. The
estimated heritability for well-being from our 1988 personality study was
0.48.12 Fortunately, twenty-six reared-apart twin pairs had taken the ques-
tionnaire twice about 4.5 years apart, yielding a retest correlation of 0.67.
Assuming a long-term well-being stability of 0.60, Lykken showed that
200 B O R N TO G E T H E R — R E A R E D A PA RT

the heritability of their stable happiness component was also 80 percent

(0.48/0.60), and that the heritability of their momentary happiness was
about 50 percent. It seemed that an MZ twin’s happiness was a better
predictor of his co-twin’s current or future happiness than was the co-
twin’s educational achievement, status, or income.
Intraclass correlations for the well-being scale are displayed in Table
10-2. They included data from some new reared-apart twin pairs that
had been studied since the publication of our 1988 personality article.
The similar MZA-MZT correlations suggested that shared environ-
ments do not affect how happy people become. Most importantly, Lyk-
ken showed that people have a characteristic well-being level. But daily
events would not significantly alter the usual happiness level to which
people revert when the glory of success or the despair from grief subsides.
This information seemed potentially valuable to patients and to therapists
trying to understand and improve their clients’ quality of life—but this
information could be misconstrued, especially if Lykken’s next words
were not read carefully.
Lykken concluded, “Trying to be happier [may be] as futile as trying to
be taller.”13 However, a close reading of the study shows that if people tried
to determine what made them happy, and minimized what made them
unhappy, then despite their genetic backgrounds their happiness levels
could be elevated—albeit not permanently—as long as they were doing
things they enjoyed. In fact, Lykken regretted that quote, noting in his
1999 happiness book, “That pessimistic conclusion is not impelled by the
data and, in fact, I believe it is wrong.”14
Lykken’s happiness research was cited in the New York Times, the Lon-
don Times, the Manchester Guardian, and in Italian, German, and South
American periodicals.15 He called the coverage “unprecedented.” The hap-
piness set-point concept was embraced by a number of researchers, includ-
ing Ed Diener, Paul Costa, and Robert McCrae. The three had found that
happiness levels were stable, and that people generally did not maintain

Table 10-2. MZA and MZT Intraclass Correlations for Well-Being

Twin Type n (Pairs) ri (Standard Error)

MZA 75 0.52 (± 0.10)

DZA 36 −0.02 (± 0.17)
MZT 647 0.44 (± 0.03)
DZT 733 0.08 (± 0.04)

Source: Adapted from Lykken and Tellegen, “Happiness Is a Stochastic Phenomenon,”

Psychological Science 7 (1996): 186–189.
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 201

unusually high or low moods.16 They also acknowledged that severe psy-
chological problems or continual bad luck could override the happiness
set-point. This makes sense because extreme situations have a greater
impact on behavior and mood than ordinary daily events. Lykken’s esti-
mated happiness heritability of 50 percent was questioned by psycholo-
gist Howard Weiss, who suggested that genetic influence on individual
differences in happiness could be as low as 25 percent or as high as 75
percent.17
Tellegen’s take on the happiness research emphasized within-person
variability despite the genetic effect: “There’s a range of oscillation
around a given person’s set-point. This means that you can be an emo-
tional Pavarotti . . . and it will still average out.”18 Lykken elaborated on
the nature of the happiness set-point by asking if it is just an average of re-
peated measures or “is there a true homeostatic process that is activated by
any deviation from the value characteristic of the individual?”19 That is,
perhaps when someone experiences a pleasant or unpleasant event the body
works to restore that person to his or her characteristic happiness level.
The media attention led Lykken “to speculate more freely” on his find-
ings than he might otherwise have done.20 He remarked on the contented
style of people doing work that brought them pleasure, singling out the
clever plumbers in his favorite shop and the skillful garbage collectors in
his neighborhood. “The people in my examples either have above-average
set-points or they transcend their set-points by doing their jobs with skill
and good humor.”21 He also urged people to act on their happiness feel-
ings, recognizing that money and status many not increase happiness and
can even lower it.
Lykken had advice for parents: “Nurturing the child’s innate pro-social
proclivities, which include the inclination toward constructive endeavor,
constitutes the most important responsibility of parents, the socialization
of their children.”22 In other words, effective parenting means being aware
of, and responsive to, children’s talents and creativity. Finding genetic in-
fluence on the happiness set-point or on any other behavior never denied
the crucial roles that families played. Everyone associated with the MIS-
TRA believed this.

Practice Effects
It was surprising to find genetic influence on practice, something parents
pressure their children to do when they are learning a new task. Most par-
ents encourage their children to repeat the alphabet, replay musical scales,
or work at whatever skills they have to get better. Most people assume
202 B O R N TO G E T H E R — R E A R E D A PA RT

that “practice makes perfect,” but the nature and outcomes of practice
are more complex. The same amount of practice may affect two individ-
uals differently, depending on their commitment, motivation, and ability.
Author Malcolm Gladwell wrote that one has to practice for 10,000
hours before becoming an expert.23 Some people might need that much
practice, but a few may be experts from the start. Galton, father of the
Twin Method, told the story of a young man who claimed first place in
a national running race without formal training or experience.24 It seems
practice and guidance can refine natural talents, but they cannot create
them.
By the mid-1990s, several reared-together twin studies had reported
a genetic influence on motor skills, but the effects of practice over more
than a single day had not been examined.25 Our studies showed a genetic
component to improvement following practice by engaging the twins in
a multiday motor skill acquisition task. The late Professor Paul (Bill) Fox
directed this phase of the assessment, which was the first of its kind in the
history of reared-apart twin studies.
A student assistant arrived in the laboratory every Thursday, Friday,
and Saturday morning of the assessment week to individually escort twins
to a special testing room. The psychomotor battery included a hole steadi-
ness test (to examine involuntary hand movement), the Purdue Pegboard
Test (to assess gross movements of the fingers, arms, and hands, and also
fine fingertip dexterity), and the Rotary Pursuit Test (to measure ability to
keep a stylus on a moving target). Data were available for sixty-four
MZA twin pairs and thirty-two DZA twin pairs.26
The Rotary Pursuit apparatus resembled an old-fashioned record player.
A disk placed on the top of a rectangular box was rotated in a clockwise
direction, at a constant speed of sixty revolutions per minute. Holding a
metal stylus, twins were instructed to keep contact with a metal spot on
the moving disk, while the length of time on target was recorded. Each of
the twenty-five trials in each thirty-minute session lasted for twenty sec-
onds, and trials were separated by ten-second breaks.
The twins’ responses were organized into five blocks per day, with
each block consisting of five trials. The MZA and DZA twins did not
differ in their performance levels, and both groups improved consider-
ably during the first day. They also improved after a break from the activ-
ity between days one and two, and two and three. Variability within each
twin group increased over the three days because practicing from day to
day improved some twins’ skills more than others.
The MZA twins’ correlations were stable and showed a slight increase
over time, whereas the DZA twin’s correlations were unstable and irreg-
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 203

ular. The higher MZA correlations indicated significant genetic influence

on performance, but there were more interesting features of these data.
The first trial on day one showed a higher heritability for the percent-
age of time on target (0.66) than the other four blocks on that day. Fox
suggested that individuals rely on their current abilities when attempting
a new task, and MZA twins would be more highly matched than DZA
twins in this regard. And, except for day one, genetic influence increased
as the trial blocks progressed—the highest heritability (0.74) occurred on
day three on the fourteenth trial block. The MZA twins maintained a high
level of similarity, whereas the DZA twins generally drifted apart. Even
more striking, the MZA twins’ similarity across trial blocks did not differ
from the block-to-block similarity of a single individual, which was not
the case for the DZA twins. A purely environmental model for all fifteen
trials was rejected.
Table 10-3 shows the intraclass correlations for the slope (change in per-
formance across days). The MZA twins’ higher slope correlations were
increasing, while the DZA twins’ lower correlations were decreasing.
Model fitting indicated a heritability of 0.66 on performance (time on tar-
get) and 0.40 on reminiscence (improvement). Nonadditive genetic factors
were suggested because of the relatively high MZA correlations.
Everyone improves with training—in fact, the lowest-scoring twin on
day three outperformed the highest scoring twin on day one. But not ev-
eryone improves to the same degree. Fox suggested that practice sessions
on this task enhanced the twins’ environmental similarity, reducing exist-
ing differences between MZA twins more than between DZA twins. When
environments are uniform, genetic differences contribute more to behavioral
differences among people. Conversely, when environments are different,
genetic differences become less salient. Because environmental similarity

Table 10-3. Intraclass Correlations for the Slope and Reminiscence for the Rotary Pursuit Test

MZA DZA
Day Measure (64 pairs) (32 pairs)

1 0.56 0.24
Slope
2 0.69 0.17
3 0.72 0.11

Note: Day 3 was added after the study began, resulting in 58 MZA twin pairs for
related measures.
Source: Adapted from Fox et al., “Genetic and Environmental Contributions to the
Acquisition of a Motor Skill,” Nature 384 (1996): 356–358.
204 B O R N TO G E T H E R — R E A R E D A PA RT

enhances heritability (the proportion of variance associated with genetic

differences), the declining DZA correlations reflected genetic effects on per-
formance and practice.
People were quite taken by these new findings. They had practical ap-
peal and could help them understand why they might do well on some
tasks with little practice and only slightly better on others despite lots of
practice. The study was reported by Reuters news service, whose article
included an interview with one of the authors, then University of Kansas
psychology professor Scott Hershberger, who had spent a year at Minne-
sota. Hershberger claimed, “Since the identical twins [were] building upon
a common genetic substrate . . . practice [was] going to have the same
effect for them.”27
The genetic influence on the effect of practice in sports performance
has not been considered, but would be expected. Identical twin Olympic
athletes Jiang Tingting and Jiang Wenwen (synchronized swimming) and
Mark and Bob Bryan (tennis) are known for their extraordinary perfor-
mances, but few nonidentical co-twins both perform at elite levels.28 The
MZ twin athletes’ rigorous and consistent practice could minimize slight
ability differences between them, increasing the heritability of physical
skills. Similar processes may apply when twins pursue academic activities,
especially those requiring repetition or rote learning. Teenage identical
twins Erin and Becky Glendaniel of Delaware have both participated in
national spelling bees in different years.29 We found a high heritability for
spelling, as discussed in a later chapter.

Personality and Psychopathology

In 1996, a group of MISTRA colleagues examined the Minnesota Multi-
phasic Personality Inventory (MMPI) personality indicators of psycho-
pathology, studied for the first time in reared-apart twins.30 The MMPI
is a widely used personality and mental health inventory, developed in
1940 by Starke R. Hathaway and J. C. McKinley.31 Interest in personality-
psychopathology associations and the genetic underpinnings of these
associations was high at the time.
The article’s lead author was David DiLalla, Southern Illinois Univer-
sity psychologist and former student of Irving Gottesman.32 DiLalla met
Bouchard at the 1987 Behavior Genetics Association meeting in Minne-
apolis when he was a graduate student at the University of Virginia. “Got-
tesman introduced us. I’m sure I appeared kind of silly. I was in awe of the
whole thing—the MISTRA had rock star status. And Bouchard was such
a striking figure.” After a postdoctoral position at the University of Colo-
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 205

rado and a clinical internship in Denver, DiLalla joined Southern Illinois

University in Carbondale.

I was interested in the genetic and environmental correlations between per-

sonality and psychopathological traits, but I couldn’t gather these data in
Carbondale. It was lucky for me that the MISTRA’s MMPI data hadn’t been
spoken for. The other plus was Bouchard’s collaborative generosity. Gottes-
man arranged for us to analyze the data, and Bouchard trusted us to de-
velop our ideas and interpretations. But he read the manuscript and made
contributions.

Other contributors to the MMPI study were University of Colorado

behavioral geneticist Greg Carey, another former student of Gottesman
who had been DiLalla’s postdoctoral mentor, as well as Gottesman and
Bouchard. Carey knew Bouchard from his graduate student years in Min-
nesota, and he recalled that Bouchard significantly influenced his thoughts
about the origins of behavior. Carey was a postdoctoral fellow at the
University of Washington in St. Louis when Bouchard began the MISTRA.
He told me, “[The study] was not a screwball idea. I had concerns about
how many twins [Bouchard] would find and if they would be willing to
come to Minnesota for one week. But I was fascinated by the Jim twins.
Case studies can generate great hypotheses, and Tom accomplished a
great deal more.”
The MMPI results were available for sixty-five MZA twin pairs and
fifty-four DZA twin pairs (thirty-eight same-sex and sixteen opposite-
sex). The MMPI includes over 500 behavioral style and preference items,
rated as “true,” “false,” or “can’t say.” The twins’ responses were com-
puter scored for the ten clinical scales (e.g., Depression and Hypochon-
driasis), the three validity scales (e.g., Lie and Frequency), and the thirteen
Wiggins content scales. The Wiggins content scales (e.g., Family Problems
and Authority Conflict) covered thoughts, experiences, and behaviors as-
sociated with psychopathology. Gottesman had studied these scales in
1977 using adolescent twins reared together and found evidence of both
genetic and environmental effects.33
Intraclass correlations for the ten MMPI Clinical Scales ranged from
0.29 to 0.64 for the MZA twins, and from 0.03 to 0.23 for the DZA twins,
consistent with a genetic influence. A nonadditive genetic influence was
suggested for the Frequency scale because the DZA correlation (−0.02)
was much less than the MZA correlation (0.52), and because it did not fit
the model’s assumption that only additive genetic effects were important.
The Wiggins scales behaved like the clinical scales in that the MZA twins
showed greater resemblance than the DZA twins. Intraclass correlations
206 B O R N TO G E T H E R — R E A R E D A PA RT

ranged from 0.23 to 0.65 for the MZA twins and from −0.12 to 0.34 for
the DZA twins. Only two scales (Religious Fundamentalism and Femi-
nine Interests) didn’t fit the additive genetic model.
According to DiLalla, there were no real surprises in the results. The
overall finding of 44 percent genetic variance in the MMPI’s clinical and
content scales agreed with previous studies, and the results came from what
DiLalla called a “fantastic resource.” In Carey’s view, the study “joined the
nexus of behavioral genetic research that sputtered to life in the ’60s and
’70s and truly came to life in the ’80s and ’90s.”
An interesting and novel feature of DiLalla’s work was the multivari-
ate profile analysis, looking at the score elevation and tracing the peaks
and valleys of the twins’ MMPI scores. The average score elevation looks
at each twin’s average scale score, while the profile shape looks at the dif-
ference of each twin’s MMPI scale from the average. This analysis showed
that genes affected both the average score elevation (how high or low
twins scored overall) and the profile shape (the form of psychopathology,
that is, whether twins scored high or low on the same scales). Such find-
ings had been reported for twins reared together, yet they were more
striking among twins who had always been apart. But DiLalla was cau-
tious in interpreting that data because the group of reared-apart twins
was a small, nonpsychiatric sample. He suggested that certain profile con-
figurations found in the study did not necessarily signal specific psychiat-
ric problems that were more or less genetically influenced. Still, the scales
associated with psychotic behavior (such as bipolar disorder) seemed more
heritable than those associated with neurotic behavior (such as disrup-
tions in memory, awareness, and identity).
The authors noted that regional associations in the MMPI had been
previously reported, a factor that could have enhanced the twins’ similar-
ity. This seems unlikely because, in most cases, both the MZA and DZA
co-twins had been raised in the same country, and the DZA twins were
still less alike. However, DiLalla and his colleagues wisely gave us a look
at the scores of the MZA twins raised in different countries. Their within-
pair differences of 8.5 and 5.1 T-score points were, respectively, equal to
and below the MZA twin group average of 8.6.34 Thus, the culture of
rearing does not necessarily affect personality or the predisposition to
psychopathology. For comparative purposes, the investigators cited MZA
twin pairs with differences above the group average, as large as 16.4
and 17.2. In one case, the twins became discordant for major affective
disorder with psychosis, and in the other case a head injury five years
before assessment had probably caused neurological damage in one
twin. These last two cases showed that MZA twins raised in the same
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 207

culture could differ in personality and psychopathology for a variety of

reasons.
Studying a large group of MZA twins raised in different countries would
be the best approach to understanding how cultural factors affect behav-
ior. But twins raised apart are rare, and twins raised in different cultures
even more so. The MISTRA had only six such pairs, four MZA and two
DZA. In 2008, Yoon-Mi Hur and I studied a pair of twenty-eight-year-old
MZA female co-twins of Korean descent, raised in South Korea and the
United States, respectively. We found both striking similarities (e.g., identi-
cal IQs and nearly identical self-esteem scores) and interesting differences
(e.g., weight difference of 11.86 pounds and dissimilar sports participa-
tion) between them. These twins did not take part in the MISTRA.35
The Minnesota investigators would revisit the MMPI in 1999. Mean-
while, other personality traits and associations among them were ex-
plored, such as impulsivity and sensation seeking.

Impulsivity and Sensation Seeking

Hur’s second study used reared-apart twins to see if the association be-
tween impulsivity and sensation seeking could be partly explained by the
same genes.36 Impulsivity is behavior expressed without thought or de-
liberation in response to objects or events.37 Sensation seeking is a behav-
ior that entails a search for new, intense, and complex experiences, and a
willingness to take risks to acquire them.38 People high in impulsivity
tend to be high in sensation seeking. When Hur began her analysis in the
mid-1990s, genetic effects of 0.40 for impulsivity39 and 0.36 for risk-
taking had been reported.40 Biological correlates common to these traits,
such as monoamine oxidase level, had also been described.41 Based on
these findings, Hur reasoned that a genetic connection between impulsiv-
ity and sensation seeking was likely.
The measure of impulsivity was the Control scale of the Multidimen-
sional Personality Questionnaire, and the measure of sensation seeking
was Marvin Zuckerman’s Sensation Seeking Form V. This form is orga-
nized into four subscales, namely Thrill and Adventure Seeking, Experi-
ence Seeking, Disinhibition, and Boredom Susceptibility. The twin sample
included fifty-seven MZA pairs and forty-nine DZA pairs (Table 10-4).
With the exception of Thrill and Adventure Seeking, a genetic influence
was indicated, especially by the relatively low DZA correlations. The
amount of genetic variance shared by Control and the Sensation Seeking
scales was 55 percent, supporting the expected link between impulsivity
and sensation seeking.
208 B O R N TO G E T H E R — R E A R E D A PA RT

Table 10-4. Intraclass Correlations for Sensation Seeking Measures and for Control

Scale MZA DZA

a
Thrill and adventure seeking 0.47 0.48a
Disinhibition 0.50a 0.10
Boredom susceptibility 0.43a 0.16
Experience seeking 0.54a 0.28b
Control 0.51a 0.05

a. P < .01.
b. P < .05.
Source: Adapted from Hur and Bouchard, “The Genetic Influence between Impulsivity
and Sensation-Seeking,” Behavior Genetics 27 (1997): 455–463.

A significant feature of this study was that it demonstrated a well-

reasoned idea about two personality traits, impulsivity and sensation seek-
ing, based on observed relationships between these two traits and their
likely biological underpinnings. Of course, males’ sensation-seeking ten-
dencies exceed those of females, making an argument for conducting anal-
yses separately by sex. But Hur’s modest sample size precluded organizing
the twins in this way.
Hur’s work had practical implications. Impulsivity had been linked to
problem behaviors such as substance abuse, conduct disorder, and suicide
attempts. Some individuals acting on impulse may inadvertently endanger
themselves and the people around them. Impulsive behavior could be a
warning sign to parents and teachers of behavior that goes beyond rea-
sonable risks. However, some people depend on heightened levels of intel-
lectual stimulation, emotional excitement, or physical danger to feel con-
tent. Matching experience to temperament is everyone’s challenge.
While Hur was completing her 1997 study, the MISTRA’s analyses of
vocational interests and job satisfaction continued. The 1994 studies in
these areas were extensions and replications of earlier work.

Replications: Vocational Interests and Job Satisfaction

Graduate student Deborah Betsworth’s 1992 PhD dissertation42 built upon
Dan Moloney’s 1991 study of vocational interests,43 discussed in Chap-
ter 8. Bouchard provided Betsworth with scores for a slightly larger num-
ber of reared-apart twins than Moloney had used. Her article appeared
in 1994.44
Betsworth gathered data from twins reared together, twins reared apart,
adoptive parents and children, and biological parents and children. The
participants had already completed the Strong Interest Inventory, but their
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 209

data were rescored to yield the scales comprising the Hansen Combined
Form Scales. She did this because the Hansen scales include all twenty-
six scales common to the different versions of the Strong. The scales she
used included six General Occupational Themes, seventeen Basic Interest
Scales, and two special scales (Academic Comfort and Introversion/
Extroversion); the special scales had not been examined in our earlier
study. However, Betsworth omitted the Adventure scale from her analysis
because of its low reliability. The median correlations for the eight differ-
ent kinships are shown in Table 10-5.
Several features of these data are worth noting. The higher MZT than
DZT, and higher MZA than DZA correlations demonstrated genetic influ-
ence. However, the lower MZA than MZT correlations for the General
Occupational Themes and Basic Interest Scales suggested shared environ-
mental effects. Analyzing all eight kinships simultaneously yielded a heri-
tability of 0.36, lower than the 0.40 to 0.50 reported in Moloney’s paper.
Betsworth also reported an environmental effect of 0.64 for vocational
interests (0.09 shared and 0.55 nonshared), whereas we had previously
found an environmental effect of 0.50 (all nonshared). Betsworth sug-
gested that Moloney’s use of three vocational interest measures rather
than one explained the discrepancies between the two studies. We had
actually reported a range of heritabilities for the different measures: 0.37
for the Basic Interest Scales, 0.44 for the Jackson, and 0.50 for the ten
vocational interest factors generated by items in these instruments. Thus,
Betsworth’s 0.36 based only on the Strong agreed with Moloney’s 0.37.
Betsworth’s median correlations for the biological family members
were just slightly higher than those of the adoptive family members who
shared no common genes. Perhaps studying vocational interests concur-
rently in parents and children produced misleading results. People’s inter-
ests change with age, and popular activities in one generation may be
outmoded in another generation. Studying twins overcomes this problem
because of their common age, and studying twins reared apart overcomes
the shared environmental confounds that plague family research.

Replications and extensions of the MISTRA’s earlier work also involved

a survey of job satisfaction and work values.45 Richard Arvey continued
to collaborate on this project, and University of Minnesota industrial rela-
tions professor Brian McCall, University of Pennsylvania economist Paul
Taubman, and University of Minnesota graduate student Marcie Cavana-
ugh were added.
Arvey replicated our earlier findings twice using two samples of twins
reared together. The first sample included ninety-five MZT and eighty
Table 10-5. Median Intraclass Correlations for Vocational Interests

Adopted Adopted Biological Biological

MZT DZT MZA DZA Parent-Child Sibling Parent-Child Sibling
Scale (1,960 pairs) (1,212 pairs) (59 pairs) (33 pairs) (283 pairs) (63 pairs) (332 pairs) (60 pairs)

General occupational themes 0.48 0.23 0.31 0.07 0.10 0.11 0.15 0.15
Basic interest scales 0.47 0.21 0.33 0.05 0.08 0.08 0.13 0.14
Academic comfort 0.54 0.24 0.46 0.25 0.11 0.04 0.21 0.22
Introversion/extroversion 0.50 0.24 0.52 0.07 0.06 0.24 0.18 0.08

Source: Adapted from Betsworth et al., “Genetic and Environmental Influences on Vocational Interests Assessed Using Adoptive and Biological
Families and Twins Reared Apart and Together,” Journal of Vocational Behavior 44 (1994): 263–278.
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 211

DZT male twin pairs from the Minnesota Twin Registry. Our 1989 pa-
per had analyzed these same data for thirty-four MZA twin pairs, yield-
ing a job satisfaction heritability of 0.30.46
As discussed earlier, intrinsic job satisfaction refers to opportunities to
use one’s abilities and talents, extrinsic job satisfaction refers to one’s
working conditions, and general satisfaction refers to overall contentment
with one’s work situation. Intraclass correlations from the replicated anal-
ysis are shown in Table 10-6.
Note that the MZA-MZT intrinsic satisfaction correlations did not dif-
fer, despite differences in rearing, gender, and sample size. Both MZ cor-
relations exceeded the DZT correlations, but all three were low for extrin-
sic satisfaction. The MZT general satisfaction correlation was lower than
the corresponding MZA correlation. Our 1989 findings of genetic effects
on intrinsic satisfaction and lack of genetic effects on extrinsic satisfaction
were confirmed. Genetic effects on general satisfaction were suggested, but
the estimated 0.16 heritability in our replication (see below) was lower
than the 0.30 heritability in our 1989 study. The variances of this measure
across the two samples were similar, so this difference may have been as-
sociated with unknown features of the two samples.
The reared-together sample was large enough to let Arvey estimate the
genetic and environmental variance in job satisfaction. The best model
for intrinsic satisfaction showed that 23 percent of the variance was ex-
plained by genetic factors, and the best model for general satisfaction
showed that 16 percent of the variance was explained by genetic factors.
A genetic effect on extrinsic satisfaction was not indicated by these data,
consistent with our previous analysis.
When extrinsic and intrinsic job satisfaction are analyzed at the same
time, the possibility that both are genetically influenced but to different
degrees is raised. Lower genetic influence on extrinsic than intrinsic job

Table 10-6. Intraclass Correlations for the Three Job Satisfaction Measures

Scales MZA MZT DZT

Intrinsic satisfaction 0.315a 0.29b −0.02

Extrinsic satisfaction 0.109 0.09 −0.01
General satisfaction 0.309a 0.19a 0.01

a. P < .05.
b. P < .01.
Source: Adapted from Arvey et al., “Job Satisfaction: Environmental and Genetic
Components,” Journal of Applied Psychology 74 (1989): 187–192; Arvey et al., “Genetic
Influences on Job Satisfaction and Work Values,” Personality and Individual Differences
17 (1994): 21–33.
212 B O R N TO G E T H E R — R E A R E D A PA RT

satisfaction would make sense because most people like having caring
bosses and good benefits. Greater variability, some genetically based,
should characterize intrinsic job satisfaction because people value oppor-
tunities for creativity and achievement differently.
Arvey’s second replication (reported as Study 2 in the same paper) used
male twins from the National Academy of Sciences-National Research
Council’s sample of white male twin veterans.47 The sample sizes were
huge—1,152 MZT twin pairs and 1,055 DZT twin pairs. These twins
completed sixteen job importance items and one general job satisfaction
item in a 1974 questionnaire specific to that study. The MZT correlation
(0.21) for general job satisfaction significantly exceeded the DZT corre-
lation (.05), although both were significant. The MZT twins’ correlations
were higher on all seventeen measures, significantly so for ten of the six-
teen job importance items (e.g., “provided much free time” and “repre-
sented a challenge”).
One exception to the finding of genetic influence concerned Family Busi-
ness, for which the similarity of the MZT (0.42) and DZT (0.39) twins
hardly differed. Of course, genetic influence on this item was not ex-
pected because most people would value their family’s business to some
degree due to family ties, in addition to (or instead of) its intrinsic interest.
After eliminating this variable, a 0.27 genetic component to overall job
satisfaction was found, close to our 1989 finding of 0.30. Arvey also found
a 0.35 genetic estimate for the remaining fifteen work values. This finding
was important because it replicated our 1992 study, using a different in-
strument and different twin sample. The fact that Family Business did not
show a genetic effect indicated that the analyses discriminated among the
different work values.
These replicated studies were important, given the criticism our 1989
job satisfaction paper had evoked from several industrial organizational
investigators. The fact that the studies were conducted with different
samples and one study had administered a different instrument increased
confidence in the findings. But none of these studies were perfect. The
model did not consider nonadditive genetic effects, and both replication
samples were male. Nevertheless, the results were encouraging, and Arvey
believed the information would be useful to employers hoping to improve
their employees’ work lives. The genetic effects, while significant, were not
substantial, inviting opportunities for change from within the job culture.
It was also apparent that a genetic perspective was starting to seep into
the minds of researchers who had not previously considered this view.
A widely cited 1993 article by University of Georgia social psychologist
Abraham Tesser reviewed evidence for the heritability of social attitudes,
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 213

and he argued for bringing genetic perspectives to psychological theory

and experimentation.48 At the same time, Tesser was aware of the slow
acceptance of a genetic view of attitudinal differences by his sociology
colleagues. Nevertheless, twin studies of earning power,49 social atti-
tudes, and religious interests50 were proliferating in the mid-1990s.
Increased interest in genes and behavior can also be explained by the
Human Genome Project (1990–2003), aimed at identifying all 20,000–
25,000 human genes and sequencing the three billion DNA subunits.51
Parallel advances in molecular genetic techniques made it possible to
search for genes tied to complex behaviors and diseases. The press closely
covered progress in these areas such that a genetic viewpoint was perme-
ating the consciousness of the public as well as the social scientists.52

Chapters and Papers, 1994 to 1997

Looking over Bouchard’s curriculum vitae for the years 1994 through 1997
shows that MISTRA-related chapters in edited volumes outnumbered
MISTRA-related empirical articles about 2.5 to 1.0. I suspect that the
chapters were accumulating quickly because Bouchard received so many
invitations to contribute to psychology handbooks, annual reviews, and
conference proceedings. Bouchard preferred analyzing data, but he had
trouble turning down interested colleagues. In those years, the team as-
sessed eleven twin pairs new to the study and twelve twin pairs who had
been studied before, for about five per year on average, which was less
than had been accomplished in several previous years. The pool of reared-
apart twins might have been dwindling, but as I will show in the last chap-
ter new sources of reared-apart twins are surfacing.
There was a total of fourteen MISTRA chapters or review papers,
mostly on intelligence and personality, published between 1994 and 1997.
They are too detailed to summarize fully, but I have prepared a selective
sampling of key material. (A complete list of MISTRA publications ap-
pears on the Web site.)

Other Significant Work

A 1994 highlight was Bouchard’s essay, “Genes, Environment and Per-
sonality,” published in a special issue of the journal Science on genes and
behavior.53 His review of historical developments in personality research
showed that, until the 1980s, the relatively small reared-together twin stud-
ies yielded heritability estimates of 50 percent. Since then, larger samples,
reared-apart twin and adoption studies, and model fitting techniques had
214 B O R N TO G E T H E R — R E A R E D A PA RT

reduced that figure only slightly to just over 40 percent. With common
environments explaining 7 percent of the variance, and nonshared envi-
ronment and measurement error equally divided among the remainder,
two-thirds of the reliable personality variance could be tied to genetic
factors. Graphs showing the same gene-environment breakdown for the
Big Five personality factors across three independent replications that in-
cluded twins reared apart and together told this important story. Bouchard
made an impressive case for the genetic and environmental effects and
structure of personality.
One of Bouchard’s points, often lost in discussion, was that finding
nonshared environmental effects on personality did not mean that such
events affected personality traits in systematic or predictable ways. Par-
enting style could, for example, be a cause of children’s behavior in some
cases, but a consequence in others. Permissiveness by parents might en-
courage independence in some children, but might result from rebellious-
ness in others. This observation recalled the “gloomy prospect” discussed
in Chapter 8.
Bouchard also defined the “next big hurdle” in personality research as
that of attempting to understand the function of individual personality
differences in an evolutionary context. “The purpose of this variation is
undoubtedly rooted in the fact that humans have adapted to life in face-
to-face groups.”54 Knowing why people differ in personality would trans-
form behavioral genetics from a descriptive to an explanatory discipline.
We would know why we differ, not just that we differ. Siblings may differ
in personality because they inherit different genes, but also because they
occupy different places in the family. Research has found that later-born
children are more rebellious or risk-taking than older, more mature chil-
dren in the family, possibly due to their need to secure parental resources
that might go to their elder siblings.55 Other evolutionary-based research
has examined the personality correlates of fertility and mate desirability.56

Also in 1994, Arvey and Bouchard authored a book chapter summarizing

the MISTRA’s studies of job satisfaction and work values.57 The MISTRA
had completed enough reared-apart twin research on work-related traits
to fill a chapter. That chapter offered a model showing how genetic and
environmental influences might affect variables studied by industrial or-
ganizational psychologists; this model is displayed on the Web site. Arvey
believes that the model is still relevant, but he would add outcome vari-
ables such as leadership, work design (physical aspects of the job), salary,
financial well-being, financial risk, and occupational switching. He would
also add a path indicating a direct reciprocal relationship between what
were labeled Genetic Differences and Environmental Differences, because
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 215

genetic expression is affected by environments and environments reflect

genetic differences. For example, a person’s artistic potential requires op-
portunities for expression, and such workplace opportunities reflect a
person’s interests and talents.
Arvey also speculated that some variables in his model might eventually
be expanded to include specific genes associated with work-related be-
haviors. I will discuss the impact of molecular genetic advance on reared-
apart twin studies in the final chapter.
Several chapters written by Bouchard in 1995 and 1996 were not about
the MISTRA in particular, but they were influenced by the MISTRA’s
findings. His chapter in the 1995 International Handbook of Personality
and Intelligence called for bringing behavioral genetic and evolutionary
perspectives to longitudinal behavioral studies.58 He suggested that re-
peated measures on twins and adoptees would disentangle the genetic
and environmental confounds inherent in most longitudinal analyses.
Bouchard’s late friend and colleague Ron Wilson showed us this with his
1980s longitudinal twin studies in Louisville—Wilson’s concept of gene-
tically influenced “spurts and lags” in mental development are still fa-
mous. Wilson found that MZT twins showed similar patterns of intel-
lectual growth and stability during childhood, while DZT twins showed
patterns that differed. The IQ similarity in adult MZA twins was even
more compelling in light of these findings because Wilson showed that
intellectual changes during development, not just scores at a given time,
had a partial genetic basis.59 It is likely that the adult MZA twins under-
went similar intellectual growth patterns in childhood and adolescence,
ending up alike when we studied them as adults.
The importance of viewing behavior from an evolutionary perspective
is a recurrent theme in many of Bouchard’s chapters. He believed that
traditional life-course approaches did not consider the adaptive nature of
individuals’ life histories. Bouchard also showed special interests in tradi-
tionalism and tolerance, identifying Robert Altemeyer’s Right-Wing Au-
thoritarianism Scale as the best contemporary measure of such traits.60
Most scientists in the 1980s believed that authoritarianism was largely
social or cultural in origin, but Bouchard favored an evolutionary ex-
planation. He stressed the universality of following rules and conven-
tions, and the fact that parents set standards for their children to follow.
Right-wing authoritarianism became a focal point of Bouchard’s work
in the late 1990s and, subsequently, the origins of what he called the
“Traditional Moral Values Triad”—authoritarianism, conservatism, and
religiousness.
A footnote to Bouchard’s chapter, “Genetics and Evolution: Implications
for Personality Theories,” caught my eye: “I believe that social scientists
216 B O R N TO G E T H E R — R E A R E D A PA RT

will eventually give up this naïve [environmentally driven, standard social

science] model in the face of evidence.”61 Bouchard asserted that Table 4
in our 1990 Science IQ paper,62 showing genetic influence on nearly thirty
psychological traits, was the clearest refutation of the idea that human be-
havior was mostly environmental in origin. He asked why the criticism
seemed “interminable,” and tried to answer the question himself: “All deci-
sions by human beings are made on the basis of information that satisfies
rather than being conclusive. A natural consequence of this fact is that it
will always be possible to argue that there is not sufficient scientific knowl-
edge about a topic to make an unalterable decision.”63
Bouchard published seven book chapters and reviews in 1997. These
publications summarized current findings on intelligence, personality, and
other psychological measures from the MISTRA and elsewhere. One of his
papers, published in The Sciences, addressed the implications of reared-
apart twin studies for human reproductive cloning.64 Human reproductive
cloning had attracted considerable attention following the 1996 cloning of
Dolly the Scottish lamb.65 I had also used the reared-apart and reared-
together twin findings to assess its possible behavioral consequences.66
Bouchard and I agreed that the lack of perfect similarity between MZA
co-twins on every measured trait meant that clones and donors would not
be exactly alike in behavior. And unlike twins, clones and donors would
belong to different generations, offering them access to different opportu-
nities and experiences. In separate papers, we both mentioned that clones
and donors would not share intrauterine environments as did twins. In
my writings I also reminded readers that MZ twins’ shared prenatal envi-
ronments mostly produced differences between them, so clones and do-
nors might actually be more alike in traits such as birth weight than MZA
or MZT twins. In fact, the degree of similarity between clones and donors
could vary across traits as we had been finding for the MZA twins. In
accordance with Experience Producing Drive Theory it was reasonable
to suppose that clones and donors might be driven toward similar people,
places, and events within their respective environments.
I did not agree with Bouchard’s statement that “clones are simply identical
twins by another name.” Identical twins are clones by definition (gene-
tically identical organisms), but clones are not identical twins because
they do not share their time of conception, prenatal environment, parents,
or generation.67

The MISTRA data were integrated into the chapters, tables, and charts of
widely used textbooks in behavior genetics and psychology.68 A graph
with IQ findings that first appeared in the 1997 behavioral genetics primer
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 217

by Plomin et al. would be reprinted with updates in later editions.69 The

MZA IQ data were now organized into “old” and “new” studies, as shown
in Figure 10-1. The “old” studies were based on reviews by Bouchard and
McGue,70 and amended by Loehlin,71 and included the findings from
Newman et al., Shields, and Juel-Nielsen.72 The “new” studies were those
by Bouchard et al. and by Pedersen et al.73 The somewhat higher herita-
bilities provided by the “new” (0.78) versus “old” studies (0.72) most
likely reflected participants’ older ages. The heritability of IQ had been
shown to increase over time, probably because small genetic effects be-
come more important as individuals gain greater environmental control.74
The graph also showed that IQ heritability increased with the degree of

1.0
Genetically related
.86
Environmentally related
.78
.8 Both .72

.60
.6
Correlation

.47
.42
.4
.32
.28
.24 .24
.19
.2

0
Together Adopted-apart Adoptive Adopted-apart Together

Relationship P-0 Sib P-0 Sib P-0 Sib Virtual “Old” “New” MZ DZ
Twin MZ MZ

Number of pairs 8,433 26,473 720 203 1,491 714 140 65 93 4,672 5,533
Genetic relatedness 0.5 0.5 0.5 0.5 0.0 0.0 0.0 1.0 1.0 1.0 0.5
Same home? Yes Yes No No Yes Yes Yes No No Yes Yes

Family designs Adoptive designs Twin designs

Figure 10-1. Average IQ correlations across different pairs of relatives.

(Adapted from Plomin et al. 2008.) Updating the “new” MZA twin correlation
yields a value of 0.75 (n = 119), based on data from Pedersen et al. (1992) and
Johnson et al. (2007). Johnson’s paper, which reports general intelligence
findings for the full MISTRA MZA twin sample, is reviewed in Chapter 13.
Pedersen’s result (n = 45, ri = 0.78) used the first principal component; the
Johnson group’s result (n = 74, ri = 0.73) used the g factor score. The virtual twin
data are from Nancy L. Segal, “Twins: The Finest Natural Experiment,”
Personality and Individual Differences 49 (2010): 317–323.
218 B O R N TO G E T H E R — R E A R E D A PA RT

genetic relatedness between people; that is, full siblings who share 50 per-
cent of their genes on average, by descent, are more alike in IQ than half-
siblings who share 25 percent. This pattern of similarity has been repeated
in several reviews and remains a robust finding.

Looking Ahead: 1998 and Beyond

In 1998, the number of empirical papers from the MISTRA increased,
and the numbers held steady even after the last pair of new twins (number
137) was studied in April 1999; Bouchard studied the last pair of follow-
up twins a month later. This upward trend was partly due to the continu-
ing collaborations with Susan Roberts and Paul Fuss at Tufts University
in Boston. Roberts received a federal grant in 1993 to study the twins’
body composition and nutrition, letting Bouchard assess more new twins
and reassess those who had visited Minnesota previously. New graduate
students wrote dissertations on the twins’ ego development and religious
interests, work they eventually turned into publications.75 And the new
reared-apart twins expanded the databases, allowing for further analyses
of continuing topics such as intelligence and personality, and analyses of
new topics such as social relationships and body asymmetries.

In April 2010, I visited Minnesota a second time and met again with Meg
Keyes, Bouchard’s former student and assistant. Keyes had worked with
Bouchard when the Jim twins were found, and again for six years as asso-
ciate director of the project (1992–1998) after I had left. She recalled the
“incredibly long hours” that each assessment took and the long prepara-
tion time before each assessment. One day, she and graduate student Katie
Corson were assembling the twins’ inventory booklets, a labor-intensive
task. Bouchard left for lunch, and as he was passing a McDonald’s restau-
rant, a young homeless man asked him for change. Bouchard said he could
not give him money, but he could give him a job. The young man spent the
rest of the afternoon making booklets with Keyes and Corson.

At the end of 1997, Bouchard sent his usual annual report to his primary
funding source at that time, the David H. Koch Charitable Foundation. It
was an unusually thick (1.25 inches) volume covering twin recruitment,
twin assessment, professional activities, scientific papers, and newspaper
articles published between 1995 and 1997. Two major articles about the
study had appeared in the New Yorker76 and in Psychology Today,77 and
a chapter had appeared in a popular science book,78 all of which were
appended to the report.
 F A M I LY, H A P P I N E S S , S E N S AT I O N S E E K I N G , A N D T H E M M P I 219

But it was the “Director’s Comments” that caught my eye in 2010 be-
cause they included the first formal sign that the MISTRA was winding
down:

The recruitment of new twins to this program will most likely cease in the
summer of 1998 after nearly nineteen years. The reasons for ceasing recruit-
ment of new twins are two-fold. The first is that funding for the center has
declined. The second is that the number of new twins being located each
year has declined.

It is still hard to believe that the MISTRA no longer exists except as

data waiting to be analyzed. Had I read those words in 1997 it would
have seemed completely unimaginable that the MISTRA would end in
the near future. It was such a vital program that shaped the careers of
many colleagues and enriched the lives of many twins. “Twin Weeks”
were fascinating, overwhelming, and exhilarating. There was rarely time
for personal matters when twins were in town, but nothing else was as
important. Findings from the MISTRA had widely infiltrated the psycho-
logical field, changing our thoughts on how genes and environments af-
fect behavior. When the study ended there would be no place to send
newly reunited pairs or separated twins in search.
In his 1997 statement, Bouchard acknowledged the financial support of
the Koch Charitable Foundation, the Tufts University Nutrition and Aging
Laboratory, and the University of Minnesota Periodontal Group. Since the
first Koch Foundation grant was made in 1985, these sources were among
those that had covered the twins’ airline fees and hotel expenses. Various
medical costs, such as the echocardiograms and allergy tests, were sub-
sumed by the different departments and no one received salaries.
Federal funding was especially difficult to obtain because the number of
twin pairs was never certain. Federal granting agencies, such as the Na-
tional Science Foundation and National Institutes of Mental Health, re-
quire detailed participant descriptions, including age, gender, and minor-
ity status. This was just not possible with the reared-apart twins who
could appear at any time. “This smothers creativity,” Bouchard said. “The
private foundations were a blessing because they can afford to take risks.”
In accepting support from the Pioneer Fund support in 1981 Bouchard
took a calculated risk that the research would weather the controversies
surrounding that source, as I will discuss in Chapter 14.
The history of reared-apart twin studies remained contentious through-
out the MISTRA’s years. The Cyril Burt affair and concerns over how
gene-behavior relationships might interfere with good parenting and social
programs were foremost in some grant reviewers’ minds. Ratings Bouchard
220 B O R N TO G E T H E R — R E A R E D A PA RT

received on federal grants could be Excellent or Poor—this wide range

and the comments associated with them most likely reflected some re-
viewers’ biases to a large degree.
When Bouchard talked to me in April 2010 about ending the study
when he did, he cited lack of funding and the decreasing number of reared-
apart twins, but he also mentioned personal fatigue. “I worked eighteen
hours a day when twins were in the lab and then I slept for two days.”
“What if you had gotten a five million dollar grant?” I asked him, “Would
the MISTRA have ended when it did?” He paused, then recalled his long
conversations with Lykken over the need for a comparison group of twins
reared together and the importance of developing the Minnesota Twin
Registry. But I think we both knew that a large grant would have ex-
tended the life of the MISTRA, a child that Bouchard had nurtured for
years and did not seem quite ready to let go despite the work involved.
Whenever I tell Bouchard about a new reared-apart pair he says I should
study them.
 CHAPTER ELEVEN

“Larks” and “Owls,” Ego Development,

and Authoritarianism (1998–2002)

S ome people get up early and work best at six in the morning
(“larks”), while others like to sleep late and work into the night
(“owls”). Most people know which type they are, but don’t know
how they got that way. In 1998, Yoon-Mi Hur and Bouchard examined
“morningness” and “eveningness”—and these tendencies, like most mea-
sured traits, showed genetic effects. This finding, which occurred relatively
late into the study, was understandably less surprising than the detection of
a genetic influence on habits, fears, and medical traits that were suggested
with our earliest pairs. This shift in thinking began with the information
processing and ophthalmology findings in 1984 and 1985. Genetic effects
on speed of response and on visual acuity had been surprising and exciting
during the early stage of the study. This shift was reinforced by the MIS-
TRA’s early 1990s findings of a genetic influence on less frequently mea-
sured traits such as religious involvement and work values, and by similar
findings from other laboratories.
Fourteen scientific papers appeared between 1998 and 2002. Other
new topics examined during those years included ego development, head-
ache frequency, dietary preferences, and authoritarian tendencies. The
MISTRA investigators also reported new findings on topics studied previ-
ously, such as creativity and religiosity.

Morningness-Eveningness
When Bouchard’s graduate student Hur analyzed the reared-apart twins’
morningness-eveningness data, only one other such twin study was avail-
able.1 That 1992 investigation had found heritabilities of 0.56 for males
222 B O R N TO G E T H E R — R E A R E D A PA RT

and 0.48 for females from reared-together pairs.2 Given that the MZA
and MZT twins showed the same degree of similarity in personality and
job satisfaction, Hur suspected that the MZA twins would also show
matching wake-sleep cycles. Had this analysis been conducted when the
MISTRA began, she might not have been so sure.
Scores on the Morningness-Eveningness Questionnaire3 were available
for fifty-five MZA twin pairs, fifty DZA twin pairs, two hundred and five
MZT twin pairs, and seventy-nine of the reared-apart twins’ spouses. The
questionnaire included thirteen items, such as “At what time in the evening
do you usually feel tired and, as a result, in need of sleep?”
The MZA correlation of 0.47 was close to the values reported in the
previous study. The MZT correlation of 0.57 was somewhat above the
MZA correlation, suggesting that shared rearing environments explain
only a small amount (10 percent) of individual differences in preference
for the morning or evening.
What was surprising was the unexpectedly high morningness-eveningness
correlation (0.45) for the DZA twins. Perhaps some DZA twins’ schedules
were not initially to their liking, but they had to acclimate due to their
working conditions, family obligations, or other matters. Another possi-
ble explanation for this result was assortative mating or mating between
spouses who matched in their waking and sleeping time preferences.
Positive mate assortment for morningness-eveningness among the spouse
pairs in our sample was found, but not enough to explain the DZA twins’
resemblance.
When the data from the three twin groups were analyzed simultane-
ously, the heritability of morningness-eveningness was 50 percent, consis-
tent with the 1992 study. There was also interest in whether age affected
this trait—perhaps younger twins slept later or older twins went to bed
earlier—but age had little effect. It seems that we become “morning” and
“evening” people because of how our individual circadian system is struc-
tured, not because we saw our mother get up early or our father sleeping
late. Sleep-wake patterns also become less regular as we age, a pattern of
change that could be partly genetically controlled.
The Morningness-Eveningness Questionnaire was completed on or
about the third day of the assessment week. Perhaps the twins should have
completed this form before they arrived so we would have known when
we could expect their best performance! Assuming that the MZA and
DZA twins were equally divided among larks and owls, our failure to do
so probably did not affect the data.
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 223

Ego Development
By 1998, various investigators on the project had published several
personality studies, but not the twins’ ego development scale. This was a
timely topic because some personality researchers were examining rela-
tionships between ego development and the Big Five personality traits, as
well as relationships between ego development and cognitive skills.
Psychologist Jane Loevinger is famous for her comprehensive research on
adult ego development.4 She saw the ego as a “lens” for viewing the social
world and ego development as the changes in those views. She character-
ized it as a “master trait” that organizes many specific personality traits.
More specifically, Loevinger viewed ego development as a developmental
process that unfolds in stages during childhood, then stabilizes during ado-
lescence and adulthood. She explained that “the search for coherent mean-
ings in experience is the essence of the ego or ego functioning . . . The ego
maintains its stability, its identity, and its coherence by selectively gating
out observations inconsistent with its current state.”5
Loevinger recognized different levels of ego development. The least ma-
ture levels are characterized by presocial, impulsive, self-protective, and
ritualistic behaviors. The most mature levels are characterized by consci-
entious, individualistic, autonomous, and integrated behaviors. The modal
level of ego development, based on American samples, is characterized by
reference to the expectations of other people regarding rules for personal
conduct and social values.6 Individual differences in ego development lev-
els can be observed at all ages after fourteen, so Loevinger’s theory could
be applied easily to the adult reared-apart twin sample.
In 1994, Minnesota graduate student Denise Newman completed the
first study of reared-apart twins that addressed genetic and environmen-
tal influences on ego development.7 Newman, who worked directly with
Auke Tellegen, said, “I was interested in ego development, and ego devel-
opment had an instrumentation that could be addressed within a behav-
ioral genetic framework. It was Loevinger’s brilliance that enabled that
possibility, and it was of course Bouchard’s uncanny knack for designing
a study with lots of different measurements in it, knack (with maybe Tel-
legen’s influence) that allowed me to realize the project was possible.”
Newman also looked at whether ego development was distinct from
cognitive ability, especially verbal skills, an issue also raised by Loevinger
and colleagues.8 This was an important feature because the instrument,
the Washington University Sentence Completion Test, relies on verbal
fluency,9 a feature that could confound ego development measurement.
However, while verbal fluency tests are usually administered under timed
224 B O R N TO G E T H E R — R E A R E D A PA RT

conditions, the Sentence Completion Test is not. As Loevinger and col-

leagues noted, the valid variance in ego development should not be fully
accounted for by intelligence. Previous associations between ego devel-
opment, general ability, and the Big Five trait of Openness to Experience
led Newman to examine the extent to which ego development provided
unique information about an individual.
The Sentence Completion Test is a projective test (a method in which
subjects respond to unstructured or ambiguous terms or images)—
Loevinger believed that measuring ego development required projection
of the subjects’ own frame of reference.10 The test also comes in male and
female forms. Sample items for males and females are:

Female: A girl has a right to __________.

Male: A good father _________________.

The cognitive tests Newman used included the Wechsler IQ test and
the verbal subtests of the Hawaii Ability Battery. The sentence completion
responses were rated independently by individuals who were blinded to
the twins’ age, sex, and zygosity. The score distributions for all 248 people
(forty-five MZA twin pairs, twenty-eight DZA twin pairs, nine individuals
from DZA opposite-sex pairs, and ninety-three of the twins’ spouses, chil-
dren, and companions) were similar to those from other adult samples.11
The MZA intraclass correlation of 0.50 suggested that half the vari-
ance in ego development is genetically influenced. The corresponding
DZA correlation was 0.22. When Newman recalculated the correlations
after statistically removing the effects of verbal ability and IQ, the values
fell only slightly to 0.45 and 0.41, respectively, for MZA twins and 0.21
and 0.17, respectively, for DZA twins. Ego development could, therefore,
be considered as separate from intellect, a finding confirmed in the bio-
metrical analysis. Genetic influence on ego development was 0.54,12 and
individuals’ nonshared experiences were nearly as important.
The twins’ ego development similarity could not be explained by simi-
larities in their rearing environments. Only one measure (mothers’ educa-
tion level) correlated significantly for both MZA (ri = 0.45) and DZA
(ri = 0.56) co-twins. Selective placement can inflate resemblance between
adopted apart relatives, giving greater weight to environmental effects
than is warranted. However, mothers’ education explained less than 1
percent of the variation in ego development.
These results were exciting. Newman also concluded that the data af-
firmed the presence of the “black box” of variance.13 That is, idiosyncratic
things that happen to people explained about half the variance in ego de-
velopment. However, the specific events (e.g., going to college or losing a
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 225

spouse) that raise or lower ego development levels were unknown—hence

the uncertainty.
Newman wondered to what extent genetic and environmental influ-
ences on ego development were constant or variable across ages, a ques-
tion she could not answer with her modest sample. She also wondered
how comparable her findings were to those from other personality studies,
a question she could answer. Previous personality studies, including the
MISTRA’s, suggested that most personality traits showed a 50 percent ge-
netic effect with little shared environmental influence, as Newman found
for ego development.
Openness to Experience (imaginative, inventive versus narrow, simple),
the Big Five personality trait with the largest additive genetic component,
was the only Big Five trait that had been associated with ego develop-
ment.14 Loevinger’s modal classification spans two levels, conformist
and self-aware. Individuals at the conformist level value group accep-
tance and can delay or modify their impulses. Such individuals might tend
toward the “narrow, simple” pole of Openness to Experience. In contrast,
those at the mature, self-aware level are more individualistic and autono-
mous and might be found closer to the “inventive, imaginative” pole. The
possibility that common genes affect ego development and Openness to
Experience, particularly at different ages, would be worth pursuing.
The distinction Newman found between ego development and cogni-
tion was consistent with a meta-analytic study of findings from forty-two
investigations. Reasons to support the conceptual and functional distinc-
tion of ego development and intelligence include the higher scores given
to shorter sentence completions, the irrelevance of speed of response, and
the fact that most adults possess the vocabulary to express significant ego
development features.15
Newman sees ego development as a “compelling and central aspect” of
personality development. “There is a very extensive empirical literature
to support a very rich theoretical story . . . ego develops from infancy
and well into adulthood; something that is not fully formed at birth be-
comes so formed by early adulthood and changes still a bit more across the
lifespan.” The similarities she observed in the twins reared apart are more
impressive in light of these remarks.

The Headache Study

New studies of the twins’ physical conditions were also progressing in the
late 1990s. A study on migraine headaches came about because a specialist
was interested.
226 B O R N TO G E T H E R — R E A R E D A PA RT

The MISTRA’s collaborator on the headache study was Dr. Dewey

Ziegler from the University of Kansas Medical Center’s Department of
Neurology.16 Ziegler wrote to Bouchard in April 1989 to say that the
reared-apart twins would be informative participants in a study of factors
affecting migraine headache, and he proposed a collaboration. Bouchard
put Ziegler in touch with David Lykken who could provide reared-together
twins, and a research agreement was reached in October.
I immediately thought of the Jim twins when I learned about the
headache study because both men had similarly described their severe
migraines, which had begun in their teens. Their similar symptoms sug-
gested that headaches could have a genetic basis. The reared-apart Ko-
rean twins whom I discussed earlier also experienced headaches, albeit of
different frequency and severity from one another. However, their
matched complaints, and the fact that their biological mother had suf-
fered migraines are consistent with genetic effects.
Several reared-together European and Australian twin studies con-
ducted between 1956 and 1995 had found that genes explained 50 per-
cent of the variance in headache susceptibility.17 However, two United
States studies, including one by Ziegler in 1975, failed to find genetic
effects.18 Farber’s 1981 review had identified four out of five headache-
concordant pairs scattered among the early MZA twin investigations,19
but no one had formally studied headaches in reared-apart twins.
Ziegler’s study included twenty-three MZA, one hundred and three
MZT, twenty DZA, and fifty-one DZT twin pairs. The reared-together
twins came from the Kansas and Minnesota registries, and they had vol-
unteered in response to notices about a headache study. The reared-apart
twins were MISTRA participants, so they were not recruited with refer-
ence to headache predisposition. Only female twins were included in the
study because headache prevalence (the frequency of individuals express-
ing a trait at a particular time) was low among the male twin respondents,
a sex difference that has been well documented.20
A telephone interview was administered to each twin beginning with
the question, “Have you ever had repeated headache (apart from blows
on the head or during acute illness) that you would consider more than
mild?” If answered affirmatively, clinical details about their headaches
were requested.
Headache resemblance was higher for MZT (0.52) than DZT twins
(0.00), reflecting genetic influence. In contrast, the MZA twins (0.52)
were only slightly more alike than the DZA twins (0.46), a possible result
of the small sample. Because the MZA correlation directly estimates ge-
netic influence, it was concluded (albeit, tentatively) that 52 percent of the
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 227

headache variance was associated with genetic factors—this was the same
value Ziegler obtained by model fitting. The lack of difference in the MZT
and MZA correlations suggested that shared environments did not affect
headache frequency—instead, environmental contributions to headaches
appeared unique to individuals, possibly related to emotional stress, fever,
sun exposure, or dehydration.21
The findings applied to women, not to men, and they did not distin-
guish between headaches with and without aura (the presence of feelings
and signs that proceed a headache). Despite these limitations, a genetic
component to migraine headache was indicated, although the relatively
high DZA correlation urged a cautious interpretation. Knowledge about
factors affecting headaches could potentially benefit the thousands of af-
fected individuals attempting to make sense of their symptoms.
Headaches, both severe and mild, can have many sources, such as eating
certain foods, eating irregularly, or not eating enough. Headaches, dizzi-
ness, and joint pain are among the symptoms associated with allergic re-
action to gluten, a protein found in most grains, breads, and cereals. A
pair of our MZA male twins suffered from “irritability” and other diffi-
culties, suggesting a food allergy. In fact, these tall, thin British gentlemen
had been experiencing the same symptoms from their allergy to gluten.
However, when they arrived in Minnesota only one twin had determined
that eating wheat products was the cause. Upon becoming a vegetarian
and eliminating wheat from his diet, this twin’s health and spirit improved
dramatically. His twin brother agreed to make the same dietary changes
when he returned home.
One month later, Bouchard received a letter from one of the twins saying
that both were feeling better and were scheduled for allergy testing. The
visit of these twins was meaningful in other ways. The letter read, “So we
grow more, even more alike . . . I think that we are now quite close in our
understanding and empathy . . . Without the basis of our visit [to Minne-
apolis] I do not think that we would have ever spent sufficient time to-
gether for this to have happened.”

Meals and Snacks

There was interest in the twins’ dietary preferences, but not just because
of their possible allergies to certain foods. Eating is a fundamental and
pleasurable human behavior. Knowing whether certain foods are selected
or rejected because of genetic factors, cultural habits, family practices, body
size image, or all four were questions that the reared-apart twins could
potentially answer.
228 B O R N TO G E T H E R — R E A R E D A PA RT

Children growing up together have limited freedom to eat what they

like, but once they become adults they are free to choose. Adults can also
decide how much and how often they want to eat. Sharing meals with the
MZA twin pairs showed that favorite foods, quantities, and meal times
were often matched. The fireman twins, Mark and Jerry, both craved
Chinese food and very rare steak. The giggle twins, Daphne and Barbara,
co-invented a signature cocktail that they called “Twin Sin,” made by mix-
ing one ounce of vodka, one ounce of blue curaçao, one ounce of crème de
cacao, and a half ounce of cream—they thought it was the greatest. Oskar,
raised in Germany, was instantly enthusiastic about the spicy foods that
Jack, his Trinidad-raised brother, offered him. These observations sug-
gested genetic effects on food preferences, even the possibility that genetic
factors can override cultural food practices.
Existing findings on twins’ dietary choices were mixed when Hur,
Bouchard, and Elke Eckert began looking at the diet data in 1998.22 Their
review of the research showed that estimates of genetic influence on nutri-
ent and beverage intake ranged from about 0.30 to 0.50, and one study
had found genetic influence on complex carbohydrate intake only. Rela-
tionships between twins’ dietary similarities and the amount of contact
time between them were also in dispute. The MISTRA had the data to
revisit these questions using sixty-six MZA and fifty-one DZA twin pairs.
The twins completed the Food Frequency Questionnaire developed for
the National Heart and Lung Institute Twin Study.23 The questionnaire’s
interview format was modified to allow the questions to be answered by
self-report rather than by personal interview. The form’s sixty-seven items
requested information on the frequency of eating foods included in stan-
dard American diets (e.g., milk and cheese) and on the timing of meals
and snacks. However, twins also listed their preferred foods that were not
included in the survey. Participants’ height and weight were also obtained
in the laboratory, measures that were converted into the body mass index
(BMI = weight in kilograms/height in meters2). Age correlated only mod-
erately (0.17 to 0.20) with various dietary variables such as calories,
protein, and fat showing that older people ate more than younger people,
but not a great deal more.
Hur compared similarities between the pairs of MZA twins, DZA twins,
and spouses across three dietary categories: nine nutrients (e.g., calories
and protein), two eating frequencies (e.g., meals and snacks), and six bever-
age intake frequencies (e.g., whole milk and coffee). The MZA twins
(0.30 to 0.35) were more alike in three of the four dietary categories than
the DZA twins (0.08 to 0.22) with one exception: the DZA twins’ nutri-
ent intake showed greater similarity than that of the MZA twins (0.35
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 229

versus 0.22). Other than sampling fluctuations, this result could not be
explained. Model fitting yielded average heritabilities of 0.30 (nutrient
intake), 0.30 (meal frequency), 0.33 (snack frequency), and 0.32 (bever-
age frequency). Thus, what we eat and how often we eat appear moder-
ately influenced by genetic factors.
An important point is that certain foods must be available if both
reared-apart twins are to eat them. Oskar’s German culture did not pro-
vide the spicy foods he eventually found so appealing. Other foods seem to
have acquired tastes. An Australian twin would enjoy spreading Vegemite
or Marmite on toast,24 condiments that would probably receive a “No,
thank you” from an American counterpart who had never tasted them
and knew their origin. In a separate case study, Hur and I also discovered
a shared distaste for fish in reared-apart Korean co-twins we had studied.
Fish was less commonly eaten in the area of the United States where one
twin was raised, but it is a large part of the Korean diet. Perhaps the
smell or texture of fish were unpleasant for the Korean-raised twin, and
could be for the American twin as well. These ideas could be tested using
larger twin samples.
The diet data produced some counterintuitive results. Hur did not find
a link between body size and amount of food intake, suggesting that me-
tabolism rather than quantity consumed is largely responsible for body
weight and fatness. This finding was consistent with the outcome of a clever
1990 twin study conducted by University of Laval researcher Claude
Bouchard—no relation to the MISTRA’s director.25 In this study, twelve
MZT male twin pairs abstained from exercise for eighty-four days while
on a diet exceeding their normal intake by 1,000 calories. At the end of
the study, everyone had gained weight, but the amount gained, which
varied from 9.5 to 29.3 pounds, was more similar within twin pairs (for
co-twins) than among twin pairs. Metabolism seemed to mediate weight
gain more significantly than diet.
One might suppose that the twins’ time together influenced their di-
etary preferences, but it did not: the mean correlation between their con-
tact and the seventeen dietary measures was zero. What the twins ate, and
when, were generally products of their genes and their unique experi-
ences. These same factors appear to be at play when family members eat
out together and can order freely. There are often as many different meals
as there are diners.
A personal story told to us by one of the MZA female twins suggested
a unique experience that could affect one’s dietary habits. When the thirty-
six-year-old twin and her sister came to Minnesota for the assessment, the
twin who was five feet three inches tall and weighed 123 pounds admitted
230 B O R N TO G E T H E R — R E A R E D A PA RT

that she had always felt too heavy. But when she met her twin sister, who
stood five feet four inches and weighed 130.5 pounds, she was surprised
to find that her sister looked “great.” This experience made her rethink
her views about her own body size.

Voices
According to family and friends, my fraternal twin sister and I are nearly
indistinguishable on the telephone, so twin studies of voice quality have
been of great personal interest. Early studies showed that MZT co-twins’
voices are nearly identical with respect to fundamental frequency (char-
acteristic highness or lowness of the human voice) and fundamental fre-
quency range, although later studies showed overlap in the level of MZT
and DZT twin similarity.26 Similar tone and pitch were indicated for
twenty out of twenty-one early MZA twin pairs for whom such observa-
tions were made.27 Shields noted that voice similarities are “frequently”
detected among DZT twins and sibling pairs, suggesting a contribution
from genetic factors combined with regional and cultural influences.28
A  1995 review concluded that genetic effects clearly influence voice
quality, assuming that function is related to genetically influenced physi-
cal structures.29
Reared-apart and reared-together twins’ fundamental frequency and
range were the subjects of a 1998 master’s thesis by Michael Hammer
at the University of Kansas.30 Twins were recorded while reading Grant
Fairbanks’s “Rainbow Passage,”31 with the second sentence analyzed due
to its high correlation (0.99) with the full paragraph. Based on a relatively
small sample (forty-nine MZT, eight DZT and fourteen MZA twin pairs),
the intraclass correlations for both frequency (0.53, 0.58, 0.59) and range
(0.30, 0.43, 0.51), respectively, were quite similar. However, vocal dis-
crimination between related and unrelated pairs (generated randomly
from the twin sample) was suggested. These results contrasted with what
Farber had called the “stunningly alike” pitch, tone, and overall vocal
characteristics observed (but not studied systematically) among the ma-
jority of fifty early MZA pairs.32 Conclusions from the small sample that
included the MZA twins are tentative, at best.

Authoritarian Attitudes
Psychological studies of authoritarianism were started at Berkeley in
the 1950s by the German emigré sociologist and philosopher Theodor W.
Adorno and colleagues. These studies were intended to understand the
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 231

endorsement by many German citizens of Hitler’s fascist regime and their

participation in his anti-Semitic programs and policies. The authoritarian
personality was characterized by conventionalism, aggression, submis-
sion, cynicism, and admiration of power and strength. One conclusion of
that study, summarized in The Authoritarian Personality (1950), was that
authoritarian personalities originated with harsh, punitive, and control-
ling parents whose children projected these tendencies onto minorities or
other stigmatized groups.33 The view that authoritarianism was traceable
to rearing environments was evident nearly forty years later in Altemey-
er’s Enemies of Freedom (1988).34
The origins of authoritarianism were particularly interesting to Bouchard
who had attended the University of California–Berkeley as an under-
graduate and graduate student between 1959 and 1966. Bouchard had
been an active participant in the free speech movement at that time and
was even arrested on one occasion for his activities. At the University of
California–Santa Barbara, where Bouchard was an assistant professor
from 1966 to 1969, he and philosophy professor Merrill Ring organized
an extensive series of antiwar marches.35
Authoritarianism was also intriguing to Bouchard’s graduate student
Kathryn Corson (formerly McCourt). Corson had originally planned to
study ingroup-outgroup phenomena in nonhuman primates at Minnesota
but changed her mind because a good measure was unavailable. “But
right-wing authoritarianism seemed related, and Bob Altemeyer was so
convinced that it [the origin of authoritarianism] was all parent to child
through the environment. Many behavior geneticists also thought that
attitudes were all about rearing environment and not genetic influence,
so it was fun to take it on,” Corson recalled fifteen years later.
Few behavioral genetic studies of social attitudes were available in the
1980s. John Loehlin and Robert Nichols failed to find genetic influence
on political and religious attitudes in their 1976 landmark study of
adolescent twins.36 However, their result might have reflected the twins’
young age and exposure to their parents’ views while still living at home.
Thoughts about the roots of authoritarian attitudes showed little change
for the next ten years. Then, the 1986 Martin et al. twin study of social
attitudes drew surprised attention to findings of significant genetic influ-
ence on conservatism, radicalism, and tough-mindedness.37 That paper
prompted Bouchard to add Altemeyer’s thirty-item Right-Wing Authori-
tarianism Scale to the MISTRA’s assessment schedule,38 which he ob-
tained from David Lykken who was distributing it to his registry twins.
Bouchard also added the Wilson-Patterson Conservatism Scale39 to the
assessment, but the results were not published until 2003.
232 B O R N TO G E T H E R — R E A R E D A PA RT

The Right-Wing Authoritarianism Scale is a self-report inventory with

nine response categories ranging from +4 (very strongly agree) to −4
(very strongly disagree), with an intermediate response of 0 (exactly and
precisely neutral). Sample questions include “It is important to protect
fully the rights of radicals and deviants,” and “Obedience and respect for
authority are the most important virtues children should learn.” In 1986,
the scale was administered to new twins and to twins returning for their
follow-up visit, and mailed to twins who had been assessed previously.
Data were available for forty MZA twin pairs and forty-two DZA twin
pairs. Data were also available for 423 MZT twin pairs and 434 DZT
twin pairs.
The first authoritarianism article to come from the MISTRA appeared
in 199940 and was based on Corson’s 1996 doctoral dissertation.41 Cor-
son (McCourt during this study) began by noting that higher authoritari-
anism was significantly associated with fewer years of parental education
(reared apart: −0.36, reared together: −0.37) and lower parental occupa-
tional status (reared apart: −0.40, reared together: −0.30). Higher authori-
tarianism was also linked to lower cognitive ability (−0.37) in the reared-
apart twins; ability data were unavailable for the reared-together twins.
When these different correlations were recalculated (variously removing
the effects of ability, education, and occupational status), they dropped
somewhat but remained significant. For example, high authoritarianism
was still tied to low parental education after removing the effects of cog-
nition on authoritarianism.
The correlational pattern among the four twin groups was somewhat
unusual. The higher MZA (0.69) and MZT (0.63) than DZA (0.00) and
DZT (0.42) correlations showed genetic effects. However, the DZT twins
showed modest resemblance, while the DZA twins showed none. If we
relied on the reared-apart twin data, then the genetic effect would be 0.69
because the MZA correlation directly estimates heritability. On the other
hand, if we used the reared-together twin data, then the genetic effect
would have been 0.42 (based on doubling the MZT-DZT correlational
difference). These conflicting bits of information had to be reconciled by
way of model fitting.
Model fitting42 showed a genetic effect of 50 percent. It also showed
that unique environmental effects plus measurement error explained 35
percent of the authoritarianism variance and shared environmental ef-
fects explained 15 percent. McCourt observed that “like witnesses at a
crime scene” the four twin types provide different information about the
origins of right-wing authoritarianism. Most importantly, she showed that
the traditional view of authoritarianism, in which rearing by authoritarian
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 233

parents leads to authoritarian children, was untenable. McCourt explored

this issue further, using retrospective measures of the rearing environment
from the Block Environmental Questionnaire and Family Environment
Scales described in earlier chapters.
McCourt organized the reared-apart twins and their spouses according
to rearing by biological or nonbiological relatives. Four environmental
measures—achievement orientation, moral-religious emphasis, organiza-
tion, and control—correlated significantly with authoritarianism (0.22 to
0.35) for twins raised by biological relatives. This might suggest that
family environment factors drive authoritarian traits in children, perhaps
through parental rearing practices and/or by parental modeling. However,
the correlations between authoritarianism and the family environment
measures were negligible for the twin adoptees, showing that the family
environment was not associated with this behavior.43 Without the reared-
apart twins, the wrong conclusion (i.e., authoritarian attitudes come from
within-family dynamics) would have been reached.
The lack of social influence on authoritarianism was further shown
by the lack of association between right-wing authoritarianism and the
twins’ contact prior to assessment. Once again, Lykken was correct in his
view that psychological research with twins gets at behaviors’ underlying
causes better than psychological research with nontwins. Bouchard’s 1999
study on religious beliefs would confirm this view.

Intrinsic and Extrinsic Religiousness

In November 1998, the John M. Templeton Foundation funded a three-
day workshop in Berkeley, California, on religious beliefs. The invitees
were researchers with access to relevant but unpublished twin data.
Bouchard’s paper on intrinsic and extrinsic religiousness from that work-
shop became one of twelve contributions to a 1999 issue of Twin Re-
search devoted to twin studies of religion, values, and health.44 Profes-
sor of Human Genetics and Psychiatry and Reverend Lindon Eaves,
who authored the introduction, called the collection “a small landmark
in the study of behavioral genetics”45 because prevailing social science
views held that religious beliefs were acquired within the rearing family.
Psychology textbooks at that time were surprisingly devoid of chapters on
religiousness, given its association with personality, depression, and physi-
cal health.
Bouchard’s paper was the second MISTRA publication on religiosity;
the first had appeared in 1990.46 Bouchard saw religiousness as a signifi-
cant and complex psychological construct—higher levels predicted better
234 B O R N TO G E T H E R — R E A R E D A PA RT

mental and physical health,47 and reduced delinquency and drug abuse,
yet increased religiosity was also associated with increased prejudice and
intolerance.48
This second religiosity study focused on intrinsic religiousness, the sig-
nificant religious values that may have personal meaning (e.g., central in-
fluence of religion in one’s life), and extrinsic religiousness, the utilitarian
religious values that may enhance social status or self-concept (e.g., social
benefits of religious participation). The distinction between the two was
made in 1967 by psychologists Gordon Allport and J. Michael Ross,49
among others. Bouchard examined the roots of these religious tendencies
in thirty-five MZA twin pairs and thirty-seven DZA twin pairs. However,
this study went beyond that purpose in following Edward O. Wilson’s
view that by “traditional methods of reduction and analysis science can
explain religion but cannot diminish the importance of its substance.”50
Bouchard addressed the issues of replication, measurement, parsimony,
and heuristics, although consilience51 (i.e., unity of knowledge) was “un-
questionably relevant but beyond the scope of this paper.”52
This study was an attempt to constructively replicate the MISTRA’s
1990 finding of genetic influence on religious interest and commitment.
The reared-apart twins completed the Age Universal Religious Orienta-
tion Scale, a modified version of the Allport and Ross Intrinsic-Extrinsic
scales. The Allport and Ross scales had been used to measure intrinsic and
extrinsic religious orientations in adults. In 1983, its authors, Richard
L.  Gorsuch and G. Daniel Venable, rewrote the twenty items from the
Intrinsic-Extrinsic scales in simpler language to allow for research with
children and adolescents.53 The MISTRA introduced other changes to ac-
commodate a wider variety of religions and to maintain a consistent for-
mat across some MISTRA inventories.54 Sample intrinsic items are “I en-
joy reading about my religion” and “I try hard to live all my life according
to my religious beliefs.” Sample extrinsic items are “It doesn’t much mat-
ter what I believe as long as I am good” and “Although I am religious, I
don’t let it affect my daily life.” The Age Universal Scale was added to the
study in the late 1990s, so a relatively small number of twins had com-
pleted it by 1999. The twins’ scores on selected California Psychological
Inventory scales as a check on the social desirability of their responses on
the religiosity scale were also examined.
With regard to parsimony, Bouchard noted that religious behaviors
were rarely studied by psychologists because they assumed that such
behaviors were part of the personality domain. However, associations
between measures of personality and religious behaviors were small. The
heuristic value of studying religious behaviors comes from the potential
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 235

insights into numerous individual and societal circumstances (e.g., physi-

cal health and delinquency).
The data indicated significant genetic influence on intrinsic religiosity
(0.43) and extrinsic religiosity (0.39). The possibility that these results
could be explained by the adopted twins’ placement into families with
similar religious practices was ruled out. Knowing that social desirability
(the tendency by some individuals to answer in ways that would be seen
as desirable by others) might affect the responses, Bouchard correlated the
twins’ two religiosity scores with relevant scales from the California Psy-
chological Inventory, namely, the Good Impression and Dicken Social
Desirability and Acquiescence Scales. The only significant correlation was
between the Good Impression Scale and intrinsic religiosity, but it was
small (0.18). The twins seemed to have answered truthfully without try-
ing to convey a favorable impression.
A curious finding was that the three contact measures (time together
before separation, total contact time, and total time apart) correlated sig-
nificantly with the within-pair difference in extrinsic religiosity for the DZA
twins, but not for the MZA twins. This meant that a greater amount of
contact between the DZA twins was associated with smaller differences in
extrinsic religiosity between them. If, for example, reunited DZA co-twins
attended religious services together because they enjoyed the attention
their twinship attracted, then their contact time would increase, and their
extrinsic religiosity difference would decrease. However, none of the pre-
vious analyses had detected such associations, so this finding deserved at-
tention. A closer look at the data showed that only early contact, not later
contact, affected the DZA twins’ similarity in extrinsic religiosity. Sec-
ondly, several of our contact measures were approximations because sev-
eral pairs had completed the inventories by mail and could have had vary-
ing amounts of contact after leaving Minnesota. It was also possible that
the contact findings occurred by chance, because they had not emerged in
other analyses.
Three features of this study stand out. The first is that intrinsic and
extrinsic religiosity appeared to be distinct religious dimensions. The cor-
relation between them was only 0.04. Therefore, it is not possible to talk
about whether or not someone is “religious”; it is important to consider
specific religious behaviors, for example, those that are intrinsic and ex-
trinsic. This conclusion was supported by the modest positive correlations
between the California Psychological Inventory’s social maturity scales
and intrinsic religiosity, demonstrating intrinsic religiosity’s buffering of
individuals against difficult psychological conditions. Bouchard did not
find meaningful associations between intrinsic and extrinsic religiosity
236 B O R N TO G E T H E R — R E A R E D A PA RT

and personality traits as anticipated if religious behaviors are distinct

from personality measures. The only possible exceptions were a modest
negative association between intrinsic religiosity and Aggression, and a
modest positive association between intrinsic religiosity and Traditional-
ism, consistent with expectations based on the nature of this measure.
The second outstanding feature of this study was the finding of sig-
nificant, but modest, genetic influence on both the intrinsic and extrinsic
scales. This was further evidence against a longstanding social view of
the origins of religious attitudes and values. It seems counterintuitive that
genes would affect religiosity more than family factors, but the data tell a
more truthful story than our beliefs about how the world should work.
Genetic effects on religiosity were replicated in the study (using a differ-
ent assessment measure), but the expectation that intrinsic religiousness
would show greater heritability than extrinsic religiousness was not
confirmed. This was surprising because our 1989 study of job satisfac-
tion found genetic effects on intrinsic satisfaction but not on extrinsic
satisfaction, suggesting that the more personally significant aspects of
work activities and interests have a partial genetic basis.
The third outstanding feature of this study was the failure to repli-
cate the emergenic effect found for religiosity in the 1990 MISTRA
study headed by Waller and Kojetin. This does not mean that an emer-
genic effect was not there—it means that additional replications with
larger samples would be required before we can conclude that one was
not present.
In the ten years that followed publication of the special Twin Research
issue, twins showed up more frequently in studies conducted by politi-
cal scientists, economists, and game theorists,55 researchers who had not
thought previously about how genetic factors might affect their measures.
Twins provided the perfect opportunity for them to test their new ideas
with new data.

Creative Accomplishment
Creativity involves considerable mental effort and sacrifice, behaviors
that vary across individuals. The underpinnings of creative activity have
interested psychologists for some time, although possible genetic effects
have been largely overlooked. Plomin’s 2008 behavioral genetics text-
book briefly reviews just ten twin studies of creativity that show very
modest genetic effects. Possible overlap between creativity and general
intelligence has been held responsible for the genetic effects on creativity
in some cases.56
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 237

Having access to extensive data from the MISTRA’s reared-apart

twins enabled the extraction of subsets of information that could an-
swer interesting questions. This approach was discovered serendipitously
by Bouchard, given the wide array of inventories that were administered.
One question was, “Are the correlates of creativity and high levels of ac-
complishment genetic in origin?” In an address delivered to the 1995 Na-
tional Research Symposium on Talent Development, Bouchard presented
reared-apart twin data that provided an affirmative answer, work that
was subsequently published in 1999.57
Bouchard began with Frank Barron’s well-known 1965 work on cre-
ativity that listed personality traits characteristic of productive scientists,
based on nine different research programs.58 Barron had also completed
twin studies showing genetic influence on verbal measures of creativity
and esthetic abilities, but not on esthetic preferences.59 Bouchard repro-
duced Barron’s list, adding the MISTRA’s measure of each personality
trait, as shown in Table 11-1.

Table 11-1. Personality Traits of Productive Scientists and the Corresponding MISTRA Measure

Barron’s Trait MISTRA Measure

Ego strength Ego Strength Scale1

Emotional stability Negative emotionality (factor)2
Independence Independence Scale3
Control Self-control3
Liking for abstract thinking; drive for Thinking vs. feeling4
comprehensiveness in explanations
Personal dominance and forcefulness of Social potency2
opinions
Nonconformist thinking Traditionalism2
Preference for dealing with things other Holland Investigative Theme5
than people

Note: MISTRA measures come from the following sources: 1Minnesota Multiphasic
Personality Inventory; 2Multidimensional Personality Questionnaire; 3California
Psychological Inventory; 4Myers-Briggs Type Indicator; 5Strong-Campbell Interest
Inventory. The MISTRA did not have measures for “special interest in pitting oneself
against the unknown, so long as one’s own efforts can be the deciding factor” and “liking
for order, method, [and] exactness together with an interest in the challenge presented by
contradictions, exceptions and apparent disorder.”
Source: Adapted from Bouchard and Lykken, “Life Achievement in a Sample of Twins
Reared Apart: Estimating the Role of Genetic and Environmental Influences,” in Talent
Development III. Proceedings from the 1995 Henry B. and Jocelyn Wallace National
Symposium on Talent Development, ed. Nicholas Colangelo and Susan G. Assouline,
81–97 (Scottsdale, Ariz.: Gifted Psychology Press, 1999).
238 B O R N TO G E T H E R — R E A R E D A PA RT

The findings were remarkable. Nearly every MISTRA creativity mea-

sure showed genetic influence in the 0.40 to 0.50 range, based on the MZA
correlations. In contrast, the DZA correlations ranged from −0.05 to
0.38. Several traits, such as Thinking versus Feeling, appeared to be
emergenic, given a high MZA correlation (0.57) and low DZA correla-
tion (−0.01). These results agreed with those in the 1993 creativity paper
reviewed in Chapter 9.60 They also added to the extant literature showing
a wide range of social environmental factors, such as level of organiza-
tional encouragement and emphasis on teamwork, that may enhance or
impede creativity.61 Challenging and supportive workplace environments
can increase creative activity, but individual differences among employ-
ees require other, possibly genetically based explanations.
Genetic influence was also evident on four other creativity measures
that were used, namely the Creative Temperament Scale (from the Cali-
fornia Psychological Inventory), the Creative Personality Scale (from the
Adjective Check List), the Esthetic Judgment Scale (from the Comprehen-
sive Ability Battery), and the Holland Artistic Theme (from the Strong-
Campbell Interest Inventory).
Bouchard found an absence of genetic influence on the Barron-Welsh
Art Scale, an instrument that assesses artistic sophistication by having sub-
jects indicate “like” or “dislike” of thirty-two different pen and ink draw-
ings. Interestingly, several previous studies of young twins had reported
higher DZT than MZT correlations, findings that Bouchard called “surpris-
ing and paradoxical.”62 Certainly, the MZA twins’ observed similarities in
home decoration and personal adornment, as well as on the creativity
measures, made the art scale findings hard to believe. I also recall inter-
viewing MZT male co-twins who paint pictures together but cannot re-
call which twin had done what in the completed works.
Barron also showed that several Minnesota Multiphasic Personality In-
ventory (MMPI) scales were high among creative writers, exemplifying the
assumed link between tendencies toward psychopathology and talent. But
he never considered whether these scales were genetically based. In the
MISTRA, four scales from the MMPI (psychopathic deviation, psychas-
thenia, schizophrenia, and hypomania) showed substantial genetic effects,
and a fifth (hysteria) showed moderate genetic effects. Perhaps assumed
links between touches of “madness” and talent have some basis in fact.
It seems that psychologists have come full circle with respect to their
take on creativity and talent. Research in this area actually began with
Sir Francis Galton’s nineteenth-century inheritance perspective. In his
1864 book Hereditary Genius, Galton tracked the accomplishments of
fathers and sons via their reputations, and concluded that eminence runs
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 239

in families.63 He came to the same decision when he compared the bio-

logical and adoptive sons of eminent figures—the abilities of adoptive
sons did not measure up to those of their adoptive fathers as was true of
biological sons. Galton eventually convinced his initially skeptical half-
cousin, Charles Darwin, that “genius” was transmitted biologically from
parent to child and was not just the product of hard work.64 But psychol-
ogy moved away from this view for many years with the introduction of
Watson’s behaviorist perspective in the 1930s. Today, no one would say
that creative talents do not require nurturance, but most would agree
that there are vast individual differences in potential skills. Some people
have to work harder to achieve the same goals.
Bouchard wished that he had included some standard creativity tests,
calling their omission “a clear mistake,” even while he did not find these
tests completely convincing. “It would have given us greater representa-
tion of so-called good creativity tests.”

Personality and Psychopathology

A personality inventory that was a first for reared-apart twins was the
Myers-Briggs Type Indicator.65 The Myers-Briggs measures four bipolar
dimensions derived from Jungian personality theory: Extraversion versus
Introversion, Sensing versus Intuition, Thinking versus Feeling, and Judg-
ing versus Perceiving. The MISTRA’s 1998 estimates of genetic influence
on these dimensions ranged from 0.40 to 0.60, consistent with what
others had found.66 Possible emergenic effects for Extraversion versus In-
troversion and Thinking versus Feeling were also observed. If replicated,
these findings would predict low levels of parent-child similarity even
though these traits are highly heritable, because parents and children do
not share the complex gene configurations that underlie emergenic traits.
The reared-apart twin sample grew and developed over time. It was
always interesting to compare findings from the different-sized samples
that were available at different stages of the study. Obtaining comparable
results could offset criticisms that the small samples precluded meaning-
ful findings. For example, the California Psychological Inventory, first
analyzed in 1990, included forty-five MZA twin pairs and twenty-six
DZA twin pairs. These data were reexamined in 1998 using additional
twins, for a total of seventy-one MZA twin pairs and fifty-three DZA twin
pairs, as well as ninety-nine MZT twin pairs and ninety-nine DZT twin
pairs.67 The earlier and later analyses yielded average heritability esti-
mates of 0.50 and 0.46, respectively, for the primary scales (e.g., Sociabil-
ity, Dominance, and Communality). The later study also examined the
240 B O R N TO G E T H E R — R E A R E D A PA RT

Vector Scales (Internality, Norm-Favoring, and Self-Realization) and Spe-

cial Purpose Scales (e.g., Work Orientation, Leadership Potential). Both
studies showed that family rearing factors and contact between twins
had little influence on the adult personality measures. The MZA twins
were as similar as the MZT twins.
More important than within-study replication was between-study rep-
lication. The California Psychological Inventory scale heritabilities were
slightly higher than those reported by Nancy Pedersen for the Swedish
reared-apart twins studied at the Karolinska Institute,68 but the reasons
were not obvious. However, the MISTRA’s findings were consistent with
findings from large-scale reared-together twin studies, such as those re-
ported by psychologists Lindon Eaves, Kerry Jang, and John Loehlin.69
Across-study consistency should have quelled criticisms that the reared-
apart twins were unrepresentative or aberrant.
The MISTRA also produced some first-time analyses of psychopathol-
ogy measures. One was a 1999 heritability study of the MMPI’s Harris-
Lingoes and Subtle-Obvious Subscales.70 The Harris-Lingoes subscales
were created by Robert E. Harris and James C. Lingoes in 1955 to over-
come the overlap among six of the MMPI’s multidimensional clinical
scales (i.e., Depression, Hysteria, Psychopathic Deviate, Hypomania, Para-
noia, and Schizophrenia) by organizing them into smaller subscales. These
subscales were intended to improve the interpretation of specific behav-
ioral and emotional problems linked to increases in the clinical scales. For
example, Hypomania was divided into Amorality, Psychomotor Accelera-
tion, Imperturbability, and Ego Inflation.
The Subtle-Obvious subscales were developed by Daniel N. Wiener
and Lindsey R. Harmon in 1948 because some items seemed to discrimi-
nate among diagnostic categories but might not seem psychopathological
to the person completing them, hence the “subtle” and “obvious” scales.71
Wiener and Harmon reasoned that someone magnifying their problems
might overendorse “obvious” items as compared with “subtle” items.
Similarly, someone hiding or downplaying their problems might not en-
dorse the obvious items, but would overendorse the subtle ones.72 This
new scheme involved the same scales reorganized by Harris and Lingoes,
with the exception of Schizophrenia.
Although the Subtle-Obvious subscales were intended to improve di-
agnostic efficacy, their usefulness has been questioned. In 1999, psychol-
ogist David DiLalla reasoned that if these subscales were substantive
then genetic variance would be expected.73 DiLalla, who had taken the
lead in our first MMPI study in 1996,74 also took the lead in this second
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 241

one. According to one of his coauthors, Greg Carey, this study was im-
portant for clinicians to know about.
DiLalla’s analysis included the same 111 twin pairs (sixty-five MZA
and fifty-four DZA) as in the 1996 study. The heritabilities ranged from
0.23 to 0.61 for the twenty-eight Harris-Lingoes subscales. However, the
surprise was finding genetic influence on the five Obvious subscales (0.37
to 0.56) and on four of the five Subtle subscales (0.27 to 0.35). Genetic ef-
fects were surprising because previous studies had shown that the Subtle-
Obvious scales did not meet their goal of improved diagnostic discrimi-
nation and showed little relationship with other relevant behavioral
scales. However, DiLalla was convinced that genetic effects on the Subtle-
Obvious scales made these scales meaningful, but perhaps not in ways
that had been supposed. He recalled that in 1974, the late psychiatric
geneticist Seymour Kety responded to Thomas Szasz’s view of schizo-
phrenia as mythical by stating, “If schizophrenia is a myth it is a myth
with a strong genetic component.”75 DiLalla applied this same reasoning
to the MMPI’s Subtle scales.
DiLalla and I had a great time exchanging stories about the MISTRA
during a telephone conversation in July 2010. He said that this 1999 paper
was one of his favorites:
The goal was really psychometric; it was not a twin thing. Behavior gene-
ticists had been accused of heritability hang-ups, but we used the twin method
to investigate substantive questions. But I am humble about the impact of
this work—it would take other, larger samples with more validity markers
to replicate our findings. But the MISTRA twins were a fantastic resource. I
greatly appreciated Bouchard’s willingness to make the data available to a
young investigator—it had a strong and positive influence on my profes-
sional development.

As the sample grew, it was clear that the study’s initial goal—determining
whether the twins’ life history differences were linked to current behav-
ioral and physical differences between them—did not cover all the types
of investigation that were possible with the data. In other words, the
MISTRA data could address many intriguing behavioral questions be-
yond those indicated in its “mission statement.” Some interested psychol-
ogy faculty and students who were not directly involved with the MIS-
TRA “borrowed” the data to test ideas of their own. A paper by Robert
(Bob) Krueger and his former student Kristian Markon was a good ex-
ample of this kind of collaboration.
Krueger was first exposed to the MISTRA in the early 1990s as an
undergraduate student at the University of Wisconsin. He took a class
242 B O R N TO G E T H E R — R E A R E D A PA RT

from Dr. Al Harkness, a visiting professor who had graduated from the
University of Minnesota. “The class had a Minnesota flavor,” Krueger
recalled. “And the MISTRA was hard to miss if you were interested in
individual differences.” Krueger met Bouchard in 1998 when he arrived
in Minneapolis for his job interview. “We just hit it off. I admired his
work, but I was a bit intimidated by him at the time.” Krueger got the job
and loved the collaborative spirit that the department offered. The idea
for a paper on personality and psychopathology came about through a
casual conversation after class with Krueger’s student Kristian Markon.
“We knew that Bouchard had the data we needed.”
Markon and Krueger used Tellegen’s Multidimensional Personality
Questionnaire or MPQ (personality) and the MMPI (psychopathology)
data in creative ways. Previous research had shown that normal and ab-
normal behaviors are correlated, but no one knew why. Markon and
Krueger found that associations between normal and abnormal person-
ality traits resulted mostly from shared genetic factors and partly from
shared environmental effects.76 For example, they found that genetic cor-
relations between paranoia (MMPI) and Achievement (MPQ), and be-
tween Social Introversion (MMPI) and Well-Being (MPQ) were 0.53 and
−0.47, respectively. This research encouraged future study of why person-
ality and psychopathology are related, looking at nervous system pro-
cesses that may be common to both.
Such studies are vital to progress in the personality field and were pos-
sible because Bouchard shared his data (albeit carefully) with interested
colleagues. Of course, those of us who worked directly with the twins
saw the data come alive in ways that it never could for those seeing it
only on a computer screen. In 2010, I asked Katie Corson, the author of
the authoritarianism paper, to come up with an interesting memory or
observation from her time working on the MISTRA. She remembered giv-
ing the University of Pennsylvania Smell Identification Test to twins.77
This is a scratch-and-sniff test that presents a series of odors with four
alternative choices for identification. (I had used this smell test in research
with reared-together twins in California, and decided that reared-apart
twin data would be a great complement to my work—another example
of wanting to study “everything possible” with the reared-apart twins.)
The interesting question for me was whether reared-apart twins would
show the same level of resemblance in odor identification as the reared-
together twins who, presumably, had been exposed to similar odors dur-
ing their lifetime.
Corson recalled, “One [reared-apart] twin would say, ‘Ah, baked bread.
That brings back such fine memories!’ And then I’d be testing the other
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 243

twin [independently, in a separate room] and when she came to the same
[item] she said, ‘Lovely! Is there anything that stirs the mind like the smell
of baked bread?’ I don’t even think bread was the ‘correct’ answer.”

Periodontology: Not All Genetic

Most journals do not welcome negative findings, but negative findings
may be informative. After our colleagues Bryan Michalowicz and Bruce
Pihlstrom found genetic influence on periodontal measures78 and on al-
veolar bone height79 in 1991, they failed to find genetic effects on peri-
odontal bacteria and temporomandibular joint symptoms in 199980 and
2000.81 The few available twin studies of these traits generally did not in-
dicate genetic effects, but the sample sizes were small, some subjects were
young, and questionnaires sometimes replaced clinical examinations. Mi-
chalowicz and Pihlstrom, therefore, believed that methodologically supe-
rior studies of periodontal bacteria and joint symptoms using reared-
apart and reared-together adult twins were worth doing, but they were
not terribly surprised by their findings.
The analysis of periodontal bacteria involved looking at dental plaque
samples for the presence of Prevotella intermedia, Porphyromonas gingi-
valis, Aggregatibacter actinomycetemcomitans, Eikenella corrodens, and
Fusobacterium nucleatum. The sample included twenty-one MZA, seven-
teen DZA, eighty-three MZT, and forty-eight DZT twin pairs. Most twins
showed the early stages of periodontal disease, and oral bacteria were
frequently detected. However, co-twin concordance for particular bacte-
rial species was low in both reared-apart and reared-together MZ and
DZ twin pairs. And the MZA twins were as similar as the MZT twins, as
they often were on psychological measures. The overall message of the
periodontal study was subtle but powerful: individuals will not develop
periodontitis in the absence of bacteria, but disease development is not
inevitable in its presence. Most remarkable was that shared family envi-
ronments do not affect the presence of oral bacteria in adults—instead,
random events unique to each individual appear to play the biggest role.
These findings contrasted with those from the 1991 study, showing ge-
netic influence on clinical and radiographic measures of periodontal dis-
ease. Therefore, other genetic factors possibly linked to the immune sys-
tem had to explain the twins’ similar periodontal disease susceptibility.
Temporomandibular joint symptoms include joint-area pain, joint
noises, tooth clenching, and tooth grinding. Slightly more than one-fourth
of twins in the 242 pairs that were studied (forty-eight MZA, thirty-five
DZA, ninety-eight MZT, and sixty-one DZT) had experienced one sign
244 B O R N TO G E T H E R — R E A R E D A PA RT

or symptom, most commonly joint noises. These data were based on a

clinical evaluation and a medical/dental questionnaire. The MZA twins
were as alike as the MZT twins across measures, showing once again that
shared family environments did not play a major role. The heritability esti-
mates, based on the MZA correlations, ranged from negligible for clench-
ing (−0.01) to low for arthritis (0.29). Thus, joint symptoms appear to be
nonfamilial.
Both sets of findings seemed surprising, even though Michalowicz and
Pihlstrom were not surprised. It made sense that MZA and MZT twins
should harbor more similar types of oral bacteria than DZA and DZT
twins. And given our previous genetic findings on dental structures and
personality traits, I expected the oral expression of nervous habits, such as
tooth clenching and tooth grinding, to show genetic effects. But the data
told a different story in both cases.
Pihlstrom maintained that the findings were interesting and added to
our knowledge base. Michalowicz suggested that this research was ig-
nored by other investigators—the negative genetic findings did make it
into the journals, but the data did not dazzle their colleagues as had the
1991 findings of genetic influence on periodontal disease measures. I sus-
pect that if they had found genetic influence on bacteria and/or joint dis-
order, their work would have received greater notice.
Many behavioral science colleagues and critics may not have known
about the periodontal studies and the lack of genetic influence on some
measures. That is understandable because psychologists are unlikely to
conduct searches in the dental literature. But that was unfortunate be-
cause the MISTRA investigators were often accused of looking for genetic
factors, emphasizing the twins’ similarities and ignoring their differences.
This was not the case—we were interested in results of any kind on any
topic that was studied. We did not decide how the data turned out, the
twins did.

1998–2002
The last annual report Bouchard prepared spanned the years 1995 to
1997, probably because after 1999 the only funding sources were private
donors and the Whitfield Institute, which ceased to exist in 2005. Further-
more, the last twin assessment took place in April 1999. Bouchard began
dividing his time between teaching and research at the University of Min-
nesota, and skiing down the slopes of Steamboat Springs, Colorado. He
retired fully in 2009, at which time he relocated permanently to Steam-
boat Springs. However, MISTRA activities continued even after the last
 “ L A R K S” A N D “ O W L S ,” E G O, A N D A U T H O R I TA R I A N I S M 245

twin pair went home. Bouchard created files documenting each twin
pair’s discovery and testing histories, and organized the hundreds of pho-
tographs that had been taken of the twins. His students processed data
from many tests and inventories that had not been analyzed. Research
team members published findings on social relationships, mental ability,
personality, sexual development, and moral values, to name a few.82
 CHAPTER T WELVE

Twin Relationships, Social Attitudes,

and Mental Abilities (2003–2005)

O utside a shopping mall, thirty-eight-year-old Marty and

Stephanie stared at each other in a bewildered daze, touching
each other’s faces and hair as if they were their own. After sev-
eral minutes they joined hands, jumping up and down with glee. Thirty-
four-year-old Jim and Trent exited their cars on a stretch of Texas high-
way. They approached one another slowly, shook hands, asked the other
simultaneously how they were “after all these years,” then started laugh-
ing. Reunited identical triplets Bob, Dave, and Eddy met one another on
two occasions when they were just nineteen—first Bob and Eddy, then
Dave and Bob and Eddy. They described their moments of meeting as
“ecstatic shock.”1
I have videotapes of the first reunions between MZA twins Marty and
Stephanie, Jim and Trent, and several other sets. I never tire of watching
these tapes or hearing their stories. The first jubilant moment when the
twins recognize each other is overwhelming. Reunions of separated twins
have also grabbed public attention because they appeal to the deep-seated
sense that family members belong together.
What people call “the twin bond”—identical twins’ extraordinary
understanding, intimacy, and rapport—most likely comes from the twins’
perceptions of their matched abilities, personalities, and temperaments.
This view comes from studies of reared-together twins, set within differ-
ent theoretical perspectives and based on twins of varied ages and gender.2
These studies consistently show closer MZ than DZ twin relationships, as
would be predicted from MZ twins’ greater physical and behavioral re-
semblance. Such findings have begun to affect decisions about separating
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 247

young adopted twins, compensating a twin for a co-twin’s wrongful

death, and even deciding if identical twins’ identical test scores reflect
independent efforts or cheating.3
The three early MZA twin studies did not systematically assess the
nature and quality of the twins’ evolving relationships. The investigators
did, however, append wonderfully descriptive accounts of the twins’ life
histories to their quantitative findings, some of which included informa-
tion about the twins’ relationships. They did this because they were in-
trigued by the human interest aspects of their cases as well as by the sci-
ence. Newman et al. claimed that their nineteen pairs would provide
writers with plenty of material for good stories.4
According to Newman et al., MZA twins Edith and Fay, age thirty-
seven, “had become very fond of each other and enjoyed impersonating
each other.”5 Betty and Ruth, age twelve, “were interrupted while dancing
on a table top in one of the examination rooms . . . In all these high-spirited
performances Ruth seemed to be the leader, but Betty was entirely co-
operative.”6 Shields’s twins Madeline and Lilian, age thirty-six, “bought
themselves identical dresses when they next met, as if they wanted to make
up now for the pleasures of twinship which had been denied them.”7 Juel-
Nielsen’s twins Palle and Peter, age twenty-two, said, “We saw something
very familiar in each other, and this familiarity was almost ‘weird.’ ”8
Forty of the fifty-four early MZA twin pairs for whom social relationship
information was available grew close to one another after meeting. This
was evidenced by their frequent contact following reunion, satisfaction
with their relationship, and other factors. Twins from fourteen of these
early MZA pairs did not become close, and information was vague or
missing for the remaining twenty-two sets. But the finding that nearly 75
percent of the pairs had bonded closely indicates that the length of shared
time does not necessarily decide how twins’ relationships progress. Unfor-
tunately, comparable information on DZA twin pairs was unavailable.9

Twin Relationship Survey

It was apparent to the MISTRA researchers that meeting a twin was a
significant, life-changing event for our participants, regardless of their
zygosity. Most twins were adoptees and were delighted to hear new details
about their biological family, and to acquire in-laws, nieces, and nephews
about whom they had not known. They were also excited to learn more
about their medical life histories. My graduate studies, which combined
twin research and evolutionary psychology, convinced me that a formal
study of MZA and DZA twins’ social relatedness would be meaningful
248 B O R N TO G E T H E R — R E A R E D A PA RT

on both theoretical and practical levels. Aside from testing the hypothesis
that MZA twins would show closer social relations than DZA twins, the
findings could help us and the twins understand their feelings with re-
gard to meeting and being with one another.
The hours spent with the Minnesota twins were great opportunities to
watch their relationships unfold. The MZA twins generally showed a bet-
ter “fit” than the DZA twins, as if they had known one another all their
lives. Their common understandings and shared laughter came more eas-
ily for them than for the DZA twins, revealing a greater sense of “we.” But
it was important to systematically record and analyze these observations
to give them scientific significance. When I arrived in Minnesota in 1982,
I asked Bouchard if a protocol was in place to capture the twins’ relations
with one another. He said there wasn’t, but he encouraged me to create
one. The result was our fifty-two page Twin Relationship Survey, first ad-
ministered in 1983.
The survey appeared initially as a thick stack of green and white fan-
fold computer paper, then acquired a leaner look as a spiral-bound book-
let with a pink cover. The number of questions grew after the first few
administrations because the twins volunteered information on topics we
had yet to consider, such as how the twins’ adoptive siblings felt about the
newly found twin. The survey’s final version contained sections on the
twins’ rearing-family situations, satisfaction in the rearing home, search for
biological relatives, and meetings with the twin. Some twins completed this
survey during their follow-up visits, and several answered it by mail.
Most twins genuinely enjoyed this part of the study because it connected
meaningfully to why they were participating in the project. They loved
learning about their similarities and differences on the behavioral and
medical tests, but their relationship with their twin touched them on a
deeper level.
The Twin Relationship Survey posed two key questions concerning the
closeness and familiarity that twins felt toward their co-twin. Each ques-
tion was answered with reference to the time of their first meeting (re-
called) and the time of their study participation (current), for a total of
four questions. Their answers ranged from (1) greater than best friends to
(6) less than someone I meet for the first time.
Twins also answered two questions with reference to the current close-
ness and current familiarity they felt toward any unrelated siblings with
whom they were raised. This set of questions was one of the most infor-
mative features of the study.
Bouchard grumbled about how much file drawer space the survey con-
sumed, so I understood his delight when he finally mailed them to me in
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 249

California. The surveys now occupy considerable space in my laboratory,

but I am grateful for the volume of information we collected because I
have used these data many times.
In 2003, twenty years after creating the Twin Relationship Survey, I
wrote the paper I’d always wanted to write—a study of social closeness
and familiarity between reunited twins.10 The analysis used data from
forty-four MZA twin pairs, thirty-three DZA twin pairs, and seven indi-
vidual twins. Based on Hamilton’s inclusive fitness theory (reviewed in
Chapter 9), I expected the MZA twins to express significantly greater
closeness and familiarity than the DZA twins, and they did. The current
ratings for both MZA and DZA twins exceeded their initial ratings, prob-
ably because the twins had had a chance to be together and become closer
and more familiar over time. The twins’ closeness scores exceeded their
familiarity scores, both when they met and currently, but the familiarity
that they felt for one another also increased with time.
The twins’ answers were anticipated, but fascinating still—the differ-
ences in ratings between the MZA and DZA pairs were greatest at the
highest levels of closeness and familiarity (i.e., closer/more familiar than
a best friend). Furthermore, the proportion of DZA twins endorsing the
highest level of current closeness (at the time of testing) was less than the
proportion of MZA twins endorsing the highest level of initial closeness
(at the time of reunion). Some DZA twins took longer to achieve the high
level of closeness that many MZA twins experienced almost immediately.
It is unlikely that this was because the reunited MZA and DZA twins
had different expectations about how they would get along—aside from
researchers, many people place importance on twinship rather than on
twin type.11 That is, they equate “twinness” with “sameness” or “close-
ness,” but this view will not make DZ twins alike or socially close.
The four social relatedness measures showed negligible or slight as-
sociations with the six contact measures (e.g., days before separation or
total contact time), except for two small but statistically significant cor-
relations. Current familiarity showed a 0.16 correlation with days before
separation, but the relationship between them was in a counterintuitive
direction, that is, more time together was associated with reduced famil-
iarity. Time from reunion to assessment showed a 0.21 correlation with
current closeness, indicating that twins studied soon after their first meet-
ing were closer than those studied later. However, many factors (e.g., the
twins’ work responsibilities and our work obligations) affected the sched-
uling, thereby reducing the meaningfulness of this finding.
A wealth of information lies in the simple contrasts between people
who vary in genetic and environmental relatedness. Comparing the social
250 B O R N TO G E T H E R — R E A R E D A PA RT

closeness between full siblings and half-siblings, or between half-siblings

and step-siblings can say something about whether genetic relatedness
plays a role. The reared-apart twins’ unrelated siblings were true research
treasures in this regard. It turned out that the twins’ current closeness
and familiarity ratings were higher for the co-twins they had recently met
than for the adoptive siblings they had always known. These results were
more striking than the twin comparisons because they were not what
logic would suggest. It makes more sense for lifelong siblings to be closer
than reared-apart twins meeting for the first time, but the data held a dif-
ferent message, one that agreed with evolutionary predictions.
The next step was to figure out the reasons behind the MZA-DZA twin
differences in social relatedness. It was expected that similarities in the
twins’ personality traits, interests, values, and/or education would be tied
to their social relatedness, but that was not the case. Perhaps the twins’
perceptions of their similarities, rather than their objectively measured
self-reported traits, triggered the levels of connectedness they expressed. A
1991 study showing that relatives’ perceived personality similarity pre-
dicted the degree of support and conflict between them affirms this pos-
sibility.12 My former University of Chicago thesis advisor, the late Daniel
G. Freedman, put it best when he suggested that recognition of common
features may foster a “sense of ‘we’ between ourselves and our fellow tribes-
men. Recognition of this sense triggers a series of emotions whose net
effect is tribal unity and the increased chance for altruism.”13
Since the MISTRA ended, my colleagues (including Bouchard) refer
newly reunited reared-apart twins directly to me. University of Minnesota
psychology professor Mike Miller put me in touch with Nina, a twenty-
four year old Korean-born MZA female twin raised in the United States.
Nina met her twin sister for the first time in Seoul at age twenty-one. I
heard the details of their reunion at Between, a trendy restaurant in the
Itaewon section of Seoul, when I visited South Korea in 2010. The twins’
first meeting occurred in a taxi cab that took the twins from the univer-
sity campus to the train station—the American twin, who was taking
classes, first had to spot the cab that carried her sister. En route to their
birth city of Busan, the Korean-raised twin indicated that she wanted to
hold her sister’s hand and then “cuddle together” while they fell asleep.
The American-born twin said that while it would be strange to cuddle
with someone she’d met for the first time, it felt normal and natural with
her twin.14
This fascinating remark beautifully captured our fundamental human
tendency for social closeness with someone like us. I wish I could have
worked it into the survey.
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 251

Personality and the Family Environment

Many MISTRA’s later analyses built upon earlier ones. By 2003, links be-
tween the articles examining the twins’ responses on the Family Environ-
ment Scale were apparent. In 1990, Bouchard and McGue found that
genetic factors had a modest effect (.31) on the recollection of family co-
hesion (how much family members were concerned, committed, and help-
ful toward each other). They also showed that about 25 percent of the
variance in consenuality in adulthood (how much a person sees the world
as others see it, as measured by the California Psychological Inventory)
could be explained by cohesion or conflict (openly expressed anger) in the
childhood home. More consensual adults came from more cohesive homes.
In 1995, Hur and Bouchard, using an enlarged reared-apart twin
sample, found significant genetic influence on memories of family cohe-
sion (0.35, based on the Family Environment Scale and 0.44, based on
the Block Environmental Questionnaire). However, they found relatively
little genetic influence on the recall of organization (importance of struc-
ture and planning) or cultural orientation (level of political, intellectual,
and cultural activities) in the childhood home. Then, in 2003, Krueger,
Markon, and Bouchard tested the idea that personality heritability might
explain the heritability of recalled family environments.15 That is, maybe
happier people are more likely to remember pleasant than unpleasant
childhood events. Since the family environments of reared-apart twins
are not shared, this study also let them see if the co-twins’ different fam-
ily environments (the nonshared variance, in this case) explained their
current personality differences.
Beginning with psychologist David Rowe’s 1981 twin studies, a recur-
rent and provocative behavioral genetic finding was that recollections of
the family environment are under partial genetic control.16 The difficulty
was that no one knew how these recollections were influenced. Krueger
wanted to figure this out. He used a range of environmental measures (i.e.,
the Family Environment Scale, Minnesota-Briggs History Record, Block
Environmental Questionnaire, and Physical Facilities Questionnaire) and
organized the scores into two factors (Status: socioeconomic and intellec-
tual/cultural status; and Cohesion: cohesion versus conflict in the rearing
home). The personality measures were the three Multidimensional Per-
sonality Questionnaire higher-order factors (Positive Emotionality, Nega-
tive Emotionality, and Constraint). Data were available for fifty-two
MZA twin pairs and thirty-eight DZA twin pairs.
As expected, genetic influences on family environment measures were
explained by genetic influences on personality traits. For example, genetic
252 B O R N TO G E T H E R — R E A R E D A PA RT

influence on constraint (emotional control) and on lack of negative

emotionality (not feeling aggressive or alienated) explained 40 percent
and 60 percent of the genetic effects on recalled cohesion, respectively.
Thus, genetic factors predisposing people to control their emotions and to
lack negativity are the same genetic factors that lead them to recall, or
believe they recall, a cohesive childhood family environment. Krueger also
found that positive emotionality explained 100 percent of the genetic ef-
fects on status (childhood family standing). That is, genetic factors underly-
ing positive emotions predispose people to recall family environments as
having been high in socioeconomic, intellectual, and cultural status.
Krueger’s other key result was that the co-twins’ differing perceptions
of their family environments had little impact on current personality dif-
ferences between them. In other words, if one MZA twin recalled a low-
conflict home and his or her co-twin recalled a high-conflict home, these
different perceptions were not associated with the twin’s differences in
emotionality. But the twin’s personalities were not exactly alike, so their
differences had to come from events that they did not share. I was in-
trigued by Krueger’s proposal that there may be a “personality set-point,”
analogous to Lykken’s happiness set-point discussed previously.17 If so,
minor daily events, such as catching a bus or losing a bet, might cause
temporary fluctuations around people’s characteristic positivity, negativ-
ity, and constraint levels. Perhaps we caught some twins on a day when
such minor events had occurred.
I liked this study because of its simple design and sophisticated reason-
ing. (It is also one of Krueger’s favorites.) Krueger also included an in-
sightful discussion of the objective and subjective elements of the twins’
answers. This discussion was necessary for making sense of how heritable
personality traits could explain heritable family recollections. Answers to
personality and rearing questionnaires reflect both subjectivity and objec-
tivity. However, if the twins’ responses were more subjective, then perhaps
their personalities affected the way they recalled their childhood homes.
Happy, upbeat twins might have focused on pleasant events, whereas sad,
depressed twins might have remembered the bad times. Alternatively, if the
twins’ answers were more objective, they might have actively influenced
the nature of their rearing homes. Twin peacemakers might have promoted
congenial home atmospheres, whereas twin troublemakers might have
created conflict.
Krueger’s was an earnest attempt to uncover links between different ge-
netically influenced behaviors. Further attempts at constructively replicat-
ing his findings are, hopefully, forthcoming.
More recent to this writing, Krueger has focused on how unusual cir-
cumstances in people’s lives can impact genetic and environmental effects
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 253

on personality.18 Reports of either very high or very low parental conflict

by adolescent reared-together twins showed that their positive emotions
were affected by their shared environment. This finding is important, but
it must be reconciled with findings of personality resemblance in some of
the adult reared-apart twins who were raised in especially different and
unusual circumstances. Active genotype-environment correlation—the
seeking out of environments compatible with inherited characteristics—
probably best explains the twins’ matched personality traits. Resiliency, in
particular, explains why MZA twin Roger Brooks, who bounced around
from home to home, showed the same sunny personality as his twin
brother Tony Milasi, who was raised in a warm Italian family. According
to Krueger, the biggest behavioral mysteries are posed by MZA twins
such as Jean and Joan, both of whom took music lessons—the twin who
had received less family pressure to perform turned professional, but the
opposite outcome was also possible. Understanding why and how rearing
circumstances produce one result over another remains a true challenge.

Social Attitudes: Conservatism and More

Irving Gottesman celebrated his retirement with a festschrift held at the
University of Minnesota in June 2001. The resulting 2004 volume, edited
by Gottesman’s former student Lisabeth F. DiLalla, included a chapter
by Bouchard summarizing the MISTRA’s findings on social attitudes.19
On the topic of authoritarianism, he reminded readers, “Although we
were pushed into the domain of genetic influence on attitudes by Martin
et al.’s (1986) work with the W-P [Wilson-Patterson] Conservatism Scale,
the first attitude-like questionnaire on which we published data was the
Multidimensional Personality Questionnaire’s traditionalism scale.”20
Our 1988 personality study reported an MZA correlation of 0.53 for
traditionalism, based on forty-four MZA twin pairs.21 A 2003 paper and
the 2004 festschrift chapter updates showed a remarkably stable figure
(0.52), based on seventy-four MZA twin pairs, with the latest MZT cor-
relation based on a study of 626 sets equal to 0.61.22 The numbers re-
flected substantial genetic influence on traditionalism. In fact, the updated
mean MZA correlations for the Multidimensional Personality Question-
naire’s eleven personality scales and three higher-order factors (0.46 and
0.50) closely matched those based on the 626 MZT twin pairs (0.46 and
0.52).23 In addition, the estimated mean heritabilities from the reared-
together study (0.44 and 0.49) were extremely consistent with the MZA
intraclass correlations which directly estimate heritability. The data on
traditionalism and the other personality scales proved to be important to
the analyses of the Wilson-Patterson Conservatism Scale.24
254 B O R N TO G E T H E R — R E A R E D A PA RT

The Wilson-Patterson Scale was brought into the study in 1986. Bouchard
also added Altemeyer’s Right-Wing Authoritarianism Scale to the assess-
ment at that time, but for some reason the authoritarianism data were
analyzed first. (Bouchard said he was a “moron” not to have included
such inventories in the first place.) At that time, the few attitudinal stud-
ies based mostly on young twins showed little genetic influence on social
attitudes,25 so researchers, including Bouchard, were not motivated to
pursue this domain. Martin et al.’s 1986 twin study on the topic left him
no choice.
The reared-apart twins completed the same twenty-eight-item version of
the conservatism scale as the Virginia Commonwealth University twins,26
not the fifty-item version used by Martin et al. The briefer form allowed
comparisons between MZA and MZT twins, and was better suited to our
crowded assessment schedule. The findings were published in 2003,27 with
an update in Bouchard’s 2004 chapter.
The conservatism scale captured a significant and distinct attitudinal
dimension: conservatism. This was shown by the high correlations of
conservatism with Altemeyer’s Right-Wing Authoritarianism Scale (0.72)
and Tellegen’s traditionalism scale (0.58), based on 338 reared-apart
twin and nontwin participants. At the same time, the Wilson-Patterson
showed low correlations (−0.16 to 0.17) with the Multidimensional Per-
sonality Questionnaire’s other personality scales, such as Absorption and
Alienation. It also showed a low, negative correlation with the IQ scores
(−0.23). The fact that both the personality scales and IQ scores were not
strongly correlated with conservatism showed that the conservatism
scale had discriminant validity—that is, that it was measuring a behavior
that differed from both personality and intelligence. Bouchard also found
that childhood rearing measures were unrelated to conservatism among
twins who had been adopted, with “Importance of Religion” to the mother
being the only exception (−0.22). However, the similarity in this measure
between co-twins was minimal, and any resemblance between them did
not affect their resemblance in conservatism.
It is exciting when new findings replicate existing ones. That is because
there is a sense that the real answer is at hand. Replicating Martin’s (heri-
tability 0.62) and the Virginia group’s findings for conservatism (herita-
bility 0.65 for males and 0.45 for females) was important because most
people believed that this attitude came mostly from the rearing family
environment. Finding that the MZA and DZA conservatism correlations
were 0.59 and 0.21, respectively, and that the genetic estimates ranged
from 0.56 to 0.59, were significant breakthroughs. A successful construc-
tive replication had been completed—but there was more to come.
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 255

In 2004, Bouchard revisited the Wilson-Patterson Scale by organizing its

twenty-eight items into three factors we labeled Sexual Conservatism, So-
cial Conservatism, and Militarism, as well as a Total Conservatism score
based on all twenty-eight items. Again, it was found that Total Conserva-
tism correlated highly with traditionalism (0.60), but not with the other
Multidimensional Personality Questionnaire scales. As expected, the three
factors and total score correlated highly with right-wing authoritarianism
(0.62, 0.42, 0.51, and 0.73, respectively). These findings demonstrated con-
vergent validity among the measures; the scales that should theoretically
be related to one another were related. The three factors and total score
also showed the expected lack of correlation with general intelligence
(−0.21 to −0.14). These findings showed discriminant validity among the
measures; the scales that should theoretically be unrelated to one another
were unrelated.
The four MZA intraclass correlations (0.51 to 0.62) exceeded the four
DZA intraclass correlations (−0.05 to 0.35), consistent with genetic in-
fluence on the specific factors comprising conservative attitudes. How-
ever, the near zero DZA correlation (−0.05) for Militarism was a surprise
low that was hard to explain. The only corroborating evidence available
were the low DZT (0.23) and full sibling (0.20) correlations for Militarism
in the Virginia study. Militarism could conceivably be an emergenic trait,
influenced by unique gene combinations. If so, it would not be unusual to
see army volunteers and conscientious objectors within the same family.
The within-study replication was important because it confirmed con-
servatism as a distinct social attitude, but did so in a new way by means of
the three factors and a total conservatism score. This work also showed
the benefit of having direct access to both the twins’ data and to their life
histories—the findings gave scientific meaning to the twins’ experiences,
downgrading the explanations based on coincidence or fabrication that
some critics had claimed. The firemen twins, Mark and Jerry, were com-
pletely against women serving as volunteer firefighters. The giggle twins,
Barbara and Daphne, who were so talkative on most issues, were totally
uninterested in politics and fell silent when the topic came up. The British
twins, Irene and Jeanette, both crossed out “tolerate” and replaced it with
“respect” on a questionnaire item concerning other people’s religions.28
These glimpses into the twins’ minds often came about casually over
lunches and coffee breaks, and we made notes of these observations.
Bouchard thought it was best to avoid controversial topics with the twins
to keep the atmosphere congenial.
Bouchard’s 2004 chapter also provided updated results for religiosity.
The MISTRA’s within-study replications and those of other investigators
256 B O R N TO G E T H E R — R E A R E D A PA RT

showed that the genetic influence on religiosity stayed at about 0.50. The
widespread belief that religiosity in adulthood was a product of family
rearing during childhood was under revision.
The idea that social attitudes are partly shaped by genetic factors is
still hard for some people to accept. This is probably because parents and
children are seen together at religious institutions, social activities, and
community events. It makes intuitive sense that children’s social attitudes
should be shaped by those of their parents. In fact, the MISTRA found
that among intact biological families the moral religious orientation and
intrinsic religiosity were substantially correlated (0.53). Again, because
biological parents and their children share both genes and environments,
these causal factors cannot be separated. Adoptive parents and children
who share no genes in common provide informative tests of relationships
between family background and religiosity. The 1999 study of religious
behavior reported a negligible relationship between moral religious em-
phasis in the childhood home and intrinsic religiosity during adulthood
(0.10) among the adopted twins.29
Children grow up and draw conclusions of their own, filtering infor-
mation through their own personalities, perceptions, and experiences. The
Virginia group found only age-related differences in conservatism until the
young reared-together twins turned twenty. As young adults, these twins
could express their own views more freely, allowing their genetic tenden-
cies to emerge. That is, children usually follow their parents’ religious
practices until they are old enough to form their own beliefs and act on
them. Laura Koenig, one of Bouchard’s dissertation students and the only
one to study only reared-together twins, showed the same effect.30 She
did not find a genetic influence on retrospective religiousness (religious
activities and interests adults recall as children), but she did find a genetic
influence on current religiousness (religious activities and interests adults
have at present).31
Koenig arrived in Minnesota in 2001 after the data collection had
ended, but she said, “I loved hearing Tom’s stories about the different pairs
and what they were like.” She often uses MZA twins Oskar (raised Catho-
lic in Nazi Germany) and Jack (raised Jewish in Trinidad) as “a great ex-
ample of how religious affiliation is environmental, and how religious im-
portance can be influenced by genes . . . There is no gene for religiousness,
but the genetic effects on personality and such can influence people’s be-
liefs. The reared-apart twins were a great way to make that point.”32
The rearing home’s religious emphasis, parents’ education, and other
family factors were just not there when it came to social attitudes and re-
ligiosity in adulthood. For those who maintained that they were, this situ-
ation exemplified what Bouchard called “an argument,” not evidence.33
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 257

Reading and General Intelligence

Wendy Johnson arrived at the University of Minnesota in the fall 2000 as
a new graduate student in the Department of Psychology. Bouchard sug-
gested that she work with Matt McGue since Bouchard would soon be
retiring. “But [Bouchard] is still here,” she told me in October 2009, im-
plying that he enjoyed the MISTRA too much to leave.
Johnson earned her PhD in 2005, then worked for Bouchard and Mc-
Gue as a postdoctoral fellow for a year and a half. As of 2010, she has
been a research fellow and reader at the University of Edinburgh, Scotland,
analyzing cognitive ability data from the Minnesota and United Kingdom
samples. Understanding the genetic and environmental structure of intel-
ligence and personality is her passion. Over the years, Johnson and I have
coauthored several articles and a chapter using the reared-apart twin
data and cohosted a symposium.34 But it wasn’t until that October day
that I heard her reflections on the MISTRA.
Johnson said, “I have never met a single reared-apart twin, but I would
love to.” Bouchard was no longer gathering data in 2000, so Johnson
knew the pairs only as identification numbers in computer files. She prob-
ably knew the files better than anyone, but working directly with twins
might help her answer questions she still wants to address: Is there some-
thing special about being a twin that affects behavior? Why do twins re-
cover from low birth weight better than nontwins? Such questions reflect
the currently changing behavioral-genetics trends—away from estimating
heritabilities and toward finding the psychological and biological mecha-
nisms underlying behavior.
Johnson was first author on three papers that combined tests and in-
ventories within and between the domains of general intelligence, special
mental abilities, reading performance, and personality. Her work pro-
duced new ideas about the structure of intelligence. One of her first pa-
pers concerned the Stroop Color-Word Test.
The Stroop Color-Word Test has been of psychological interest since
its introduction in 1935.35 It is a test of mental attention and flexibility
that assesses how well people can inhibit or stop one response to say or
do another response.36
The 1978 version of the Stroop,37 added to the MISTRA in 1986, pres-
ents participants with three pages. The first page displays color words
(green, red, and blue) printed in black ink, the second page displays rows
of Xs printed in various colors, and a third page displays color words
printed in different ink colors (e.g., the word red would be printed in
green). Interference on this test refers to whether a word’s meaning, such as
red, affects the subject’s ability to name the color if it is printed in green.
258 B O R N TO G E T H E R — R E A R E D A PA RT

The color-word interference score is determined by the number of colors

on page three that are read aloud correctly in forty-five seconds. This score
has been hypothesized to reflect individual differences in verbal pro-
cessing. Interference is greatest during the elementary school years when
automaticity (performing an activity using little focused attention) with
color words is new, but performance improves over time. Reading skill
has also been linked to Stroop test performance but mostly in younger
populations and clinical samples. Other Stroop measures have been stud-
ied, such as the word score (number of words on page one read aloud
correctly) and the color score (number of colors on page two read aloud
correctly).
Only two twin studies using the Stroop test were available when John-
son began her analyses. Greater MZT than DZT twin resemblance was
found, but the samples were small, and the twins were elderly.38 Thus,
the MISTRA was the first study to consider genetic effects on Stroop test
performance in a middle-aged twin sample, and to relate Stroop scores to
mental ability, reading, and personality. This was also the first time that
data from the two reading tests, the Slossen Oral Reading Test and the
Woodcock-Johnson Reading Mastery Test, were used.39
Reading tests were added to the MISTRA in the late 1980s in col-
laboration with Professor Jay Samuels in the Department of Educational
Psychology. This work began when Bouchard was hosting a visitor from
abroad who wanted to meet Samuels. The three met in Elliott Hall, and
Samuels asked Bouchard if he had included a reading component in the
assessment schedule. In the mid-1980s, there were some tape recordings
made of the twins in various settings, but not the more informative tests
of word recognition, reading speed, and comprehension that interested
Samuels. “Tom realized that this was an opportunity to do something,”
Samuels recalled. In the weeks that followed, Samuels and I assembled
three oral reading passages, a spelling test, and two reading tests that were
used in educational assessments. According to Samuels, “We did it [the
reading component] a gorgeous wonderful way. We chose passages that
could be read at advanced high school or college levels. We did not use a
multiple-choice format—we had the twins read the passages out loud and
tell us what they remembered [known as active recall]. Doing two things
at the same time—deciding and comprehending—gets to the issue of
reading fluency.”
Samuels and I also designed a spelling test. We selected thirty-seven
words from Dr. Richard Venezky’s computer-generated list that could
either be spelled phonetically (e.g., department) or had unique letter com-
binations that were not rule-related (e.g., echo). We thus assessed rule
mastery and visual memory. Samuels called the data “an international
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 259

gold mine.” Johnson made great use of this gold mine to produce her three
studies. Her paper on genetic and environmental influences and correlates
of the Stroop test utilized the Wechsler IQ scores, selected special mental
ability tests, and the two standard reading tests. It included fifty MZA twin
pairs, thirty-seven DZA twin pairs, as well as twins’ spouses and compan-
ions for a total of 271 participants. It remains the largest study of its kind.
Genetic influence was found on every Stroop test measure, ranging
from 0.35 for interference to 0.50 for words.40 The Stroop was not devel-
oped with the idea of assessing genetic and environmental influences
on performance, a feature yielding results that were both unbiased and
persuasive. Johnson also found that most of Tellegen’s personality scales
showed little relationship to Stroop performance. This was somewhat sur-
prising because previous studies had shown a relationship between Stroop
test performance and impulsivity. Interestingly, the IQ score, most men-
tal ability measures, and the reading test scores were significantly corre-
lated with Stroop test performance. It was possible that the twins’ simi-
larities in mental abilities were responsible for the similarities in their
Stroop scores; however, the twins’ correlations hardly changed when
the effects of the mental performance scores were statistically removed.
Thus, the different Stroop measures are valid ability constructs, in and
of themselves.
Johnson’s interests led her to consider a second problem: Is the genetic
factor structure (i.e., the different ability components or dimensions) of
general intelligence and reading performance the same for children and
adults? This analysis also offered an opportunity to look at the heritabili-
ties of reading and word recognition.41 In order to maximize the differ-
ence between our general intelligence and reading measures, she chose
three fluid ability tests from the special ability battery—pedigrees, flexi-
bility of closure, and induction.42 These three tests reflect abilities that are
not closely tied to formal education, whereas reading is a learned skill. The
reading measures included the two word-recognition tests (Slossen and
Woodcock), the three reading passages, and the spelling test that Samuels
and I had assembled.
Happily, in 2009 I found old copies of the spelling words and reading
passages in my files. In the 1980s and early 1990s, I could almost recite
them by heart after having listened to the twins repeat them so many
times. The following list includes some of the words and sentences from
the spelling tape, exactly as the twins would have heard them.

• Department—He worked in a department store—Department

• Echo—I heard the echo—Echo
• Nuisance—The small noisy child was a nuisance—Nuisance
260 B O R N TO G E T H E R — R E A R E D A PA RT

The person on the tape reading them was Bouchard. Twins were instructed
to write each word on an answer sheet once it had been repeated.
Twins then read three passages aloud and told us everything they re-
membered about what they had read. The first sentences from each pas-
sage are reprinted here to provide a sense of the topics we covered:
• A medical problem that seems to afflict women more than men is
sensitivity to cold.
• Popular opinion holds that success at work depends upon continu-
ally pushing for what you want and refusing to take no for an
answer.
• When researchers first began to study reading, many ideas about its
nature were generated from the method of introspection.
The twins’ scores on the spelling test were the number of words they
spelled correctly. Their scores on the reading passages were created by
assigning values to the number of different points they presented in each
passage during their retelling; their tape-recorded sessions were rated by
independent judges. This approach to reading comprehension relies on
active rather than passive recall and is generally equivalent to the more
commonly used multiple-choice method.
Johnson found that the genetic factor structure of general intelligence
and reading performance was the same for adults and for children. A strong
relationship between general intelligence and reading performance in these
adult twins was also found. Interestingly, the twins’ reading comprehen-
sion skills were linked to their general intelligence but not to their read-
ing performance. It could be that, in adulthood, recognizing words does
not guarantee text comprehension—some people can read fluently without
grasping the concepts behind what they read. The participants’ spelling
skills were linked to both their general intelligence and to their reading
performance, not surprisingly as words with both phonetic and nonpho-
netic letter combinations had been included. Words with nonphonetic
letter combinations are harder to spell, so it made sense that the more
intelligent, better readers were also better spellers.
The heritabilities for the reading measures were higher than those es-
timated for children in prior studies (0.21 for spelling to 0.45 for word
recognition).43 Our values were 0.77 for the Slosson, 0.74 for the Wood-
cock, 0.76 for spelling, and 0.51 for reading comprehension. This was
not very surprising because the heritability of general intelligence increases
with age. But the relatively high DZA correlations on the Woodcock
(MZA: 0.72, DZA: 0.70), Slosson (MZA: 0.73, DZA: 0.60), and reading
comprehension (MZA: 0.36, DZA: 0.59) were puzzling because DZA
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 261

twins are expected to be only half as alike as MZA twins. Of course,

higher than expected DZA twin resemblance on some mental ability mea-
sures had been observed in the past. None of the possible explanations
we could think of, such as assortative mating or selective placement,
could be confirmed.
These intelligence and reading studies are good examples of how the
MISTRA’s rich data files enabled many types of analyses, over and be-
yond the heritability studies. In contrast, the investigators of the three
early reared-apart twin studies focused mostly on the twins’ behavioral
similarities and dissimilarities. This observation is not intended to detract
from the contributions made by these early studies. Clearly, the breadth
and diversity of their tests were less than those of the MISTRA, and the
sophisticated quantitative methods that allow genetic and environmental
dissection of measured traits were unavailable. The MISTRA was able to
explore many avenues of inquiry because of its comprehensiveness. Here
is another example, also from Johnson’s work: the conclusion of her
third article is stated in its captivating title “Just One g.”44

Johnson’s third analysis addressed continuing controversies within psy-

chology, namely, the identification of a general intelligence factor called
“g” and its measurement. Since the beginning of the modern mental test-
ing movement in 1905 with French investigators Binet and Simon,45 re-
searchers have debated the evidence for a general intelligence factor ver-
sus separate mental abilities as well as the best methods for intellectual
assessment. General intelligence, or g, refers to a shared factor that under-
lies human abilities and explains why tests of different skills (e.g., vocabu-
lary and comprehension) are correlated.46 Competing models recognize
separate abilities or talents that function independently or somewhat in-
dependently.47 The MISTRA investigators reasoned that finding similar g
factors based on different clusters of mental ability tests would demon-
strate that there is, in fact, “just one g.”
Three different mental ability batteries had been completed by the
436 MISTRA twins, spouses, and others: the Wechsler IQ test, the Ha-
waii Ability Battery, and the Comprehensive Ability Battery.48 Here, the
twins were treated as separate individuals without reference to their relat-
edness. In a combined analysis, the three g factors correlated highly, ranging
from 0.99 to 1.00, values that are rarely seen in psychological research.
Based on this finding, there appears to be a single g—of course, this g does
not capture the full range of abilities, nor does it pinpoint the biological
processes underlying human intellect. But the findings made a strong case
for including g in future models of intelligence. According to Johnson,
262 B O R N TO G E T H E R — R E A R E D A PA RT

“These results provide the most substantive evidence of which we are

aware that psychological assessments of mental ability are consistently
identifying a common component of general intelligence.”49 Bouchard
worded this conclusion more forcefully in his 2009 paper, asserting that
“the previously widespread belief that different intelligence tests yield es-
timates of quite different intelligences is simply false.”50
Johnson and Bouchard used the twins’ and spouses’ mental ability data
to probe the structure of human intelligence in other ways.51 They com-
pared the fit of the data to the three extant major intelligence models:
the Cattell-Horn fluid-crystallized model, the Vernon verbal-perceptual
model, and the Carroll three-strata model. This was one of the few MIS-
TRA analyses that specified a hypothesis at the outset—the Vernon model
was expected to fit the data best. Beyond that, the work was exploratory
in nature, although it generated other hypotheses. Bouchard waited years
to do this analysis—statistical techniques were too rudimentary when the
study began, and the sample needed to grow.
The Cattell-Horn model recognizes two types of intelligence: crystal-
lized (the intelligence acquired from learning and culture) and fluid (the
genetically based capacities unrelated to experience). The Vernon model
recognizes a general intelligence factor and two main ability groupings:
verbal-educational and special-practical-mechanical. The Carroll model
includes a general intelligence factor, broadly specialized abilities, and
narrowly specialized abilities that relate to a specific domain in the second
group. As expected, the Vernon model fit the MISTRA data best, although
Johnson and Bouchard enhanced it by adding memory and image-rotation
factors. They called it—appropriately—the Verbal-Perceptual-Image
Rotation Model, noting its consistency with the idea of coordinated
function across brain regions and the importance of brain laterality (spe-
cialization of mental function across the brain’s left and right hemi-
spheres) in mental performance. The g factor contributed substantially to
the verbal, perceptual, and image-rotation components, which then con-
tributed to eight more specialized skills (verbal, scholastic, fluency, num-
ber, content memory, perceptual speed, spatial, and rotation). More will
be said about this model in the next chapter.
Johnson replicated this model in a 2005 study using Thurstone and
Thurstone’s 1941 data from 710 Chicago schoolchildren.52 She repli-
cated it again in 2007 using a battery of forty-six tests available for 500
professional seamen from the Royal Dutch Navy.53 And she did once
more in 2008 using a different five-test battery and sample.54 The MISTRA
twins had launched a continuing research program—it was usually the
other way around, with the MISTRA testing new methods and findings
with the twins.
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 263

Tony Vernon, professor of psychology at the University of Western

Ontario in London, Canada, is an expert in the study of intelligence and
editor of the journal Personality and Individual Differences. According to
Vernon, “[The Verbal-Perceptual-Image Rotation Model] is the first ‘new’
model of the structure of intelligence to come along in a long time. It showed
itself to be a better fit to the data than Thurstone or Cattell or P. E. Vernon,
though it was mostly closely aligned with the Vernon model.”55
I never thought I would say this, but the fact that the original g findings
on the model were based largely on reared-apart twin data was inconse-
quential. Data from good nontwin samples could have—and have—
accomplished the same thing.

Johnson made an interesting point as we ended our interview. Unlike many

other MISTRA colleagues, she said Bouchard may have compromised the
scientific rigor of the study by publicizing the MZA twins’ unusual simi-
larities. “You don’t need quirky stuff—it does not establish anything about
behavior. The anecdotes gave detractors something to attack. You need to
show striking observations more systematically. Of course, I say this now
with 20/20 vision of hindsight which Bouchard could not have had at that
time.” Johnson referenced a group picture of identical twins in which the
children’s hands are positioned similarly within pairs, but differently be-
tween pairs. It appears in a 1982 introductory genetics textbook by the
late geneticist Edward Novitski.56
Some people might claim that the twins in the picture simply followed
the photographer’s instructions regarding their hand placement, but the
similarities were mostly within the pairs. One might also suggest that each
pair’s parents had instructed them on how to hold their hands when they
were being photographed. However, Johnson speculates that while we are
all socialized to fit into our environments, no one instructs us on how to
hold our hands. Of course, we really don’t know what the twin children
were told to do, but as Bouchard said in his last lecture, observing similar
hand postures in MZA twin pairs is the best answer to objections that
hand holding is a learned behavior. That answer is shown in Figure 12-1.
Meaningful information is stored within each pairs’ signature habits.
But contrary to Johnson, one could argue that it was appropriate for
Bouchard to discuss these habits publicly. Rare similarities between MZA
twins are provocative, suggesting genetic effects, possibly emergenic in
nature. The fact that the same physical and behavioral quirks are more
often found in MZA than DZA twin pairs suggests, but does not prove,
that genetic influence is at work, providing a rationale for further study.
It is also the case that the MZA twins’ unusual observations were not
publicized exclusively by Bouchard.
264 B O R N TO G E T H E R — R E A R E D A PA RT

Comment
The Association for Psychological Science publishes a slim journal called
Current Directions in Psychological Science. Articles appear on a mostly
invitation-only basis, although interested contributors can bring poten-
tial topics to the editor’s attention. In a 2004 article, Bouchard identified
several key trends and big questions.57 He noted that behavior geneticists
had always assumed that genetic influence would vary from trait to trait,
and that behaviors such as social attitudes would show little heritability.
But the two recent surprises have been the generally moderate level of ge-
netic influence across most traits. According to Bouchard, “It seems rea-
sonable to suspect that moderate heritability may be a general biological
phenomenon rather than one specific to human psychological traits, as
the profile of genetic and environmental influences on psychological
traits is not that different from the profile of these influences on similarly
complex physical traits and similar findings apply to most organisms.”58

Figure 12-1. Five MZA twin pairs shown with CBS host Lesley Stahl on the
set of 48 Hours, November 15, 2003. Note the similar hand and leg positions
of the twins seated in the front row. The male twins to the left and the female
twins in the center of the second row did not participate in the MISTRA.
(Photo credit John P. Filo, CBS News.)
 T W I N R E L AT I O N S H I P S , AT T I T U D E S , A N D M E N TA L A B I L I T I E S 265

Bouchard also asserted that the name of the field—behavior genetics—is

an “unfortunate misnomer.”59 The studies are neutral with respect to gene-
tic and environmental influences. Now that the heritabilities of most traits
have been established, interest in uncovering the underlying molecular
and environmental mechanisms is paramount.
 CHAPTER THIRTEEN

Sexual Development, Fluctuating Asymmetry, Body Size,

and the Structure of Intelligence (2006 and Beyond)

I am looking at the calendar as I write this last chapter on

the MISTRA’s findings. It is July 1, 2010, and I know that later
today thirty-five-year-old Hanan Hardy will travel to Morocco to
meet her twin sister Hassania for the first time. The twins, who appear to
be MZ, were separated at birth because their biological mother could not
care for them. Hanan was reared in the United States by their mother’s
close friend and enjoyed middle-class privileges; Hassania remained in
Morocco and was sold to another family when she was just one week
old, living in poverty most of her life. Hanan’s adoptive family kept her
birth details hidden, but Hanan learned the truth several weeks earlier
from her twin sister who found her on Facebook. Hassania knew she had
a twin, but she had been unable to find Hanan without that vital re-
source. The MISTRA’s reared-apart sample might have been much larger
if the Internet had been available for public use in the 1980s.
In the “old days,” we would have done our best to bring Hanan and
Hassania to the university, hiring a translator to help Hassania with tests
and interviews. The twins’ dramatically different rearing situations
would have put genetic explanations of behavior to the strict tests our
critics had clamored for. Based on other MZA pairs with very different
backgrounds, we probably would have found some striking behavioral
similarities between the Moroccan twins. We would have also learned
more about how their homes, communities, and cultures mixed with
their genetically based proclivities to produce differences between them.1
Reared-apart twins are no longer visiting Minnesota, but I keep track
of all the cases I discover. I conduct “mini-assessments” with some of
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 267

them, just as Hur and I did in 2008 for the Korean twins. I also hope to
study Hasan and Hassania some day, albeit at a distance. Most of the MIS-
TRA investigators also miss the excitement and frenzy of those whirlwind
twin weeks, but they are making the most of the available data. Many
papers appeared in the poststudy period, and there are more to come.

Sexual Development
During one of their premeeting telephone calls, Hanan and Hassania told
each other that they weren’t feeling well, and learned that they had started
their periods at the same time. According to Hassania, they believed that
their matched cycles and physical complaints were amazing. In spite of their
menstrual discomfort, I could tell that they were having a good time dis-
covering things they had in common.
The twins’ coordinated menstrual timing was not surprising. No one
had studied menstrual synchrony in twins, although reared-together twin
studies have reported genetic influence on age at menarche, the onset
of menstruation (MZT ri: 0.65 to 0.97 and DZT ri: 0.18 to 0.50).2 Far-
ber found a 9.3-month age difference in menarche for twenty-eight MZA
female twin pairs scattered throughout the early studies. She also noted
that the twins described similar symptoms, both common (e.g., irregular-
ity and emotionality) and uncommon (e.g., fainting and vomiting) as did
Hanan and Hassania.3 However, a formal study of menarche in reared-
apart twins had not been conducted until 2007.4
The MISTRA had relevant data for twenty-seven MZA female pairs
and thirty-one DZA female pairs, taken from the Sexual History Timeline.
The same data were available for thirty-three MZT female twin pairs and
fourteen DZT female twin pairs from Lykken’s Minnesota Twin Registry.
I had a particular interest in the potential findings on age at menarche,
due to some relevant work in evolutionary psychology, as I will explain
later in this chapter. I analyzed the data in 2007 in California with Dr. Joanne
Hoven Stohs, my colleague in the department of psychology. Stohs had
never before worked with twins and acquired appreciation for genetic ef-
fects on behavior by doing so.
The four-group (MZA, DZA, MZT, DZT) design confirmed our ex-
pectation of genetic influence on age at menarche. The MZA twins (ri:
0.56) were more alike than the DZA twins (ri: 0.16), and the MZT twins
(ri: 0.70) were more alike than the DZT twins (ri: 0.41). The MZT and
DZT twins were also more alike than the MZA and DZA twins, respec-
tively, suggesting shared environmental effects, possibly through diet or
exercise.
268 B O R N TO G E T H E R — R E A R E D A PA RT

I wondered about features of the reared-together twins’ cohabitation

that may have led to their synchronous menstrual timing. A 1971 study
by biological psychologist Martha McClintock showed that female col-
lege students living in the same dormitory acquired menstrual synchrony
over time.5 The significant factor in McClintock’s results was that the
women had spent time together, but the underlying mechanism was un-
certain. Of course, menstrual synchrony is not the same as age at men-
arche, but cohabitation could affect these measures in similar ways. Far-
ber noted that twins who had met prior to puberty had a more similar
age at menarche than twins who had met later, suggesting that contact
during adolescence could affect this measure. However, she did not ex-
amine associations between age at menarche and contact frequency.
The mean ages at menarche were 12.50 years for the reared-apart twins
and 12.86 years for the reared-together twins, similar to the age of 12.70
years reported for the general population.6 However, our MZA twins
reached menarche significantly earlier (12.24 years) than our MZT twins
(12.80 years), a finding for which we had no explanation.
When Stohs and I began our study, a large body of work had been gen-
erated on the Belsky-Draper hypothesis.7 This evolutionary-developmental
hypothesis predicts that girls from father-absent homes should experience
menarche earlier than girls from father-present homes, because it would
be biologically and psychologically advantageous for them to leave their
disorganized environments to begin families of their own. The presence of
a stepfather and lower levels of parental investment have also been linked
to girls’ earlier age at menarche, whereas parental warmth and approval
have been linked to later menarche. Research evidence has both sup-
ported and challenged these findings.8
I was skeptical of the Belsky-Draper hypothesis because I favored twin
research showing genetic effects on age at menarche. Furthermore, a 2006
study by former University of Virginia graduate student Jane Mendle
showed that the daughters born to twin mothers did not differ in age at
menarche if one daughter was raised by a stepfather. The daughters of
each twin were cousins, but they would be genetically equivalent to half-
siblings if their mothers were MZ twins. Fortunately, data to further test
the ideas generated by this body of work were available from the MIS-
TRA’s Twin Relationship Surveys. Questions concerning parental under-
standing, sense of belonging in the family, and childhood happiness and
anxiety were relevant to our goal.
None of the childhood satisfaction and adjustment measures had a sig-
nificant bearing on age at menarche, except understanding by father. We
showed that greater paternal understanding was associated with earlier
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 269

age at menarche, contradicting the Belsky-Draper hypothesis. Our find-

ings were more consistent with research linking favorable living condi-
tions, possibly greater resource availability, to earlier menarche.9 Thus,
the data supported a genetic explanation of individual differences in age
at menarche. Parents who achieve earlier sexual maturity may be the ones
to engage in risky sexual behavior, transmitting these tendencies to their
daughters.
In 2009, Stohs and I went on to study genetic effects on age at first inter-
course using the MISTRA twins.10 Age at first intercourse had never been
studied in reared-apart twins although several reared-together twin stud-
ies were available. These studies had found genetic effects on this mea-
sure, but the extent of genetic influence varied widely (0.00 to 0.72 for
males and 0.17 to 0.49 for females). Age also made a big difference in the
degree of genetic influence—most studies found higher heritabilities for
younger twins than for older twins who also experienced their first inter-
course at later ages. This finding probably reflected an easing of sexual re-
strictions among the more recent, hence younger, cohorts at the time of
the study. A more permissive sexual climate would allow greater freedom
for expressing individual differences in sexual behavior.
Stohs and I approached our study the same way we approached our study
of age at menarche. Using data from the Sexual Life History Timeline, we
examined resemblance for age at first intercourse for thirty-four MZA twin
pairs, twenty-five DZA twin pairs, and twenty-three individual twins. We
then examined age at intercourse in relation to the timing of other behav-
iors, such as age at first kiss. We also examined life history events consid-
ered meaningful within evolutionary and developmental frameworks, such
as peer and family relationships. We limited this second set of analyses to
females because of our previous work on age at menarche.
The twins’ mean age at first intercourse was 18.80 years. Consistent
with previous twin studies, older twins engaged in their first intercourse
at later ages (19.50 years) than younger twins (17.60 years). Also as ex-
pected, the MZA twin pairs (0.34) showed significantly greater resem-
blance than did the DZA twin pairs (0.22). These values did not change
when we separately examined the MZA males (0.39, n = 13 pairs) and
females (0.32, n = 21 pairs). Thus, the heritability of age at first inter-
course, expressed directly by the MZA correlation, was 0.34, consistent
with the findings from reared-together twin studies.
Some evolutionary-based studies have found that adverse family factors
predicting earlier age at menarche also predict earlier age at intercourse.
However, in 2002, psychologist David Rowe showed genetic influence on
both as well as a significant genetic correlation (0.72) between them,
270 B O R N TO G E T H E R — R E A R E D A PA RT

challenging an exclusively family-dynamic interpretation of the findings.

That is, genetic transmission may affect the timing and association be-
tween these measures. Some developmental psychologists see the connec-
tion between age at menarche and sexual activity as contingent upon peer
relationships and the distancing of parents and children during adoles-
cence and young adulthood. Again, items in our Twin Relationship Sur-
vey helped make sense of these associations.
Not surprisingly, twins who had intercourse earlier in life experienced
their first kiss, first marriage, and first child at earlier ages than those who
had intercourse later in life. The more interesting findings were that twins
who were less happy than their childhood peers and less fulfilled in their
childhood homes had intercourse at younger ages than their more con-
tented and fulfilled counterparts. These results were consistent with de-
velopmental perspectives, but could also reflect genetically based person-
ality traits that propel these individuals toward earlier sexual activity. The
quality of the twins’ relations with their parents was unrelated to age at
first intercourse, suggesting the greater effects of peer relations on early
sexual activities.
Our research on menarche and first intercourse demonstrated the util-
ity of using behavioral genetic designs for examining evolutionary con-
cepts and findings. As indicated earlier, these disciplines have functioned
separately for long periods, but there is evidence of a rapprochement be-
tween them.

Fluctuating Asymmetry
If lines are drawn down the center of our bodies, it is unlikely that any of
us are physical mirror images. Hand preference, foot size, and wrist cir-
cumference are not exactly the same on our left and right sides. These de-
partures from physical equivalence, known as fluctuating asymmetry (FA),
have been associated with genetic and environmental stressors,11 such as
inbreeding and mutations (genetic), and pesticides and food deficiency
(environmental).12 Fluctuating asymmetry has interested evolutionary psy-
chologists because higher levels are thought to reflect increased stress sus-
ceptibility and reduced developmental stability. For example, women have
been shown to prefer the scents of low FA men to the scents of high FA
men, especially during females’ high fertility phases.13 Increased fluctuat-
ing dermatoglyphic asymmetry has also been associated with greater gene-
tic risk for schizophrenia.14
In the late 1980s, Bouchard added a battery of anthropometric mea-
sures to the MISTRA assessment schedule. The addition was not done with
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 271

FA in mind because evolutionary-based studies exploring associations

between FA and human psychological traits did not appear until the early
1990s.15 Fluctuating asymmetry was, however, of interest to behavioral
ecologists, human biologists, and psychiatrists in the 1980s.16 The an-
thropometric measures were added to the MISTRA’s assessment schedule
mainly because extensive physical measurements had never been made
on twins reared apart. However, we noted mirror-image effects in the
twins’ physical traits in order to explore questions of physical asymmetry
at a later time.
In fact, there was a precedent for doing such work. In 1937, Newman
et al. had recorded his MZA twins’ sitting height, head length, head
width, and cephalic index, as well as the usual height, weight, and finger-
print measures.17 The only significant correlation Newman reported was
between differences in the twins’ weight and differences in their physical
environment (0.60). However, as a biologist, Newman was also interested
in physical asymmetries: “A good many pairs show[ed] mirror-imaging in
eye defects, tooth anomalies, in master-eye, in a number of anthropomet-
ric measurements, and in motor and strength tests of the two hands.”18
However, the only documentation of these data is found in the twins’
separate case histories included at the back of their book—only a handful
of these measures were examined systematically.
Newman et al. found no difference in the MZA (n = 19 pairs) and
MZT (n = 50 pairs) twin pairs’ correlations for standing height, sitting
height, weight, head length, and head width. However, the mean within-
pair weight difference (5.87 pounds) between the MZA (9.90 pounds)
and MZT twins (4.03 pounds) was statistically significant, a finding they
variously explained by differences in the MZA twins’ diet, climate, and
care while they were growing up. The height similarity of the MISTRA’s
reared-apart twins (ri = .96) was quite comparable to that of reared-
together twins (ri = .94), based on a 2011 analysis. In contrast, the weight
similarity for the MZA twins (ri = .69) was relatively less than that of the
MZT twins (ri = .87), suggesting shared environmental effects on body
weight.19 These analyses used a subset of mostly female MZA twin pairs
(n = 29, 72 percent female) and a comparison group of MZT twin pairs
(n = 49, 76 percent female), studied in collaboration with Tufts University
researchers. The relatively low female MZA weight correlation is consis-
tent with our early findings and those of other researchers.
Our array of seventy-seven anthropometric indices let us look at FA
heritability and other features as they became popular research topics.
In 1997, researchers reported that people with higher FA had lower IQs,
a  finding that led to attempts at replications. The degree of association
272 B O R N TO G E T H E R — R E A R E D A PA RT

between FA and IQ varied across studies, which used fairly small sam-
ples, so we attempted a study of our own.
The twins were asked to partially remove articles of clothing during the
anthropometric assessment that was conducted by two examiners in a pri-
vate room. One examiner made the measurements while the other exam-
iner recorded them.20 The metal calipers and other instruments were often
cold, causing some twins to laugh or gasp slightly. But they accepted these
procedures with the same good humor that they showed throughout the
week. Some twins even thought the anthropometric assessment was inter-
esting and fun.
I was a frequent anthropometric examiner along with a graduate stu-
dent assistant. This was an interesting session because it revealed the
separate physical similarities in the MZA twins that together contributed
to their generally matched appearance. We literally hugged some of our
heavier twins in the process of measuring their body circumference. We
spotted the same clefts on the chins of MZA twins Trent and Jim, and the
same receding hairlines on the big heads of the fireman twins, Mark and
Jerry. I wish we had performed these procedures on the nineteen-year-old
MZA male triplets Bob, Dave, and Eddy who visited Minnesota in 1980
before we conducted anthropometry studies. A photo of the three, dis-
played in Lykken’s former laboratory, revealed that they had mirror-
image smiles—one triplet’s mouth curled up higher on the side opposite
to that of his two brothers. One of the triplets was left-handed, and we
might have found other reversed physical features among them, such as
birthmarks and moles. We ended up with anthropometric data for eighty-
eight twin pairs, sixty spouses, and twenty-seven nontwin relatives.
Johnson, Bouchard, and I concluded that there was a problem with
most studies that had measured genetic influence on FA. The problem was
that the studies had used FA measures of only one trait. We pointed out that
single measures might not capture the FA of many traits affected by the
same stressors. In other words, stress might affect head height and head
width and arm length. A literature survey found that the estimated heri-
tability of FA, based on a single trait, is 0.03.21 In contrast, one exceptional
study that looked at parent-child resemblance for FA, based on an eight-
trait composite, showed a moderate heritability (0.38), although this fig-
ure was possibly inflated due to assortative mating.22 However, the diffi-
culty with that study was the confounding of genetic and environmental
influence, something that would not be problematic with reared-apart
twins. We reasoned that if the heritability of a composite based on many
traits was higher than that of a single trait, then the FA of each of the
many single traits must be affected by the same factors. In other words,
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 273

the effects of stress on the FA of each trait could be small but might “add
up” across traits. We also reasoned that the factors affecting the FA of the
traits may have a genetic basis.
We created a ten-trait composite that included reliable measures such
as foot breadth, knee height, and ear length.23 We also created three ad-
ditional FA indices, one measure based on the absolute difference between
left and right sides for each trait, another measure that statistically ad-
justed for the direction of trait asymmetry, and a measure that statistically
adjusted for each twin’s variability in the single trait measures. Our results
matched our reasoning. The three indices showed low heritability for the
ten single traits (0.00 to 0.17), but higher heritabilities for the ten-trait
composite (0.27 to 0.30). The use of multiple traits for studying the heri-
tability of FA was a better approach than the use of single traits, as called
for by Belgian biologist Stefan Van Dongen.
Van Dongen said, “The determination of hDI2 [the heritability of devel-
opmental instability] is one of the most challenging in evolutionary biol-
ogy.”24 This is because FA may not accurately reflect developmental in-
stability because it does not capture all sources of stress on the individual.
The reared-apart twins, just by letting us pinch them with calipers and
wrap them in tape measures, were up to the challenge.
In 2008 we completed another study of FA, one that failed to replicate
the reported negative relationship between FA and general intelligence.25
Some researchers had linked a larger left-right side difference in physical
traits to lower intellect, although one contradictory study found greater
physical asymmetry among males with higher IQs. The twins’ intelligence
test scores (a summary measure based on forty-two mental ability tests)
and the anthropometric data were used to test this idea. In order for
this relationship to hold, there had to be a genetic correlation between
FA and ability, that is, the same genes that affected FA had to also affect
intelligence.
We did not find this. We did find that FA was heritable, something we
had shown in the previous study. We also showed, both in this analysis
and in previous IQ analyses, that intelligence was heritable. However,
we did not find a genetic link between the FA and intelligence measures
as shown by the near zero correlations between them (−.01 τo .04). Thus,
FA and intelligence appear to be affected by different sets of genes.
These results made sense. That is because FA is a more likely reflection of
a person’s reproductive fitness or capacity to have children, and that has not
been tied to intelligence. People at all IQ levels can have families. We were
confident in our results because the sample was larger than previously stud-
ied samples, and because intelligence was assessed in more than one way.
274 B O R N TO G E T H E R — R E A R E D A PA RT

An important next step in this research program would be to deter-

mine if more physically asymmetrical MZA twins are less healthy and/
or less reproductively successful than their more symmetrical co-twins.
My late colleague, Linda Mealey, had already linked increased physical
attractiveness to reduced facial asymmetry within pairs of MZT twins.26
This was a great technique because it separated the effects of FA from
other physical attractiveness indicators, such as age and hair color that
are matched for MZ twins. However, Mealey did not find that the more
facially symmetrical twins recalled better parental treatment from child-
hood through age sixteen than their less facially symmetrical co-twins.27
She suggested that childhood asymmetry may differ from later asymmetry
due to developmental changes. The greater number of physical and be-
havioral measures currently available for the MISTRA twins would allow
many informative analyses along these lines. For example, it would be
possible to see if the more symmetrical MZA male co-twins were health-
ier or had fathered more children than their less symmetrical co-twins.28

Body Size, Nutrition, and Health

In 2009, the first set of results from the collaborative health and nutrition
study, conducted with Dr. Susan Roberts at Tufts University, became avail-
able. In April 2010, I visited Roberts and her project manager Paul Fuss
at their downtown Boston campus. I wanted to learn more about this
research that had begun after I had left Minnesota.
On August 18, 1993, the first of thirty-seven MZA twin pairs arrived
at Tufts’s Human Nutrition Research Center. They came to participate in
a four-day study of genetic and environmental influences on energy me-
tabolism or body fat mass. Fifty MZT twin pairs, mostly recruited from
the New England area, also took part. As Fuss explained, the study fo-
cused on individual differences in calorie intake versus energy expendi-
ture, “and what hangs in the balance is body weight and body composi-
tion [amounts of fat, muscle, and bone].” The study’s practical significance
would be the identification of factors associated with body fat mass, lead-
ing to better treatments for metabolic conditions.
I spent about forty-five minutes with Roberts until she left for a series
of academic meetings. She is a small, slim woman with a British accent
who loves her job and works hard at it. Roberts studied nutrition as an
undergraduate student at the University of London, served as a research
technician in Africa, and earned a PhD at Cambridge University for her
work on mathematical models of energy expenditure. She arrived in the
United States as a postdoctoral fellow at the Massachusetts Institute of
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 275

Technology, where she developed interests in weight regulation. This

culminated in the collaborative reared-apart/reared-together twin study
that she called a “stepping stone” toward finding effective weight-control
interventions.
Roberts told me she learned about the “famous MISTRA” from the
professional literature in 1990 and soon after arranged to visit Bouchard
in Minnesota to suggest a joint project. The MZA twins struck Roberts as
“ideal research subjects” for the study she envisioned. Her initial goal was
to study MZA twins and adoption triads composed of adopted children,
their adoptive parents, and their biological parents. This plan was ulti-
mately not feasible because adopted children’s biological parents are diffi-
cult to find, especially their fathers, so this group was replaced with reared-
together twins.
The research arrangement worked well for everyone—Roberts’s 1993
five-year grant of more than one million dollars paid for twins to visit
Minnesota for the five- to seven-day assessment, then fly to Boston. By the
time twins left Minneapolis, they would have recovered from any jet lag
they might have had initially, something that would have interfered with
Roberts’s findings. The twins’ were housed and fed in rooms next to the
laboratory so the data could be gathered under highly controlled condi-
tions. Twins received a $260 stipend for their time as well as an opportu-
nity to tour one of America’s most beautiful cities. “We brought in every
pair that Tom gave us,” Roberts said. “We could have gone on longer, but
our grant cycle ended.”
Roberts did not collect data from the twins herself, but she greeted
them all and found them to be “delightful.” “There was a real joy in their
reunion, and no strife between them,” she told me. As part of the protocol,
the twins independently selected all of their meals on the first day and
were especially pleased when they picked the same items. She recalled
that twins in one of our British pairs each wrote down two items that
were not on the menu—hot chocolate and bananas—even though there
was no place on the menu to write alternative choices. These twins had
traveled to the United States on Virgin Atlantic Airlines and were so ex-
cited about coming that they wrote to the airline’s founder, Sir Richard
Branson, to tell him. They received a letter back and showed it proudly to
the Tufts University staff.
Roberts, who had had considerable experience with research partici-
pants, observed that the twins were “more into [the study] than nont-
wins. They feel special. And the MZAs were definitely more into it than
the MZTs . . . But they could not try to be alike on our measures,” Roberts
assured me. The twins slept in separate rooms in the somewhat sterile,
276 B O R N TO G E T H E R — R E A R E D A PA RT

dormitory-like quarters of the Nutrition Center, and never ate seated to-
gether, although they ate in the same dining room. Every procedure and
questionnaire was completed while the twins were apart. A possible ca-
veat was that the twins’ body composition, food intake, physical activity,
and energy metabolism were studied under controlled conditions rather
than at home in their regular setting. This situation, and their possibly
similar array of food choices from the somewhat limited menu, may have
slightly inflated the heritabilities of the findings. Furthermore, the major-
ity of both MZAs and MZTs were female, limiting the generalizability of
the results. “It’s hard to get men,” Roberts said. Most medical and behav-
ioral studies, including the MISTRA, enroll more females than males for
several reasons. Women are more likely to seek medical attention, women
may gain greater enjoyment from the social aspects of research, and the
sex difference in longevity yields a higher proportion of women in all
adult age categories.
After interviewing Roberts, I spent the rest of the day with project
manager Paul Fuss, hearing about the study, touring the laboratories, and
inspecting the equipment. I also looked through their photographs of the
MISTRA twins, now familiar figures but in unfamiliar settings. Fuss said
that the twins charmed him. “What struck me was that they were strang-
ers to us, yet they were so willing to come here. They took a real leap of
faith, based on Bouchard’s encouragement and their trust in him.” Like
Roberts, Fuss sensed that the MZA twins “felt even more special than the
MZTs . . . they understood their value as MZA twins: it was a unique
opportunity for them to contribute to science.”
Fuss specifically mentioned the switched-at-birth Canadian twins
Brent and George, and George’s brother Marcus, who had accidentally
grown up as George’s twin29 (brief mention of these twins’ compelling story
can be found in Chapter 1). The three had been placed in the same tem-
porary foster home as babies. When the elderly couple caring for them
became overwhelmed by their responsibilities, they decided to move Brent
and George to a different home, but they inadvertently moved George and
Marcus. The twins’ mother returned to claim them when they were two
months of age, never suspecting that a mistake had been made; conse-
quently, Brent went home with the adoptive family most likely intended
for Marcus. The switch was discovered when the twins and Marcus turned
twenty and a student who saw Brent at a college club meeting mistook him
for George. According to Fuss, “George was very sensitive to Marcus.
They came to Tufts shortly after the switch was discovered.” Fuss also
remarked on how physically different Marcus was from the MZA twins
Brent and George.
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 277

The Tufts University protocol was comprehensive, using only state-of

the-art procedures. The complete list of tests and procedures, organized
into the five categories of nutrient intake, body composition, energy ex-
penditure, metabolic parameters, and substrate oxidation, is available on
this book’s Web site.
Twins could consume no food or beverages after 7:00 p.m. on Sunday,
the day they arrived, because testing began the following Monday. Their
food intake was strictly monitored to precisely measure their energy up-
take and energy expenditure. Each day began at 6:30 a.m. when the re-
searchers recorded participants’ fasting gown weight and vital signs. The
day ended after dinner, leaving their evenings free with one exception—
they could not eat and could only drink water after 9:00 p.m. Breakfast,
lunch, and dinner were scheduled for 8:00 a.m., 12:00 noon or 12:30
p.m., and 5:00 p.m., respectively, except on Thursday, when their insulin
test required them to fast between 8:00 a.m. and noon for the collection
of timed blood samples. The various laboratory tests and clinical proce-
dures occupied the mornings and afternoons. Although there was some
flexibility to the schedule, the strict ordering of the tests could not be
changed because some later tests depended on the completion of earlier
ones. Twins knew that they could decline to participate in any part of the
study, as allowed by informed consent procedures, but few did.
The body composition assessment received the most attention from
the investigators. One morning, the twins lay on a cot with leads (used to
record voltage) attached to their right hand and foot. An electrical current
passed between them, relaying information about their quantity of body
fat to an attached monitor. Later in the week, they sat in a tank of warm
water and submerged their heads five or six times for a measure of body
density. Body fat is less dense than water, so someone with more body
fat weights less under water. However, bone and muscle are denser under
water, so someone with more muscle weighs more under water.30 (Twins
were told to bring bathing suits for this part of the study.) Twins would
also lie on a padded table twice over a forty-minute period for a full body
scan that measured their fat distribution and bone density.
The most amazing apparatus was the BOD POD, a white jellybean-
shaped capsule that could assess body density (body volume plus body
weight) more accurately and in less time than underwater weighing. The
device relies on the displacement of air rather than water for body compo-
sition calculations. Figure 13-1 shows a pair of reared-apart twins in the
BOD POD, seemingly ready to launch into outer space.
The “bubble” was less dramatic than the BOD POD, but also informa-
tive. The bubble was a clear, lightweight canopy placed over the twins’
278 B O R N TO G E T H E R — R E A R E D A PA RT

Figure 13-1. MZA female twins Jennifer Mitchell (left) and Margaret Wil-
liams, in BOD PODS. These British twins met for the first time at age fifty-seven
after one discovered a time of birth recorded on her birth certificate—the unique
indicator in Great Britain that one is a twin. They instantly felt like “old
friends” and in correspondence to Bouchard listed similarities in “food, clothes,
shops, religion, music, nail-biting, and excessive use of the telephone.” The twins
celebrated their seventieth birthday together at a party held at Jennifer’s home
in May 2011. (Photo credit Paul J. Fuss.)

heads before breakfast to measure their resting energy expenditure, or

the thermic effect of feeding. When a person consumes calories, his or her
energy expenditure increases so that the body can assimilate nutrients.
Twins could breathe normally and watch television or read while this
thirty-minute procedure took place. However, on one day the assessment
continued after breakfast until noon, with the bubble on for twenty min-
utes out of every half-hour interval.
Throughout the four days of testing, twins sampled specific liquids
and drank “palatability beverages.” The liquids were part of a taste test
in which different amounts of milk, cream, and sugar had been com-
bined. Twins rated the pleasantness and sensory qualities of these liquids
accordingly. The palatability drinks, ingested before lunch on three sepa-
rate days, tested the effects of same-volume low fat, fat plus carbohydrate,
and zero calorie concoctions upon food intake. The twins’ meals were
carefully weighed and measured, as were the bits of food they left behind
on their plates. Anthropometric measurements were also taken on the
first morning, and urinalyses were requested at preset intervals.
I knew many of the participating pairs, such as female twins Debbie
and Sharon who were memorable for their religious similarities and dif-
ferences discussed in Chapter 7. Debbie wrote, “We had THE best time
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 279

there. We would go back in a heartbeat!” Sharon agreed: “Well, ditto on

everything Debbie said. We LOVED IT. Everyone there made us feel
so special—like we were the first and only twins they ever studied. They
took a ton of blood work, and at the end Susan went over it with us
showing how closely matched we were on things like cholesterol, consid-
ering that Debbie was macrobiotic at the time.”
I asked Roberts to comment on Debbie and Sharon’s resemblance
because one twin was eating mostly grains and vegetables on her macro-
biotic diet. Roberts was not surprised by their similarity because choles-
terol level has a demonstrated heritability of 55 percent. The fact that
Debbie was macrobiotic was inconsequential. Roberts also recalled a twin
pair composed of one “junk food addict” and one “healthy eater” who
matched across most measures except for body weight.
British twins Mary Holmes and Elaine Allin, whose religious back-
grounds were also described earlier, had visited Boston for the study. Mary
wrote about the twins’ experiences in her wonderful 2008 book, Being You:
“They were totally obsessed with taking our vital signs . . . It appeared that
if a hair was out of place vital signs had to be taken.”31 Mary went on to
describe the twins’ last evening in Boston once the tests ended. All dressed
up and joined by Mary’s spouse, Tony, the twins spent hours at the bar
of the Jacob Wirth restaurant, enjoying Tropical Sunrises because the
bar had run out of their favorite, Bailey’s Irish Cream. Mary described
the rest of that evening. “I fell into bed but did not sleep. The room
spun, I felt horribly sick. Stultified by alcohol I stumbled blindly to the
bathroom. As I was throwing up I thought, with a measure of glee, ana-
lyse that.”32
What Mary did not know until the next day was that her vital signs had
been closely monitored for good reason. She had had a thyroidectomy a
few years prior, and the Tufts University physicians discovered that her
thyroid function was highly underactive, dangerously slowing down her
metabolism. She was advised to seek medical attention when she returned
home where she was put on medication. “I felt numb realising now why
they had stopped me from doing some of the physical tests. I felt grateful
and guilty simultaneously . . . Even though the intense research had been
a chore and the experience often disagreeable, I felt an overwhelming debt
of gratitude to Tufts University for their skilled and meticulous care.”33
The staff also looked out for other twin pairs with specific needs. One
of the few MZA twin pairs that was not part of the MISTRA had dif-
fered dramatically in their radiation exposure. One twin had spent time
in Chernobyl, assisting victims of the 1986 nuclear power disaster. Even
though the radiation emitted by the scheduled body scan was equivalent
280 B O R N TO G E T H E R — R E A R E D A PA RT

to just two days of natural background radiation exposure, the Tufts

University physicians omitted it in her case out of caution.
There are many reasons for people to be grateful to Tufts University for
findings that may significantly improve peoples’ lives. In 2009, Roberts’s
colleague, Dr. Sonya Elder, examined genetic and environmental influ-
ences on cardiovascular disease, the leading cause of death in the United
States.34 Significant genetic influence was found on all the risk factors she
measured. These risk factors included plasma lipid variables (56 to 77
percent), blood pressure (38 to 45 percent), and glucose and insulin mea-
sures (43 to 52 percent). Many people think that high-fat diets and seden-
tary lifestyles are the main reasons why some people develop heart disor-
ders, but Elder challenged this view by concluding that genetic variation
in the risk factors she measured is the primary reason why some people
have the disease and other people do not. But she maintained that small
dietary and lifestyle changes could significantly reduce cardiovascular risk
in people with a family history of heart disease.
Elder also wondered whether the metabolic syndrome, which involves
increased cardiovascular disease risk, hypertension, glucose intolerance,
and visceral fatness and affects over 34 percent of American adults, was
influenced by the same factors as cardiovascular disease. She determined
that waist circumference, plasma glucose, plasma triglyceride concentra-
tion, and diastolic blood pressure were affected by the same genes, pos-
sibly those relating to insulin resistance. Insulin resistance is a possible
underlying cause of the metabolic syndrome, so cardiovascular disease
and the metabolic syndrome could have a common origin.
The Federation of American Societies for Experimental Biology annu-
ally hosts small research groups to discuss results from ongoing projects.
These summer sessions are held in various locations around the world.
In 2009, Elder presented a study on the twins’ eating behaviors at one
of these meetings.35 Her study differed from Hur’s previous work that
had focused on the twins’ nutrient intake. Instead, Elder targeted the
twins’ hunger, disinhibition, and restraint. Hunger refers to feelings of be-
ing hungry and the behavioral consequences of these feelings. Hunger dif-
fers from disinhibition, which is a decrease in dietary self-control due to
emotions, alcohol, or other events. Restraint refers to the purposeful
limitation of food intake that ignores physiological cues. Factors affecting
these three eating behaviors could help, or hinder, weight-conscious indi-
viduals in their quest for a desired physique. The information was ob-
tained from true-false items in the Three-Factor Eating Questionnaire.36
Disinhibition was assessed by items such as “I usually eat too much at
social occasions, like parties and picnics.” Restraint was assessed by items
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 281

such as “I deliberately take small helpings as a means of controlling my

weight.” Hunger was assessed by items such as “I am usually so hungry
that I eat more than three meals a day.” Another part of the questionnaire
asked about the frequency of related eating behaviors.
Elder showed that a person’s environment was the most important
influence on disinhibition and on restraint. Disinhibition is a strong pre-
dictor of weight gain and can cause some people to add forty pounds
between twenty and fifty years of age, probably because metabolism be-
gins to slow down. Roberts was especially excited by the Disinhibition re-
sults. The finding that disinhibitory eating was largely learned gave Rob-
erts confidence that effective behavioral interventions to curb this type of
eating are likely. She explained that the learned component of disinhibi-
tion is the reward, the dopamine surge that follows this type of eating. But
she was surprised by the results, expecting cultural norms to have played
a greater role in disinhibition—cultures affect social conventions such as
appropriate meal times and acceptance of food from others.
It may be prudent to be more cautious than Roberts in thinking about
finding ways to control disinhibitory eating. Many learned behaviors, such
as closing a car door and stopping at a red light, would be difficult to
change. The effectiveness of eating-control interventions may rely partly
on genetically influenced personality traits such as conscientiousness and
agreeableness and thus would need to be tailored to each individual’s be-
havioral profile.
It was surprising to find that Disinhibition did not show genetic influ-
ence. Disinhibitory eating would seem to have an impulsivity compo-
nent, a behavior with demonstrated genetic effects (0.40).37 In reconcil-
ing these results, Roberts referred to Elder’s finding that some people are
more aware of their hunger than others; the internal locus for hunger had
a substantial genetic component (0.50).38 Roberts suggested that impul-
sivity may be situation specific and might not apply to eating situations.
This is an interesting idea, but one that needs testing. Impulsive eating
might lessen at formal dinners or family gatherings where eating habits
can be observed, but it may increase at home or at work when no one is
watching.
Roberts wished that her twin research could have continued as a lon-
gitudinal study rather than a cross-sectional one.39 “I am an intervention
person,” she said, meaning that studying people over time would help in
the development and monitoring of dietary controls. Roberts was also
excited, but overwhelmed by the quantity of accumulated data, confident
that ten more papers could be written. She listed the heritability of food
preference, energy intake, satiety, and physical activity as future topics,
282 B O R N TO G E T H E R — R E A R E D A PA RT

with the goal of exploring common genetic links between physical activ-
ity and body fat. She takes pride in having brought methodological
strength to her area of research.
Roberts seemed “hooked” on MZA twin studies and, like most inves-
tigators, regretful that she had not gathered more data at the time of the
study. What more could have been added? Fuss suggested that if the
study were conducted today, the researchers would have included addi-
tional genotyping components for linking DNA markers to body compo-
sition characteristics and chronic illness susceptibility. This would help to
identify people at risk for cardiac and metabolic diseases, allowing pre-
ventative measures to be introduced more quickly.
In summing up his experiences with the MZA twins, Fuss said, “I can
say without exception that there were no bad experiences. [Working with
the MISTRA twins] was a life-changing experience. It was unusual in
that you knew the twins so well in such a short time. More than any other
study I have worked on, the data were truly humanized.” Fuss also en-
joyed his role as “travel guide and tour agent” for many twins who were
visiting Boston for the first time.
During interviews with Roberts, Fuss, and many other MISTRA inves-
tigators, I found that the focus often shifted away from the research to
the twins’ life stories. For many of these researchers, it was probably the
first time that they experienced the bringing together of great scientific
stories and dramatic life events. “This study was hard to top,” Fuss con-
fessed. He is no longer in touch with the twins as before, but he loves
receiving their holiday cards.

Facial Expressivity
The efforts of many individuals enriched the breadth and scope of the
MISTRA. One such contributor to the study was psychologist Paul Ekman
from the University of California–San Francisco. Ekman is renowned for
his work on the universal linkage of human emotions and facial expres-
sivity. Because of him, eighteen MZA and ten DZA twin pairs watched
movies while research staff watched movies of them watching movies.
Other contributors to the study were psychiatrist Kenneth Kendler and
his colleagues from Virginia Commonwealth University in Richmond.
MZA twins Roger and Tony conveyed as much emotional content with
their faces as with their words. They were expressive and responsive in
conversations, making it possible to track their emotions as the topics
changed. Lucky and Dianne, Daphne and Barbara, and Sharon and Debbie
were just as demonstrative. These twins variously smiled broadly, rolled
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 283

their eyes, and shook their heads. Twins in some other pairs were harder
to “read,” because they showed fewer facial emotions during their inter-
actions with the staff.
As recently as 2007, little behavioral genetic work had been done on
individual differences in facial expressivity. The MISTRA’s Emotional
Reaction Study, designed by Ekman, was the first twin study to assess
genetic and environmental influence on spontaneous emotional expres-
sions.40 During the informed consent procedure on the first assessment
day, we told the twins that we would film them during one activity only,
but we would not say which one. We assured them that we would tell
them once the activity had ended.
Ekman selected three short films designed to elicit emotional reactions.
Twins first watched three one-minute clips of pleasant subjects, which
included a gorilla playing at the zoo, ocean waves hitting a beach, and
a puppy playing with a flower. These clips were followed by two two-
minute films, one of men hurt badly in an accident and one of a medical
training scene involving burn treatment and surgeries. A concealed video
camera captured the twins’ responses.
The Facial Action Coding System designed by Ekman and his col-
league Wally Friesen allows for the analysis of forty-four different action
units (expressions). Two certified judges viewed the films, each rating the
expressions of just one co-twin in a pair to avoid bias. For ease of analy-
sis with our small sample, the judges organized the action units into two
positive emotions (happiness and surprise) and three negative emotions
(sadness, anger, and disgust).
Genetic influence was found on the variability in positive emotional
expressions such as surprise but not for happiness. In contrast, genetic
effects on the variability in negative emotional expressions were not
found. It is likely that most of the twins showed similar signs of sadness,
anger, or disgust when viewing the accident victims or burn patients,
overwhelming any individual differences. It was also found that response
duration was more genetically influenced than response frequency. Per-
haps people react similarly to some events as they unfold, but vary in the
time taken to express or dampen their emotions. Regardless, the herita-
bility of the facial expressivity measures showed a wide range (35 to 75
percent), so additional work with larger samples is needed.
This study reinforced my observation that the MZA twins’ emotions
and emotional expressions matched when they first met. Looking closely
at some twins’ reunion tapes, I could see that the facial expressions of
joy, shock, interest, and disbelief were alike between co-twins. However,
I was especially interested in the reunion experience of Jack Yufe (the
284 B O R N TO G E T H E R — R E A R E D A PA RT

MZA twin raised Jewish in Trinidad) who met his twin brother Oskar
Stohr (raised Catholic in Nazi Germany) at age twenty. This event was
not taped, but Jack recalled that he was slightly embarrassed to see some-
one “wearing” his face.41 It may also be that Jack and Oskar’s initial fa-
cial expressions and his co-twin’s perceptions of his expressions shaped
the cool tone of their twinship at that time. A wonderful film, Oskar and
Jack, shows the twins together for the first time in Germany, but it does
not capture their moment of meeting.42
Oskar and Jack did not meet again until they came to Minnesota at
age forty-six. When they arrived at the Twin Cities airport on separate
flights, each twin wanted to see his brother before his brother saw him—
just as it was twenty-six years earlier.

General Intelligence and Special Abilities

Studies of general intelligence were a mainstay of the MISTRA’s research
program, and not just because the origins of intelligence had a long tradi-
tion in individual differences research. It was because ongoing controver-
sies over genetic explanations of ability, the structure of g, and the nature
of sex differences in some mental skills focused research attention on
intelligence.43 Everyone was interested in the MISTRA’s findings, even
those who criticized the ideas and methods. But what gets published also
reflects who is in the laboratory and when. Because former graduate stu-
dent and postdoctoral fellow Wendy Johnson was interested in the ori-
gins of general intelligence, she pursued these interests often and from
several angles.
Key articles in Johnson’s later series of intelligence studies were replica-
tion of evidence for “just one g”44 and replication of the Visual-Perceptual-
Image Rotation model of mental ability structure,45 referenced in Chapter
12. Another key accomplishment was a 2007 analysis of genetic and en-
vironmental influences on the Verbal-Perceptual-Image Rotation model.46
This was the most complete MISTRA study on intelligence since data
collection ended. It included seventy-four MZA twin pairs and fifty-two
DZA twin pairs, for a total of 126 pairs.47
The Verbal-Perceptual-Image Rotation model is composed of four
levels or strata. They include general intelligence or g as the fourth stra-
tum, three broad categories called Verbal, Perceptual, and Image Rotation
as the third stratum, eight specific abilities (e.g., Number Content Mem-
ory and Fluency) as the second stratum, and specialized skills that may
reflect specific learning experiences or events as the first stratum. The
strata are arranged hierarchically with g at the top. The second strata
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 285

factors are variously linked to the third strata factors; for example, Flu-
ency is associated with Verbal, and Number is associated with both Ver-
bal and Perceptual.
The portion of genetic variance associated with the factors at each strata
was high and consistent across measures (69 to 79 percent). The only ex-
ception concerned Content Memory, a second strata factor that showed
relatively little genetic influence (.33). Content Memory includes immedi-
ate and delayed recall of information, as tested by the ability to recall
pictures of common objects immediately after and some time after seeing
them. It also includes remembering meaningless and meaningful associa-
tions between things, as tested by the ability to recall pairs of figures and
numbers, and pairs of items and descriptors. It is possible that Content
Memory is affected more by experience than by genes. The best waiters
and waitresses are those who easily remember the orders from a table of
eight. Memory skills may develop the longer one is a restaurant employee,
but it is also possible that people with good memories seek positions that
let them use their talents.
The proportion of genetic variance shown by g was a curious number
(0.77). Anyone who recalls the Cyril Burt controversy involving Burt’s pos-
sibly fabricated reared-apart twins, assistants, and data knows the signifi-
cance of this number. Burt’s alleged fraud was suspected because several
replications of his MZA IQ analyses yielded correlations of 0.771.
Johnson’s paper had an impact on the way psychologists thought about
intelligence. Professor Tony Vernon believes that “Tables 6 and 7 in this
article will become classics,” based on the strong evidence for genes shared
among the Verbal-Perceptual-Image Rotation factors. In thinking about
her article, Johnson said, “The main response I got was that everyone
wanted to know what heritability we had gotten for the g factor, making
me realize how much that particular tidbit was buried within the paper.”
In fact, this figure (0.77) occurred in the last line of a substantial table,
and was referenced only briefly in the results. I’m referring to it now be-
cause it was exactly what Burt had found: a 0.77 heritability means that
77 percent of the general ability differences among people are associated
with genetic differences between them. It also means that about 23 percent
of the differences are associated with differences in their environments,
plus measurement error. Recall that studies of intelligence based on other
behavioral genetic designs tend to produce a lower estimate of genetic in-
fluence because they estimate heritability indirectly. The MZA correlation
estimates heritability directly.
The MZA and DZA g factor correlations were also omitted from the
article, but I obtained them from Bouchard when I visited Minnesota in
286 B O R N TO G E T H E R — R E A R E D A PA RT

October 2009 (MZA: 0.73, DZA: 0.47). I imagine that other readers
wanted them as well. The MZA g correlation was nearly identical to the
estimated heritability of 0.77. Bouchard also provided the twins’ WAIS IQ
correlations (MZA: 0.62, DZA: 0.50). The larger MZA than DZA cor-
relation was consistent with genetic influence on IQ, but the MZA value
and MZA-DZA difference were lower than what was found for the g
factor correlations. This may reflect the greater reliability of the g factor
score, but other explanations are possible.
By 2010, Johnson had thought more about other aspects of her study
since its original publication:

I’ve come to understand much more clearly how very little genetic correla-
tions can actually tell us about how systems of traits come together. If a ge-
netically influenced disease causes cognitive decline and that disease occurs
in a sample, there will be a genetic correlation between cognitive function
and the disease, even if it’s the disease that causes the cognitive problem di-
rectly, without any overtly common genes between cognitive function and
the disease.

For example, inherited eye disorders, such as the macular dystrophies,

cause visual loss in children, teenagers, and adults.48 Knowledge of im-
pending blindness understandably causes depression and apathy, behav-
iors that could interfere with mental performance. The connection be-
tween the eye disorder and either depression or lowered ability could be
conceivably, but incorrectly, explained by shared genetic factors.
Johnson continued, “So today, I might separately examine the heritabil-
ity of g and the patterns of genetic correlations, with specific emphasis on
all their possible interpretations.” I am sure that discussion of these issues
will affect how researchers interpret their findings in the future. Even
genetic appearance can be deceptive. Johnson has since addressed these
topics in subsequent research papers.49
Although it is not reported in Johnson’s article, the IQ differences be-
tween MZA co-twins are of interest, relative to those from previous MZT
and MZA twin studies. The average IQ difference between MZT twins is
6 points, although the range can be considerable.50 The average within-
pair IQ differences for twins in the three previous studies varied from
7.25 to 14.00 points, with a range of 1 to 24 points for individual pairs.
The within-pair difference for MZA twins (n = 70 pairs) in the MISTRA
was 7.07 (standard deviation = 5.83), with a range of 0 to 29 points.
These findings further underline the prenatal and postnatal environmen-
tal effects on general intelligence, the importance of assessing cognitive
ability using different samples, and the need to include multiple ability
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 287

measures in psychological research. The DZA within-pair IQ difference

for the MISTRA twins (n = 53 pairs) was 8.78 (standard deviation = 7.09),
with a range of 0 to 25. The average DZT twin difference is 10 points,
but also varies across studies.51
Bouchard included a simple, but extraordinary table in his 2009 review
paper of research on g.52 The table lists the MZA general intelligence cor-
relations from the three early twin studies, the Swedish Adult Twin Study
of Aging, and the MISTRA. Collectively, these studies provide further
support for the finding that individual differences in intelligence have a
substantial genetic basis. The weighted mean of the IQ correlations from
these five different studies, conducted in four different countries, is 0.74
(n = 187 pairs). This was another example of constructive replication, as
was Johnson’s article reporting a value of 0.77 for the g factor heritability,
in which Johnson used different measures of the twins’ intelligence levels
than the other studies. Results from different psychological investiga-
tions rarely come that close.
Because of the quality, breadth, and availability of the twin and spouse
data, Johnson and Bouchard also used it to explore sex differences in
mental skills as well as relationships among general ability, specific abilities,
and interests. Some of this work did not have a “twin angle”—nontwins
could have been used in some analyses, assuming they had completed the
same extensive test battery.
Sex differences in special mental abilities, but not in general intelli-
gence, have been observed for decades. Males tend to excel in visual
and spatial tasks, while females generally excel in verbal and perceptual
speed tasks.53 These are average differences only—there is plenty of over-
lap among the sexes, making it impossible to know someone’s sex from
his or her test scores. A given male could perform above the female mean
on a verbal fluency test, and a given female could perform above the male
mean on a mental rotation test. In fact, our highest scorer on the Shepard-
Metzler Rotation task, an activity in which males typical excel, was a
female twin.
Building upon her Verbal-Perceptual-Image Rotation model studies,
Johnson examined the structure of mental abilities after statistically con-
trolling for the effects of general intelligence.54 She relied on data from
the forty-two mental ability tests comprising the Hawaii Ability Battery,
Comprehensive Ability Battery, and WAIS IQ test. Three underlying dimen-
sions emerged—rotation-verbal, focus-diffusion, and content memory—
each involving different abilities. People with different mental skills are
found at different places along these dimensions. For example, people
close to the rotation pole of the rotation-verbal dimension did better at
288 B O R N TO G E T H E R — R E A R E D A PA RT

spatial and rotation tasks than they did on vocabulary, spelling and ver-
bal interpretation tasks, and vice versa.
Johnson found that males clustered toward the rotation and focus
poles of the rotation-verbal and focus-diffusion dimensions, while fe-
males tended toward the verbal and diffusion poles. Females also clus-
tered toward the “good memory” pole of the content memory dimension
more often than males. None of this was surprising. These results matched
the findings in the psychological literature showing small, but consistent
sex differences in these measures. Of course, both males and females are
expected to be found at any position along the poles, just not in equal
numbers.
The more interesting results from this study concerned sex differ-
ences in the residual scores. Residual scores were derived by stripping the
forty-two mental ability test scores of the effects of age, and both age and
g. Once this was done, sex differences in the residual scores were greater
than the sex differences in the original scores. The fact that sex differ-
ences in ability increased once g was removed suggested that g may rep-
resent a general problem solving ability that hides sex-related differences
in specific skills. The residual scores also showed similar degrees of gene-
tic influence (0.38–0.70), as did the original test scores (0.33–0.79), with
Memory Content again showing the lowest heritability (0.38). This re-
sult suggested that residual abilities follow systematic patterns that are
biologically rooted in the brain.
Johnson produced evidence that the tests did not measure the same
things in the same way for people positioned differently along the poles.
People presented with the same problem might try to solve it differently.
One person might mentally rotate an image to determine its appearance
from another angle, while someone else might draw a series of figures
to try to find a solution. One person might take a “mental snapshot” of
an image he or she wished to remember, while someone else might ver-
bally describe the image to himself or herself.
The origins of sex differences in mental ability have yet to be fully
known, but sex differences in brain structures, physiological functions, and
neurological features might hold answers, the subject of a subsequent pa-
per by Johnson and Bouchard.55 Male brains are more lateralized than
female brains, meaning that males’ different mental functions are found
more exclusively in the brains’ left hemisphere (verbal) and right hemi-
sphere (nonverbal). Females’ brains are relatively less lateralized, so their
verbal and nonverbal functions are distributed across both hemispheres
to a greater degree than are males’.56 Hormonal influences also affect
mental functioning. For example, females show better spatial skill per-
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 289

formance during the menstrual phase of their menstrual cycle compared

to the midluteal phase, consistent with normal fluctuations of estradiol
and progesterone.57
Biological measures could be examined in conjunction with psycho-
metric measures. Substantial correlations have been reported between
general mental ability, total brain volume, and volumes of gray and white
matter in specific brain regions.58 Johnson created two novel WAIS
scores ([Block Design − Vocabulary]/Full Scale IQ and [Block Design + In-
formation − Coding − Digit Span]/Full Scale IQ) that correlated highly
with the rotation-verbal dimension (0.69). She suggested that these novel
scores could be useful in biological brain studies.59 For example, it may
be that these scores, which are associated with peoples’ positions along the
rotation-verbal pole, reflect sex differences in brain structures and pro-
cesses, an idea that could be tested with brain-imaging studies. In 2008,
Johnson’s group went on to show associations between the rotation-verbal
and focus-diffusion dimensions and gray matter in two brain regions.60
There are many studies on the nature and origin of mental skills, inter-
ests, and occupational pursuits. However, these areas have remained
largely separate from one another. Understanding associations between
mental abilities and interest-occupation groups was the focus of the next
MISTRA study.

There is a common belief that people are interested in activities in which

they perform well. But the causal direction is unclear—are we interested
in what we are good at, or are we good at what we are interested in? It is
easy to understand how we might look for opportunities that allow ex-
pression of our best skills. MZA twins Tom and Steve (who were not in
the MISTRA) were both weight lifters who owned their own gyms, one
in New Jersey and the other in Kansas. Tamara and Adriana performed
separate modern dance routines, one in New York City and the other on
Long Island, until their chance reunion let them dance together.61 The
twins in both these sets had the body builds conducive to the physical
pursuits that engaged them, as well as the resources and support. But
exposure to activities is not enough. The fireman twins, Mark and Jerry,
never adopted their mothers’ interests in opera and ballet, openly show-
ing disdain for these activities. The twins preferred going fishing, watch-
ing television, and drinking beer with their firehouse friends. Perhaps
there were commonalities among these twins’ interests and occupations.
Johnson used latent class analysis to find out.
Latent class analysis is a statistical procedure that provides groupings
of individuals, rather than groupings of variables. For example, many
290 B O R N TO G E T H E R — R E A R E D A PA RT

people are interested in athletics, but some people like noncompetitive

activities such as individual weight-training or floor exercise. These
people might also enjoy editing manuscripts or computer programming,
occupations that are practiced indoors and are noncompetitive. Simi-
larly, Johnson was interested in identifying groups of people whose inter-
ests and occupations had common features.62 She was also interested in
whether sex differences in mental skills affected sex differences across
interest-occupation groups. MISTRA data from the Strong-Campbell
Interest Inventory and Jackson Vocational Interest Survey were available
for this purpose. Johnson also used the twins’ mental ability test scores
derived from the Verbal-Perceptual-Image Rotation model analyses dis-
cussed previously.
Johnson’s study was exploratory in nature with no specified hypothe-
ses or predictions. Using latent class analysis, she created eight interest-
occupational groups, ranging from “adventure” to “personal care,” but
there was nothing magical about this number—her choice of groups
generated by the data was somewhat subjective and practical.
Johnson found greater differences among interest groups in general in-
telligence than in abilities. Higher intelligence groups also showed greater
diversity of interests than lower intelligence groups. More job options
may, therefore, be available for smarter people who, as Johnson notes,
probably capitalize on the features of their jobs that they enjoy most. In
fact, most college professors love the intellectual challenge inherent in their
occupation, but variously prefer its teaching or research opportunities.
Johnson also discovered greater sex differences in the twins’ interests
than in their abilities. Overall, fifty-six percent of men were classified into
groups that included 1 percent of women, and 70 percent of women were
classified in groups that included 7 percent of men. Interest-occupation
groups that were especially sex-polarized were Adventure (0.01 women,
0.99 men) and Administration (0.94 women, 0.06 men).63 Social pres-
sures and biological factors, but probably both, could account for this
finding. Selective attention to events in the environment may also con-
tribute to these differences. We know that female infants prefer to look at
faces and male infants prefer to stare at mobiles.64 Male infants are also
more successful than female infants on age-appropriate mental rotation
tasks.65 If and how these early sex differences lay the basis for later sex
differences in verbal and spatial skills are unanswered. One possibility is
that personality traits mediate our choice of activities. The data to test
this idea exist, and maybe one day Johnson will get to them.
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 291

The Traditional Moral Values Triad

The MISTRA’s poststudy period from 2006 and beyond included studies
on many unrelated topics, as in the preceding years. There was value to
this variety; it reflected the wide range of behaviors examined by the col-
laborators when the data were available and when they had the time
to explore it. Throughout this book I have emphasized empirical twin
analyses, but provocative themes were also found in some of the chapters
published in volumes organized around special topics. Unfortunately,
such volumes draw less attention because they often take longer to pub-
lish than journal articles, making the information somewhat dated, and
these books are expensive. In two MISTRA-related book chapters on reli-
gious behavior and beliefs, Bouchard and his former student Laura Koenig
brought name and definition to one of Bouchard’s main interests—the
Traditional Moral Values Triad.
It was unusual for Bouchard to think in threes after years of thinking
in twos, but the research findings were responsible for this one adjust-
ment. Psychologists had defined three major attitudinal and belief dimen-
sions, namely, authoritarianism, religiousness, and conservatism, all of
which the MISTRA had studied separately or in combination.66 Bouchard
united them in a bigger way as traditionalism or the Traditional Moral
Values Triad. His justification was not just the 0.50–0.70 correlation among
the various measures of these dimensions—he used evolutionary reason-
ing to explain why the three formed a likely and sensible cluster.67
In the early and final versions of his book chapters, written in 2009 and
in 2010, Bouchard asked three simple but profound questions that he be-
lieves are confronted by children in modern society: (1) “Who is in charge?”
(2) “What does he/she want?” and (3) “What do I do?” All three questions
concern family and group relations, and probably arose as human society
and human intelligence evolved.
Bouchard identified the adaptive human tendency to obey authority as
the mechanism that could illuminate the three questions posed above. If
the reference is the family, then the answer to question one is the parent
or caretaker. In support of this idea, I recall an amusing but telling con-
versation between one of my colleagues and her belligerent seven-year-
old son to whom she asserted forcefully, “I am the adult, and I make the
rules.” Alternatively, if the reference is the greater universe, then the an-
swer is supernatural beliefs, God(s), or agents outside the visible world.
Bouchard noted that such views were reasonable in light of available
knowledge. In fact, he noted that his own “favorite entities, genes,” re-
garded as having causal influence on behavior, were neither visible nor
tangible until genetic advances made them so.
292 B O R N TO G E T H E R — R E A R E D A PA RT

Bouchard asserted that obedience underlies traditionalism and is a

likely adaptation, given its universality in childhood. Not all children are
equally obedient, and some are defiant, but those who follow the rules
are generally more socially successful than those who do not. Evidence
suggesting that Traditionalism may enhance reproductive fitness comes
from links between higher conservatism, higher fertility, and earlier child-
bearing as well as genetic influences on fertility and social attitudes.68 The
studies on age at menarche and age at first intercourse, described earlier
in this chapter and elsewhere, showed genetic effects. No one knows for
certain how attitudes and fitness measures mesh, but Bouchard believes
that twins can provide answers: longitudinal analyses of twins’ reproduc-
tive fitness correlates and social attitudes could clarify connections between
the two. Does having children enhance conservative attitudes in their par-
ents, or does being conservative enhance childbearing? Longitudinal stud-
ies do not reveal causal connections, only experimental studies do—but
twin studies are experiments that are created naturally.69

Bouchard and colleagues authored several chapters and reviews on intel-

ligence, intelligence tests, heritability, and other topics between 2006 and
2010. A survey of organizational behavior showed that twin research
had advanced since our 1989 job satisfaction paper, but many behaviors
such as factors affecting job and occupational switching were still un-
explored.70 The job histories that were gathered from the twins, begin-
ning with the part-time jobs they held while still in school, would enable
such a study. Perhaps a MISTRA colleague reading this will be inspired
to take a closer look.
Bouchard also authored a tribute to developmental behavioral geneticist
Sandra Scarr, arguing for strong inference as a strategy for scientific ad-
vance.71 Based on a 1964 article by John R. Platt,72 strong inference in-
cludes the systematic application of alternative hypotheses, the design of
multiple experiments, the carrying out of these experiments, and the re-
peating of this process. Bouchard identified quantitative genetic analysis as
exemplary of the strategy of strong inference.

Birth Order
In 2008, psychologist Shirley McGuire from the University of San Fran-
cisco and I attempted a study that I had thought about for some time—a
test of Frank Sulloway’s birth-order theory using MZA twins.73 Sulloway
showed that firstborns tend to be traditional and conventional, while later-
borns tend to be risk-taking and rebellious.74 It appeared that a novel test
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 293

of his model could be done using an MZA twin design because birth-order
effects would be unconfounded by genetic influences on personality. Many
co-twins occupied different birth-order positions in their separate rearing
families. We reasoned that if MZA co-twins placed in different birth-order
positions showed Sulloway’s expected birth order and personality relation-
ships, this would add to research demonstrating sibship position effects on
personality. Birth order could then be considered part of the nonshared
environmental influences on personality, partly explaining differences
between co-twins. We used birth order and personality data from the
MISTRA and from the Swedish Adoption Twin Study of Aging.
We found that co-twins in different birth-order positions did not show
meaningful personality differences, but the sample sizes were too small
to permit firm conclusions. In fact, the small numbers of twins in the
various birth-order combinations reduced our comparisons to just same
birth-order versus different birth-order groups. Moreover, our personality
data were self-ratings rather than ratings made specifically in comparison
with the co-twin, which are more sensitive to birth-order effects.75 Other
factors, such as twins’ age at adoption, gender of siblings, age spacing,
and rearing status of siblings (biological or adoptive) also needed to be
considered. Some studies are theoretically compelling but practically un-
doable, and this may be one of them, at least at this time.

Handedness
The MISTRA’s handedness data were not published at the time of this
writing, but they were important to include in this book, given the gen-
erally higher proportion of left-handedness in twins than nontwins (14.5
percent versus 9.9 percent)76 and the similar frequency of left-handedness
in MZ and DZ twin pairs.77 Increased left-handedness in twins has been
variously linked to delayed zygotic splitting among approximately 10
percent to 25 percent of MZ twins, the increased prenatal adversities
(e.g., premature birth) affecting both types of twins (but especially MZ
twins), and associations between familial left-handedness and twinning.78
However, a 2003 study questioned the higher rate of left-handedness in
twins, having found similar rates among twins and their nontwin
siblings.79
The MISTRA twins completed the Oldfield Handedness Inventory, a
twenty-item self-report form that asks subjects how often they use their
left or right hand to perform various unimanual (e.g., writing and drawing)
and bimanual (e.g., peeling a potato and opening a bottle) tasks.80 Rat-
ings are made on a nine-point scale (1 = left always to 9 = right always).
294 B O R N TO G E T H E R — R E A R E D A PA RT

Beginning in 1982, we assessed the twins’ hand preference, both by ob-

serving them perform the tasks specified by the twenty items and admin-
istering the self-report form at another time. This procedure let us assess
the validity of the self-report; item analysis for the first 113 individual
twins showed inconsistencies in some items, although not necessarily in
handedness classification. The analyses reported here used the behavioral
assessment data except for cases in which only the self-report form was
available.81
Handedness data were available for 247 twin individuals, 94 males and
153 females. The twins’ handedness preferences were categorized using
the Crovitz-Zener scoring and classification scheme (right: 14 to 30,
mixed: 31 to 40, left: 41 to 70) based on the first fourteen items of our
inventory.82 There were 200 right-handers (81 percent), 24 left-handers
(10 percent), and 23 mixed-handers (9 percent) among the MISTRA
twins. Organizing the twins according to right-handedness and non-
right-handedness yielded 47 (19 percent) non-right-handers, nearly twice
as many as the expected 10 percent of non-right-handers in the general
population. The proportions of non-right-handed twin individuals were
similar for the MZA (16 percent) and DZA (23 percent) twins.83 The
proportions of hand-concordant and hand-discordant pairs, across hand-
edness categories and overall, are shown in Table 13-1. The proportions
of nonconcordant pairs did not differ between the MZA (24 percent)
and DZA (35 percent) twins.84
The 24 percent hand-discordance among the MZA twin pairs matched
the 25 percent figure reported in the literature for MZT twin pairs. The
35 percent hand-concordance of DZA twins is somewhat high but con-
sistent with the equal number of non-right-handedness in both twin
groups. Overall, these early results agree with previous findings of elevated

Table 13-1. Handedness Concordance and Discordance for MZA and DZA Twin Pairs

MZA (%) DZA (%)

Handedness 74 pairs 55 pairs

Right-left 15 18
Right-mixed 8 15
Left-left 0 2
Left-mixed 1 2
Mixed-mixed 3 5
Right-right 73 58
Hand-concordant 76 65
Hand-discordant 24 35
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 295

left-handedness in reared-together twin samples. Our MZA handedness

findings also agree with those of Shields and Juel-Nielsen who observed
hand-discordance in 16 percent and 18 percent85 of their samples, respec-
tively; in contrast, 45 percent of twenty MZA twin pairs studied by New-
man et al. (original nineteen pairs plus one pair identified after the study
was published) were opposite-handed. Methodological differences may
explain the discrepancy—Newman et al. assessed handedness by wrist
tapping and finger tapping tasks, whereas Shields administered handed-
ness items and a handedness questionnaire, similar to the MISTRA.86
Juel-Nielsen did not specify his handedness assessment procedure.
The reared-apart twins’ handedness data are relevant to the funda-
mental principle underlying twin methodology, namely, that the trait-
relevant environments of MZ and DZ twins must be equivalent in order
to examine the heritability of a trait. However, a subset of MZ twins
differ in handedness because of delayed division of the zygote, a process
that does not apply to DZ twins. Therefore, it would be inappropriate to
use a conventional MZ-DZ twin comparison to assess genetic influence
on hand preference, as I will illustrate.
In 1957, biologist David Merrell observed, “The discordance between
identical twins proves the importance of environmental factors while the
discordance between a left-handed child and his completely right-handed
cultural and familial environment indicate the importance of innate fac-
tors in the development of handedness.”87 Subsequent family and adop-
tion studies of handedness have shown genetic influence on hand prefer-
ence,88 but conventional twin comparisons would suggest otherwise—as
would the data from the MISTRA.

Mental Ability Again

In late March 2011, Irving Gottesman sent me a new mental ability pa-
per from the MISTRA, authored by Johnson and Bouchard.89 This paper
was neither a literature review nor an empirical analysis. It began with a
concise survey of research on the structure of human ability and a brief
description of the MISTRA’s mental ability measures and findings from
previous reports. A new finding concerned the adjustment of the sample’s
mean IQ score from 109.7 (standard deviation = 11.8; range: 79 to 140),
based on 1995 norms, to 101.3 (standard deviation = 14.8; range: 61.1
to 139.9), based on the average rate of secular change.
But this new paper was unlike any other that Bouchard or anyone else
from the study had published previously because it furnished the correla-
tion matrix generated by the forty-two mental ability tests completed by
296 B O R N TO G E T H E R — R E A R E D A PA RT

the MISTRA participants. Johnson and Bouchard explained, “The pur-

pose of this paper is to make the MISTRA dataset available to other re-
searchers so that it can be used both to generate new ways of thinking
about mental ability and to attempt to replicate findings generated in
other datasets.”90 The inclusion of the data matrix was surprising at first
because the MISTRA data were hard to come by and the data had never
been publicly released—but Bouchard was all about replication, and the
time was right. It was interesting to speculate as to what would be said
by critics who claimed Bouchard was too protective of his data.
A second 2011 paper by doctoral student Jason Major at the Univer-
sity of Edinburgh, together with Johnson and Bouchard, used the reared-
apart twins’ mental ability test data to show why single-factor models do
not adequately represent g.91 Data supplementary to this article are avail-
able online.

Nearly Final Comment

It is almost impossible to believe that, by this point in the book, I have
surveyed the MISTRA’s entire publication list. The most satisfying part
of this process has been the long conversations with friends and col-
leagues who couldn’t resist revisiting this exciting and challenging time
in their careers. It is gratifying to have rekindled the interest of some
collaborators in data they had forgotten about or didn’t know existed.
Michalowicz will be exploring associations between oral diseases (e.g.,
tooth loss) and cardiovascular health (e.g., blood pressure) using data from
his periodontal studies and the Tufts University body composition battery.
He explained that relationships between periodontal disease and heart
disease may arise through common genes. Hanson believes the twins’
stress test data are unique and should be analyzed for measures such as
heart rate and blood pressure at each three-minute stage. Few people com-
plete a 21-minute graded exercise test, so the variability in these measures
might reflect genetic factors, with implications for health and training.
Johnson wants to examine assortative mating for intelligence and the
stability of g across test batteries. The collaborators at Tufts University
are planning a paper on predictors of energy intake and dietary composi-
tion using the twin data, an effort led by doctoral student Lorien Urban.
Sonya Elder, also from Tufts University, will be completing her study of the
twins’ eating behaviors. Bouchard is planning an analysis of the twins’
creativity involving their drawings of a person and a house, captured on
videotape. I hope to explore associations between the twins’ spatial abili-
ties and hand preference. Several articles and chapters on the heritability
S E X U A L D E V E L O P M E N T, A S Y M M E T R Y, B O D Y S I Z E , I N T E L L I G E N C E 297

of behavior, general intelligence, body fatness, and religiosity are in press

or under review.92 I also hope to explore Bouchard’s treasury of tapes if
and when they are transformed to DVD. In addition to the Juel-Nielsen
interview described in Chapter 3, I watched the only other tape that had
been converted—a 1979 semistructured life events interview with one of
the Jim twins by Tellegen and Gottesman. Such interviews were informa-
tive sessions, revealing significant features of the twins’ lives and showing
their movements and gestures as they spoke.
The ongoing and planned analyses of the reared-apart data show that,
although the MISTRA may be complete, it is not really over. The impact
of this project on research and reasoning about the origins of individual
differences in intelligence, personality, attitudes, interests, values, talents,
and so much more will undoubtedly persist far into the future.
 CHAPTER FOURTEEN

Questions, Answers, and Twin Studies of the Future

N o one could satisfy the MISTRA’s critics while the study

was in progress, but I have the benefit of time. Over thirty years
have passed since the project began, and over ten years have
passed since it ended. The bitter 1970s nature-nurture debates have
largely (but not completely) subsided as evidence of genetic influence on
behavior has accumulated. Most serious scholars have accepted the body
of work produced by the MISTRA and related studies, and many indi-
viduals have rethought the significance of their own findings in light of
this work. Researchers have moved on to explore new and interesting
questions about factors affecting behavior, some of which I describe at
the end of this chapter. What follows are summaries of the challenges,
objections, and accusations voiced by the MISTRA’s critics and responses
to them. Some issues discussed earlier are revisited here because of their
importance—and because some serious misunderstandings of twin re-
search designs, findings, and implications remain.1

The MISTRA’s Data Collection Procedures: Fair or Biased?

The MISTRA’s data collection procedures were criticized based on as-
sumptions rather than facts. Some critics assumed that the twins exchanged
information during the assessment, the intelligence tests and personality
inventories were flawed, and the occasional use of the same tester for
both co-twins yielded biased data.
Beginning with the first assessment day and as part of the informed
consent procedure, Bouchard routinely requested that the twins refrain
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 299

from discussing the research activities until both had completed them.
Everyone was confident that the twins and their companions understood
the reason for this and fully complied.
To the extent that the staff and budget allowed, every effort was made
to keep the data free from potential biases by following procedures set by
Bouchard. Twins were always seated apart as they completed their inven-
tory booklet in the MISTRA’s laboratory or elsewhere. A staff member
was always present during these sessions to answer the twins’ questions
privately, if any arose. Two staff members independently reviewed each
twin’s inventories for completion and clarity, noting questions that needed
additional attention from the twins.
The use of separate examiners (unrelated to the study) and simultane-
ous testing of twins were strictly followed for the IQ testing because of
its controversial nature and because the early reared-apart twin investi-
gators had been criticized for sometimes testing both twins. Simultane-
ous and separate (or consecutive) testing by different examiners was in
place for most activities, with some exceptions, such as the medical/
psychiatric interviews administered to the early twin pairs and David
Lykken’s psychophysiological battery that the twins did in reverse order
on different days. Kevin Haroian who tested both twins in Lykken’s lab-
oratory told me that “the computer controlled the presentation. We never
saw any systematic differences between the twin tested on Monday and
the twin tested on Tuesday.” Several other exceptions included taking the
joint twin photographs, videotaping a joint interview immediately after
the individual interviews, and administering the special mental ability
battery (due to concerns over examiner differences in administration).
Some tests, such as the general physical examination, chest x-ray, stress
electrocardiogram, special mental ability battery, and psychomotor tasks,
were administered to co-twins by the same physician, psychologist, or
staff member. Testing related individuals can conceivably bias research-
ers’ expectations for the second participant based on the performance of
the first. However, what seems like reasonable inference can prove insig-
nificant upon closer inspection.
Bouchard pointed out long ago that those who raised such objections
were the same people who reexamined the data with full knowledge of
the twins’ IQ differences.2 In other words, critics who categorized the
early pairs by age at separation, contact, and other factors had access to
these twins’ IQ scores, so their categorization may not have been done
blindly.
The burden of proof lies with the critics—bias must be demonstrated,
not assumed. The critics of reared-apart twin studies never conducted a
300 B O R N TO G E T H E R — R E A R E D A PA RT

formal test of their assumptions, but I did. In 1991, I compared IQ intra-

class correlations for MZT and DZT twin children (all of whom I tested
myself) with twin children from a different sample (in which I tested only
one twin). The intraclass correlations hardly differed between sample 1
(MZT: 0.84, n = 68; .DZT: 0.54, n = 34) and sample 2 (MZT: 0.88,
n = 28; DZT: 0.54, n = 34). I concluded that when examiners follow
well-prescribed administrative and scoring rules tester bias is negligible
or absent.3
Several of the collaborators commented on the possibility of bias in their
particular assessments. Paul Fuss, the project manager of the Tufts clinical
nutrition study, explained that the same two testers recorded each twin’s
anthropometric measurements. Both examiners assessed each twin in re-
verse order, and the data were averaged together. Fuss emphasized that
testers did not check the results until data were available for both twins.
The same individual recorded each twins’ maximal oxygen uptake, but
Fuss said it was difficult to see how an investigator could bias a machine.
Leonard Heston administered medical interviews to both twins until
he left Minnesota in 1990. “I presented the questions in the same way to
each individual—doctors learn to do this during their second year of
medical school. There was a possibility of bias, but if so it did not operate
in an important way.” Heston’s comment offers a realistic understanding
of potential bias in the project’s methods and its effects on the data. No
data are perfect, but given the care with which each individual’s informa-
tion was gathered there is little chance that the MISTRA findings were
inaccurate or misleading in any way. The lack of bias shown by my IQ
test experiment most likely applies to findings in other behavioral do-
mains. The MISTRA’s findings generally showed excellent agreement with
those from other twin studies, which would not have happened with poorly
gathered data.
The intelligence tests, personality inventories, and most of the interest
questionnaires that were administered had been used widely in prior re-
search. The reliability and validity of these instruments had been well
established. Bouchard’s choice of these instruments was largely guided by
their use in previous twin studies so that they could enable comparative
findings.
The results from the IQ data analyses were extremely consistent with
the findings from the three previous reared-apart twin studies and the
study conducted in Sweden. Furthermore, twin data are more informa-
tive in conjunction with data from other kinships varying in genetic and
environmental relatedness. It is difficult to explain the much greater IQ
similarity of MZA twins (0.74), relative to that of DZT twins (0.60), bio-
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 301

logical parent-child pairs (0.42), half-siblings (0.31), adoptive siblings

(0.29–0.34), virtual twins (0.28), adoptive parent-child pairs (0.19), and
cousins (0.15) without reference to genetic factors.4 In fact, there has
never been a convincing environmental explanation as to why genetically
identical individuals meeting as adults are more behaviorally alike than
unrelated individuals raised together. Research also shows that unrelated
individuals raised together become less alike in intelligence over time.5 A
consistent result from the MISTRA’s analyses was that features of the
twins’ rearing homes made negligible contributions to their behavioral
resemblance. This point was often lost in some of the harsh media treat-
ments of the MISTRA.
As an aside, when Bouchard delivered his last lecture to University
of Minnesota students on April 16, 2010, I was there, as was my former
postdoctoral colleague Kimerly Wilcox. When Bouchard showed the 0.77
MZA IQ (g factor) on the screen, Wilcox sent me a knowing smile. We
both remembered that we had given Bouchard a t-shirt with the number
“.771” printed across the front, as I mentioned earlier. Even this contro-
versial figure associated with Sir Cyril Burt’s twin study remains a robust
finding.
Some critics argued that personality inventories do not faithfully cap-
ture the behaviors people express in real life. Of course, inventories are
imperfect, but they are superior to observational data gathered on just a
few unrepresentative occasions. Inventory respondents indicate traits,
attitudes, and interests that are most characteristic of their “self,” provid-
ing a summary view of their behavior over time. However, depending
upon the investigators’ goals, observational studies can be very informa-
tive. I am a long-time member of the International Society for Human
Ethology, whose members study behavior by naturalistic experimenta-
tion and observation. High-quality ethological studies require a large
number of subjects for testing hypotheses about very specific behaviors.
The reared-apart twins were in Minnesota for only one week, so their
time and ours had to be used very efficiently to learn as much about them
as possible. Had the twins’ personality and interest assessments been
limited to a few observational sessions, the likely yield would have been
unrepresentative or atypical material. Moreover, the MISTRA was ex-
ploratory in nature, so an observational approach was not suited to the
study’s goal. However, in addition to the standard interviews and inven-
tories that were administered, the twins’ facial expressions were recorded
while they watched films, they were photographed in unposed standing
postures on the first assessment day, and they were videotaped in un-
structured individual and joint interviews.
302 B O R N TO G E T H E R — R E A R E D A PA RT

Every major domain of behavior that was assessed, such as intelligence,

personality, and attitudes, was accomplished by multiple measures. Within-
study constructive replications generally produced comparable findings,
which could only have occurred with good-quality data. A final point is
that the standard measures that were used, such as the Wechsler Adult
Intelligence Scale and Minnesota Multiphasic Personality Inventory among
others, had been designed to assess individual differences in behavior
with no reference to genetic effects. The fact that the MISTRA and other
twin researchers found genetic influence on these measures further sup-
ports the unbiased nature of the data. Of course, despite effective safe-
guards, biased data collection is possible in any psychological study.
In 1974 correspondence, James Shields (investigator of the 1962 Brit-
ish reared-apart twin study) addressed similar criticisms of his work that
were raised by Leon Kamin (the individual who had ignited the contro-
versy over Cyril Burt). Shields wrote, “The criticisms boil down to poor
tests, experimenter bias, and insufficient separation of the twins. Even if
these criticisms had as much force as Kamin claims, they would not lead
to the conclusion of zero heritability, but rather that the degree of herita-
bility cannot be estimated satisfactorily.”6

The Reared-Apart Twins: Was the Sample Size Problematic?

The fact that the reared-apart twins comprised a collection of cases rather
than a systematically gathered sample was always acknowledged by the
MISTRA investigators. I used the term “sample” throughout this book for
the sake of convenience only, as explained in Chapter 2.
Many critics claimed that the reared-apart twin sample was too small
to generate valid findings. There is little question that twin samples com-
posed of several hundred pairs or more are desirable in psychological
research. However, the United States lacked a twin registry, which forced
Bouchard to rely on recruitment methods that yielded a smaller number
of pairs. Nevertheless, the quality of the twin pairs that were studied was
excellent—the sets were separated earlier in life and reunited later than
twins in the three previous reared-apart twin studies and in the Swedish
study whose researchers benefited from registry access. The MISTRA
twins were also older at the time of assessment than the twins in the in-
vestigations by Newman et al. and Shields.
Sample size can, and did, pose difficulties for some analyses. The smaller
number of subjects sometimes prevented assessments of the differences
among twins organized by sex and zygosity, as in Johnson et al.’s 2007
mental ability study.7 The DZA twins showed higher than expected
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 303

resemblance on militarism and on some mental ability measures. An-

other problem linked to the modest sample size was the large variation in
heritability estimates, as in Johnson et al.’s 2005 study of intelligence and
reading performance.8 These DZA twin findings could have reflected as-
sortative mating, placement effects, or sample fluctuation. Placement
effects could usually be ruled out as explanations, but the DZA twins’
similarity on some measures could not be resolved completely; therefore,
a larger DZA twin sample would have been desirable. These limitations
were discussed in the publications, noting that the DZA twin pairs were
more often typical than atypical of DZ twins generally.
It is important, but also unappreciated, that heritability estimates are
more accurate and efficient when they are based on data from twins reared
apart than twins reared together, and can be derived from fewer numbers
of pairs. These points were developed by Lykken and his colleagues in an
unpublished 1981 manuscript.9 Lykken showed that only fifty MZA twin
pairs, compared with four to five hundred MZT twin pairs and four to
five hundred DZT twin pairs, can estimate heritability with the same de-
gree of confidence. That is because heritability is reflected directly by the
MZA intraclass correlation, whereas heritability is calculated indirectly
by doubling the difference between the MZT and DZT intraclass correla-
tions. The MISTRA had studied eighty-one MZA twin pairs by the end of
the study, thirty-one more than the fifty MZA twin pairs specified by Lyk-
ken. Improving methods for estimating heritability was a goal of Lykken’s
throughout his career.10

MZA Twin Resemblance: Genetic Influence or Selection of Similar Pairs?

In their 1937 book, Newman et al. recalled:

The first case studied by us of identical twins reared apart gained consider-
able publicity through no fault of ours, for the twins themselves gave their
photograph and their life stories to an enterprising local reporter, who sent
his news story to an American newspaper, from which it was copied far and
wide. Since our name and address were given, we were deluged with all
sorts of letters from twins and about twins, a few of which furnished valu-
able clues.11

Bouchard’s experience of identifying twins, especially in the early

years, did not differ greatly from that of Newman et al. with respect
to the press. The Jim twins were featured in national and international
newspapers, leading to other new cases. Bouchard’s appearances on talk
shows and other media had the same effect. As the project progressed,
304 B O R N TO G E T H E R — R E A R E D A PA RT

some twins were located through social services and reunion registries,
sources that would not necessarily attract visually or behaviorally similar
twins. The eighteen twin pairs who met through mistaken identity (sev-
enteen MZA and one DZA) were very similar in appearance, but there is
no connection between physical and behavioral resemblance, as has been
discussed.
The methods by which MISTRA participants (and reared-apart twins
in other studies) were identified have been criticized for attracting or fa-
voring assessment of the relatively more similar MZA twin pairs.12 That
may have been somewhat true of the study by Newman’s group because
those investigators were only interested in studying MZA twin pairs, pos-
sibly excluding dissimilar MZA sets. But their possible exclusion of dis-
similar MZA pairs is only an assumption, one I was guilty of until I read
Newman’s 1940 case report of a reared-apart female pair he identified
after publishing his 1937 study. Newman noted that he had recently
discovered a reared-apart male pair in which one twin was eager for re-
search while his twin was “somewhat of a hoodlum and refuses to sub-
mit to examination.” Newman wrote, “This is unfortunate in view of the
fact that the two brothers now seem to be so different in their personality
traits.”13
Juel-Nielsen recruited only MZA twin pairs, but did so through a
population registry; as a result, he was able to identify all possible pairs
born in Denmark between 1870 and 1910, although he included several
pairs born after 1910.14 Shields’s appeal for twins via the British Broad-
casting Corporation yielded ninety-two pairs, which he reduced to forty-
one MZA twin pairs after omitting duplications and exclusions (nine-
teen), no replies (ten), refusals (five), incompletely investigated cases (four),
and dizygosity (thirteen). Shields also found three additional MZA twin
pairs (for a total of forty-four) and two additional DZA twin pairs from
other sources15
The charge that the MISTRA sought the more similar MZA pairs fails
because Bouchard decided early on to study DZA same-sex and opposite-
sex twin pairs. It is true that the majority of the separated twin pairs (59.1
percent) were MZA, whereas the natural twinning rate predicted that 30
to 35 percent of the pairs should have been MZ.16 Of course, there is no
way of knowing if twin separations occurred equally among MZ and DZ
twins. It is conceivable (but unproven) that MZ twins are separated more
often than DZ twins because of the greater prenatal hazards to which
they are subject and their consequent poorer early health, which imposes
greater emotional and financial burdens on parents.
It is likely that the outnumbering of DZA twins by MZA twins in the
MISTRA also reflects MZA twins’ greater ease of finding one another
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 305

due to their matched appearance.17 However, to the extent that there was
an overrepresentation of MZA twins, it could be also argued that both
the less visually similar pairs and dissimilar pairs were attracted to the
study. That is probable, given that DZA twins (who did not look alike)
also volunteered for the research. In fact, twins from the second pair,
assessed two months after the Jim twins, were DZA females.
Many MZ twins, even those growing up together, are uncertain of
their twin type, partly because they and their families magnify small dif-
ferences between them. This phenomenon is well known to investigators
studying twins. However, MZ twins misclassified as DZ by their parents
or themselves receive similar behavioral ratings, consistent with their true
zygosity as opposed to their perceived zygosity.18 A number of the reared-
apart twins were uncertain of their twin type after they met, as were we.
Laboratory analyses of the twins’ blood samples took time, so some twins
finished the assessment and went home before they, or we, knew their true
twin type.
It is possible to suppose, however, that the more physically similar
MZA twins did visit the laboratory. The idea that twins who look more
alike are more behaviorally similar than those who look less alike due to
their more similar treatment by others has been raised.19 If true, psycho-
social factors rather than genetic ones would explain the resemblance
we observed between them. However, such reasoning has been weakened
by analyses showing that similar treatment by others does not cause MZ
twins to be alike.20
Science rests on data, not dialogue. Just as Juel-Nielsen compared his
registry and nonregistry twins and found no significant differences be-
tween them, the effects of various family characteristics (e.g., socio-
economic status, or rearing with a relative or nonrelative) and contact
measures (e.g., age at separation or age at reunion) on the MISTRA twins’
trait similarities were measured. Such factors contributed little to the ob-
served variance in measured traits and little to the similarity between the
twins. Some journalists, such as NBC’s science correspondent Robert Ba-
zell, complained that some MZA twins had had contact before being as-
sessed, implying that this limited the genetic interpretations of the find-
ings.21 The investigators always emphasized that the variable contact across
the twin pairs was beneficial because it allowed analysis of the extent to
which contact was associated with similarity.
In October 2010, I discovered a letter in Bouchard’s files written to
him in February 1979 by reared-apart twin Roger Brooks. Brooks and
his twin brother, Tony Milasi, had reunited fifteen years earlier after a
twenty-five year separation, and were interested in participating in the
MISTRA. Brooks had read about the study in the New York Times and
306 B O R N TO G E T H E R — R E A R E D A PA RT

after explaining his circumstances wrote, “As for the ‘contamination’ [the
fact that we have been in contact since meeting], I strongly believe that a
truly comprehensive study of the effect of the environment and heredity
can be enhanced by carefully following the subsequent lives of these twins.”
Brooks had it right.

Media Attention: Bouchard Manipulated the Press

and the Press Manipulated the Data
This reporter [Lisa Lynch] never spoke with me to check her facts! The article
was written as though I was interviewed. (Bouchard, November 20, 1996)

This handwritten note runs across the top of one of the most egregious
articles written about Bouchard and the MISTRA.22 Lynch’s only factual
words were those suggesting that the MISTRA was “one of the most
controversial dispatches in the nature-nurture debate.” The MISTRA saw
other outrageous articles, notably one by John Horgan in Scientific Amer-
ican discussed later in this chapter. These two articles were more extreme
than most, but they convey the kinds of criticisms to which the MISTRA
was subject from time to time. A sampling of Lynch’s statements (L) are
listed below, followed by my responses (R) to them.
(L) Bouchard studied intelligence with Arthur Jensen when Bouchard was
a student at Berkeley.
(R) Bouchard did not know Jensen in those years, nor did he conduct
studies of intelligence. A footnote to one of Bouchard’s 1998 papers reads,
“Jensen had been on the Berkeley faculty in Educational Psychology (lo-
cated in Tolman Hall, the same building as Psychology) while I was a stu-
dent but, as far as I can recall, I had not met him.”23

(L) Bouchard manipulated the press because newspaper coverage attracted

twins to the study. The press, in turn, manipulated the findings to imply that
Bouchard’s data had resolved the nature-nurture debate.
(R) Reared-apart twins were found in many ways, not just through the
media. Bouchard talked openly to members of the press because the findings
were exciting and he believed it was important to explain them publicly.
He was pleased that news reports attracted twins because as many pairs as
possible were desirable for the data analyses, but he never sought media
attention.

(L) MZA twins Oskar and Jack, raised Catholic in Nazi Germany and
Jewish in Trinidad, respectively, “had to be at least slightly interested in me-
dia attention or at least appearing as ‘twin-like’ as possible.”
(R) Oskar and Jack had many similarities, but each disparaged the his-
torical and political views held by their brother, a sentiment that caused
friction between them. They were hardly “twin-like” in that respect.
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 307

(L) Bouchard eventually backed away from the media because journalists
alleged, based on our findings, that “we cannot change who we are.”
(R) None of the MISTRA colleagues suggested that genes prevent behav-
ioral change or improvement, only that not everyone benefits equally from
the same experiences or interventions. Bouchard discussed the study with
any journalist or media outlet interested in talking to him.24

A 1996 article on science writing by the late sociologist Dorothy Nel-

kin in the medical journal The Lancet continued some of Lynch’s themes.
Nelkin stated that the MISTRA investigators “sought press publicity for
their studies of identical twins when their work was rejected by profes-
sional journals.”25 The source of this misinformation was Lynch’s article.
Ironically, Nelkin also stated “many accusations of inaccuracy can be
traced to reporters’ efforts to present complex material in a readable and
appealing way.”26
At the same time, many newspaper, magazine, and television journal-
ists were informed, open-minded, and fair in covering the MISTRA. They
include among others Edwin Chen (New York Times Magazine, 1979),27
Constance Holden (Science, 1980),28 Lew Cope (Minneapolis Star Tri-
bune, 1983),29 Lawrence Wright (New Yorker, 1995),30 and Jill Neimark
(Psychology Today, 1997).31 Excellent television documentaries were pro-
duced by WGBH in Boston (“Body Watch,” 1987) and WNET in Newark
(“Innovations: What Makes Us Tick? The Biology of Personality,” 1989).

Findings: The MZA Twins’ Similarities Occurred by Chance

In 1995, New York Times columnist Frank Rich detailed some curious
parallels between then President Bill Clinton and then House Speaker
Newt Gingrich, born three years apart.32 Both men had been favored
eldest children, had had strong doting mothers, had participated in stu-
dent protests, and had avoided the draft. The idea that two people near
in age and of the same sex, chosen randomly, can match across different
traits was a frequent counterargument to the possibility that some of the
MZA twins’ unusual similarities were genetically influenced. The critics’
reasoning was that eighty-one MZA twin pairs completed approximately
15,000 questions, so many findings of similarity could have occurred by
chance.
Bouchard agreed that two unrelated individuals from the same country
could display some remarkable resemblances.33 However, if chance was
chiefly responsible for the MZA twins’ behavioral similarities, then the
DZA same-sex twins should have been just as similar, but they were not.
In addition, none of the Clinton-Gingrich similarities are as rare or as strik-
ing as most of those displayed by the MZA twins. Finally, as mentioned,
308 B O R N TO G E T H E R — R E A R E D A PA RT

the effects of age and sex were removed from the data prior to analysis in
order to control for these common factors.

MISTRA Publications: Scientific Journals versus Media Coverage

Bouchard kept regularly updated lists of the MISTRA’s publications that
he forwarded to anyone upon request. He also mailed packets of his pub-
lications to all individuals requesting them. That is why a 1993 Scientific
American article by John Horgan provoked such an angry reaction from
him. And that is also why a letter disputing Horgan’s remarks was au-
thored by Matt McGue and cosigned by sixteen behavioral scientists
from eight institutions in the United States, Australia, Sweden, and the
Netherlands; the letter appeared in a later issue of the magazine.34
Bouchard prepared an unpublished memo, “Accuracy in Science
Reporting,” in response to Horgan’s piece that he circulated to his close
colleagues. He also presented his remarks to a 1994 seminar, “Science
and the Powers,” in Stockholm, Sweden.35 Horgan’s charges were based
largely on claims by Kamin. Misstatements about the MISTRA’s publica-
tion history were also made by Professor Val Dusek (D)36 as well as by
Horgan (H),37 as listed below, followed by a response (R).
(D) With the exception of the Eckert et al.’s 1986 homosexuality paper,
“the scientific data and methods of analyses upon which these conclusions
are based have not yet been published in a referred scientific journal.”38
(R) Bouchard and his colleagues published five other scientific papers be-
tween 1981 and 1987, and several book chapters in professional collections.

(H) The MISTRA twins were motivated to exaggerate their similarities

for media attention or to please the investigators.
(R) In his memo, Bouchard asserted that, “the twins would have to be
pretty talented to exaggerate their similarities in IQ, special mental abilities,
reaction time to a variety of tasks they have never seen before, responses to
thousands of questions that they have never considered, as well as cardio-
vascular functioning, cavities, etc.” In fact, many of the twins declined me-
dia opportunities in the interest of privacy.

(H) The MISTRA presented anecdotes as evidence.

(R) Bouchard countered that carefully chosen anecdotes were described
in the context of considering plausible mechanisms behind them.

(H) The MISTRA misrepresented the twins’ contact frequency and rear-
ing similarities.
(R) Bouchard had sent a packet of MISTRA publications to Horgan, so
Horgan should have been aware that selective placement did not explain the
MZA twins’ IQ similarities. Bouchard emphasized that the effects of con-
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 309

tact and upbringing were not downplayed, but rather were measured with
reference to the twins’ behavioral resemblance.

Project Funding: A Review of the Sources

Most people will be surprised to learn that the MISTRA operated more like
a mom and pop business than a major scientific investigation—sometimes
twin pair to twin pair. That was not the public perception—in 2007, psy-
chology professor William H. Tucker wrote that the MISTRA was “assisted
by millions of dollars in grant funding from numerous sources,” but that
was misleading.39 The MISTRA received a total of $2,330,720 in research
support over a twenty-five year period; some funds were awarded after the
study ended in 1999, and some collaborators used their own awards to sup-
port only specific parts of the project. The largest single grant, awarded in
1996 when the study was close to ending, was for $155,455. The average
funding per year was only $93,228; in contrast, Susan Roberts who studied
the twins’ body size and nutrition at Tufts University received $210,443 per
year for five years. The MISTRA’s different funding sources and the dollar
amounts are listed in Appendix A, Table A-1.
Funds were shockingly modest in size, though steady enough to keep
the MISTRA in business for twenty years. Fortunately, many medical costs
were absorbed by the investigators working with us. Even when the proj-
ect was close to $10,000 in debt, Bouchard kept it going, despite the fact
that the cost of assessing one twin pair with spouses was $6,935, not in-
cluding staff members’ salaries.40 Bouchard once used his own money to
purchase camera film from a Seattle company that charged half the price
of Kodak. (The film quality was awful.) Bouchard had once considered
mortgaging his home, but his wife Pauline wouldn’t stand for that. Due
to some bookkeeping issues in the University of Minnesota accounts of-
fice, he had overspent his funds by $20,000, a deficit eventually covered
by Pioneer funds.
There are three main reasons why funding for the MISTRA was diffi-
cult to obtain. The most significant reason was the timing of the study.
Even though a genetic perspective was in the air in the 1970s and 1980s,
many people viewed genetic explanations of behavior as discriminatory
and neglectful of what good parenting and interventions could accom-
plish. Some reviewers were, therefore, concerned about the potentially
explosive nature of findings that might support genetic influences on IQ
and other behaviors. The second reason was grant reviewers’ doubt that
a sufficiently large reared-apart twin sample could be identified. Federal
grant proposals require demographic details of potential subject pools,
310 B O R N TO G E T H E R — R E A R E D A PA RT

data that Bouchard could neither know nor predict because reunited twin
pairs surfaced at any time. The third reason was that Bouchard, as he was
the first to admit, did not “think big” when it came to grants, carefully
crafting his budgets to match the number of assessments he could realisti-
cally complete.
In 1981, the financial backing of the MISTRA became contentious
among academics, journalists, and the general public when Bouchard ac-
cepted a grant from the Pioneer Fund. The Pioneer Fund, based in New
York City, had a reputation for supporting research with a racial bent. I
will look at the background of the Pioneer Fund and its role in the MISTRA
later in this chapter. First, it is important to know that, as principal inves-
tigator of the MISTRA, Bouchard received awards and donations from
nine separate sources between 1979 and 2004—however, the publicity
surrounding the Pioneer Fund left the false impression that that agency
was the project’s sole benefactor.
The MISTRA was also funded by grants awarded to the colleagues in
periodontics (Pihlstrom and Michalowicz) and clinical nutrition (Roberts).
The twins’ travel, room and board, lost wages, and honoraria (approxi-
mately $5,100 per pair, with spouses, $2,600 without spouses) were cov-
ered by these grants, but the majority of funds supported the investigators’
operating costs. These grants also supported assessments of reared-together
twins with which we were not as directly involved. I have summarized these
separate funding sources in Appendix A, Table A-2. A detailed summary of
all yearly sources of MISTRA funding is presented in Table A-3.

Funding Sources
Initial Funds
There is a story behind every grant Bouchard received on behalf of the
MISTRA. The University of Minnesota Graduate School was actually re-
sponsible for financially launching the MISTRA. As a well-respected ten-
ured professor, Bouchard received several thousand dollars from the Uni-
versity of Minnesota’s graduate school in March 1979 to study the Jim
twins. There was no way to know what Bouchard would learn from the
Jim twins or where those findings would lead, but universities can risk
small sums of money for promising projects. It was a smart decision be-
cause the MISTRA ultimately enhanced the school’s reputation for first-
class research and productivity. Bouchard’s scientific curiosity was also
satisfied, something worth a lot to an investigator.
In the wake of the publicity surrounding the Jim twins, Bouchard
had a casual chat with his department chair, John (Jack) Darley Sr. Darley
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 311

told him, “You will find more twins, so you will need more grants.” Darley
telephoned the Spencer Foundation, a Chicago-based organization that
supports research on educational issues and values. The next meeting of
the awards committee was ten days away. Bouchard submitted a grant and
received $32,000. At about this time, Bouchard also secured a $10,000
grant from the publishing house Harcourt Brace Jovanovich, and $5,000
from various business sources.
“Jack Darley knew lots of people,” Bouchard told me. Bouchard could
not recall exactly how the Harcourt funding came about, but it could have
been a casual conversation with Darley in the corridor. In fact, Bouchard
kept a “Jack Darley Grant Help” file that included Darley’s many letters
written to granting agencies on Bouchard’s behalf. In it, I discovered an
April 27, 1979 letter written by Bouchard to then University of Minnesota
President Peter McGrath, in which Bouchard praised the faculty and staff
who gave so generously of their time during the Jim twins’ assessment.
Bouchard also submitted several grant proposals to the National Science
Foundation. But as Irving Gottesman pointed out, a reared-apart twin
study was “not their cup of tea.”

The National Science Foundation

The National Science Foundation (NSF) was created by Congress in 1950
“to promote the progress of science; to advance the national health, pros-
perity, and welfare; to secure the national defense . . .”41 It supports work
in many fields, including mathematics, computer science, and the social
sciences. But federal agencies tend to be quite cautious in allocating their
resources. At one time, the NSF provided immediate funds for time-limited
projects, such as a University of Minnesota study on people’s response to
the failed prediction that flooding would end the world on December 21,
1954.42 Based on this precedent, Bouchard applied to the NSF for funds
to study the Jim twins. However, the NSF no longer supported individually
proposed projects that require urgent funds. Bouchard described this
problem in a 1979 letter to Richard Atkinson, then director of the NSF.
He noted that the Minnesota graduate school funding for the Jim twins
came “on the basis of one phone call,” and that the subsequent publicity
generated additional pairs.43
The NSF did not fund Bouchard’s 1979 proposal, “A Study of a Pair of
Monozygotic Twins Reared Apart,” submitted when the Jim twins were
discovered. Excerpts from the three NSF reviews, which yielded two rat-
ings of “excellent” and one of “fair,” are reproduced below.44 My assign-
ment of reviewer numbers was arbitrary.
312 B O R N TO G E T H E R — R E A R E D A PA RT

Reviewer 1 described the proposed work as “an unusual opportunity to

obtain valuable developmental information . . . although its analysis will
not remake any general theories. (Excellent)

Reviewer 2 claimed that such “case studies carry weight that is all out of
proportion to their general methodological status.” (Excellent)

Reviewer 3 argued that it was “impossible to perform any statistical tests on

these data, and therefore to rule out explanations of the findings other than
those preferred by the investigators.” (Fair)

Federal proposals demand enormous amounts of detailed prepara-

tion, so the fact that Bouchard assembled the materials within weeks of
the March 1979 study of the Jim twins put him at a disadvantage. Re-
call that the Jim twins were found in February 1979 and studied in
March of that same year. A letter from the NSF to Bouchard dated April
5, 1979 (less than one month later) brought news that his proposal had
been declined.
A handwritten note scribbled onto one of the reviews, presumably by
the grants officer, read, “By the time I got the award document through the
process here the research had already been done. So the thing is moot (and
was declined).” As it turned out, the research was not finished, so the Jim
twins returned to Minneapolis five months later. Perhaps Bouchard should
have revisited this proposal when he invited the Jim twins back. But even
if timing had not been an issue, a favorable outcome was uncertain.
Reviewer 3 predicted a biased interpretation of the data, despite the
fact that he or she recognized the uniqueness of the research opportunity
elsewhere in the review. Furthermore, no study offers a “conclusive test of
existing theories” as he or she demanded; rather, studies provide findings
that confirm or challenge previous or existing ones, advancing knowledge
in the process. Even Reviewer 1 who was generally positive claimed, “its
analysis will not remake any general theories.” Of course, the MISTRA
and other studies ultimately did “remake” theories of behavior by showing
both genetic influence and minimal shared environmental effects across
virtually all measured traits.
Bouchard applied again for NSF funding in 1979, 1980, and 1982, but
these attempts were unsuccessful. The widely ranging reviews of his ini-
tial and revised submissions mirrored the strong disagreement over what
reared-apart twin data could contribute. His initial proposal received eight
total ratings of excellent (4), very good (1), good-fair (1), and poor (2). His
resubmission received seven total ratings of excellent (2), very good (2),
good (1), good-fair (1), and very poor (1). One negative reviewer did not
enter a rating.
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 313

The positive comments to the proposal were stunning. “It is the kind
of study one often dreams about.” “Its significance is enhanced as a result
of the semi-demise of the Burt data.” “This study is destined to become a
classic in the field regardless of the particular results.” One laudatory re-
viewer who admittedly knew “nothing about genetics,” was excited
about the MZA data, but was “not clear” on what the DZA data could
contribute. On the negative side, the small sample, the assumed contact
between co-twins, an inflated budget, and a slim chance of finding “any-
thing new” were variously mentioned by the more negative reviewers. One
feared that the work would “fan the controversy regarding heretibility
[reviewer’s incorrect spelling repeated several times] of intelligence . . . re-
jection is the only intellectually defensible course for NSF.” This last com-
ment contrasted sharply with another reviewer’s belief that “if the investi-
gators can indeed come up with twenty additional monozygotic pairs . . .
NSF should consider putting up the whole $150,000 to let them do it.”
In 1981, Bouchard received a $50,000 NSF grant for his work on the
MISTRA. However, his 1982 proposal for continued funding was re-
jected. Once again, the reviewers’ ratings displayed a wide range of opin-
ions as to the merits of the work: Excellent (3), Very Good (2), Very
Good-Good (1), and Fair (2). A sampling of their comments is insightful.

Reviewer 1: “This study eminently deserved support. It is concerned with a

most important topic and it is going about the scientific task in a most in-
formed and competent way. The results already achieved have been of great
interest and I would anticipate the study, when extended may prove to be
evidentially unsurpassable within the next fifty years or so.” (Excellent)

Reviewer 2: “Using a variety of instruments to evaluate the same traits(s) is

the wisest approach . . . I disagree with them sharply on which tests are ap-
propriate and on how to evaluate information.” (Very Good)

Reviewer 3: “This ongoing study has been reviewed frequently with mixed
reactions and has received considerable publicity. Both factors make it diffi-
cult to be critical . . . At least three of the many questions that can be raised
leave me luke-warm about this effort. First, the questions of self-selection . . .
A second and related point, is that variables in the childhood environment
that may be important are not assessed . . . A third concern is raised by the
absence of a theoretical perspective and the reluctance to specify the ques-
tions that will be asked of these data.” (Fair)

Negative feedback is not a bad thing when it is well reasoned and of-
fered in the spirit of scholarly criticism. Bouchard relished these kinds of
exchanges. But emotions and unawareness motivated some responses to
a new reared-apart twin study. One especially insulting 1980 review
314 B O R N TO G E T H E R — R E A R E D A PA RT

suggested, “There is no chance whatever that this research, which throws

the kitchen and bathroom sinks at a small number of MZAs, will [make]
any advance at all over Shields [1962]—who, by the way, also studied
adults, and more of them. I have actively resented the newspaper and
magazine propaganda efforts of these investigators.” As indicated earlier,
another reviewer complained about the small sample size, then suggested
that Bouchard first conduct a “feasibility study” to secure 150 MZA and
150 DZA twin pairs! Bouchard wrote in the margins, “Reviewer does
not know what he/she is talking about?” That reviewer also expressed
great admiration for Lykken, bemoaning the fact that Lykken’s talents
were not directed toward a richer data set. Another reviewer, who had
complained about the lack of attention to the effects of childhood envi-
ronment measures on behavior, seemed unaware of the many relevant
questionnaires and inventories the twins completed from the start.
With reviewers so sharply divided, it is easy to understand the MISTRA’s
funding difficulties. Negative reviews often overtake positive ones when
committees allocate limited funds on a competitive basis. Looking at the
rating distributions of Bouchard’s two grants confirms this view—in both
cases, there were more excellents than poors and more excellent/very
goods than fair/poors, but the negative ratings carried greater weight.
Although some reviewers’ comments seemed unjustified, the MISTRA
was not a perfect study. The twins were identified in numerous ways, forc-
ing the investigators to assess the representativeness of the reared-apart
twins by comparing their characteristics with those of reared-together
twins. The number of DZA twin pairs was small, making some findings
hard to interpret and some analyses impossible to complete. Childhood
rearing factors were obtained retrospectively and thus were subject to
possible distortions and inaccuracies. However, the analyses were con-
ducted with the various caveats in mind, as do all investigators engaged
in behavioral science research.

The Pioneer Fund

The Pioneer Fund was started in 1937 in New York City by Wycliffe Pres-
ton Draper. Draper was born in 1891 to a wealthy Massachusetts fam-
ily.45 He was a well-educated individual whose family’s fortune let him
pursue his interests in archeology, anthropology, and behavior without
responsibilities or obligations. In the 1920s, Draper became interested in
the genetic basis of intelligence and in eugenics—efforts to improve pop-
ulation quality by encouraging more intelligent individuals to reproduce
more and less intelligent individuals to reproduce less. Eugenics had
gained the attention of many psychologists and geneticists in the early
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 315

1900s, both in the United States and abroad.46 In his will, drafted in
1960, Draper left five million dollars in research funds to the Pioneer
Fund. Draper passed away in 1972. Over the years, the fund has supported
projects on the hereditary basis of intelligence, personality, and social orga-
nization, with reference to both individual and group differences.47
The Pioneer Fund has a controversial past.48 It was publicly accused of
racist policies after the Harvard Educational Review published a 1969
essay, “How Much Can We Boost IQ and Scholastic Achievement?” by
University of California–Berkeley psychologist Arthur Jensen.49 Jensen
suggested that the average fifteen-point IQ gap between black and white
populations could be associated with genetic factors. Although Jensen
had not been funded by Pioneer until 1973,50 individuals critical of his
essay discovered that the fund had supported previous research on race
differences in intelligence. The story played out many times in the me-
dia.51 Because of the Pioneer Fund’s reputation for funding race-related
research, its recipients allegedly supported racial discrimination and
anti-interventionist policies with their work. The Southern Poverty Law
Center included it in their list of hate groups.52 Many academics, jour-
nalists, and others learned about the Pioneer Fund from the press.
The Pioneer Fund contacted Bouchard in 1980 or 1981—he had never
heard of the fund prior to that time. His acceptance of their research
grants intensified critics’ charges and accusations that the MISTRA’s pro-
gram was driven by eugenics, motivated by politics, and flawed in design.
Some MISTRA colleagues and students worried that the project and their
good names would be hurt if Bouchard accepted Pioneer Fund support.
Gottesman recalled:
I brought [the Pioneer Fund’s past and present] to Bouchard’s attention. I felt
that the study was strong enough to stand on its own and that we ran the risk
of bad publicity if we accepted [their] funds. I would have turned it down and
spent more time raising funds and writing grants. But I was not in a position
to micromanage, and now I can’t second guess what might have happened
[had he declined the funds] . . . I was happy to get the benefits of the MMPI
data and become a co-author on many papers, but I have not changed my
mind about the Pioneer Fund. My name is on papers that acknowledge Pio-
neer Fund support, but that shows my confidence [in] the data.53

Gottesman also saw irony in the situation. The Pioneer Fund and the
MISTRA’s critics “got no reinforcement for any ideas about race differ-
ences in behavior because race never entered this project. We were con-
cerned with the scientific study of individual differences . . . To my knowl-
edge, no one connected to the project ever had a political agenda.” The
Pioneer Fund includes among its mission the study of group differ-
ences based on sex, social class, and race. The MISTRA’s goal was to
316 B O R N TO G E T H E R — R E A R E D A PA RT

find associations between differences in the twins’ life histories and the
twins’ behavioral differences.
Heston, who had once received a small Pioneer Fund grant, said there
was “no reason for Bouchard not to accept the funds—he could not have
done the study otherwise.” Auke Tellegen, who was not a Pioneer Fund
recipient, believed that accepting Pioneer Fund support was “politically
risky,” but that Bouchard needed the funds. McGue spoke for Lykken
when he said, “David’s argument was that we could take bad money and
do good things with it.”
Bouchard agrees with Lykken’s statement and believes people can
make their own judgments. “The left was killing us, which is why most
of our funds came from private foundations.” Bouchard was willing to
accept money from any source, but he emphasized the responsibilities of
acknowledging funding sources and describing exactly what the research
accomplished. Bouchard also recalled his 1987 taping of the ABC’s news
program 20/20. The producers spent several days filming an MZA twin
pair, then confronted Bouchard in a hostile manner about his Pioneer
funding.54 “I lost my cool—slightly. I told them that since I am being
taped for TV they could ask me any questions they wanted, but they
folded and went home. Unless they could uncover something dirty they
were uninterested.”
Former graduate student Susan Resnick, who worked on the MISTRA
when it began, was “not happy” about the study’s association with the
Pioneer Fund. While working on the MISTRA, her support came from a
National Institute of Health National Research Service Award, not from
Pioneer. Resnick is a co-author on only one publication, the first prelimi-
nary description of the MISTRA’s project design and findings, which ap-
peared in 1981 prior to Pioneer Fund support.55 She recalled, “Bouchard
was very upfront about it—he felt that it was the only way to keep the
study going.”
Minnesota psychology professor Matt McGue had also weighed the
implications of accepting Pioneer Fund support:

The Pioneer Fund never influenced the study, and the study could never
have been done without it. But there are legitimate concerns. The Pioneer
Fund had a racial agenda. The 1980s was a more politicized time with re-
spect to race, and some students worried that a Pioneer Fund association
would “taint” them. True to his nature, Bouchard heard them out. My feel-
ing then was that Bouchard was wrong to take it [the Pioneer Fund grant],
but he proved to be right . . . what [Pioneer] got out of it was minimal.

University of Western Ontario psychology professor and editor of

Personality and Individual Differences, Tony Vernon, also has received
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 317

Pioneer Fund support but was not associated with the MISTRA. Vernon
believes that racist allegations against the Pioneer Fund were “question-
able” at best, given that there has been “no solid evidence.” Vernon and
others also noted that the Pioneer Fund imposed no restriction on the inves-
tigators’ procedures or publications.56 According to MISTRA colleague
Greg Carey, “Bouchard told me that the Pioneer Fund never told him what
to write. I trust Bouchard, and I know when he says something he means
it.” Had Bouchard failed to find genetic influence on the behaviors that he
studied, the Pioneer Fund would not have interfered with his reporting.
One of the benefits of writing this book has been the opportunity to ob-
tain professional and personal perspectives from each investigator, not just
about the science but also about how the science was done. Every colleague
stands by the findings, including those who tried to dissuade Bouchard
from accepting Pioneer Fund grants. The money partly supported the sala-
ries of many project associates, secretaries, and research assistants, includ-
ing mine. As Bouchard said in 2009, “If not for Pioneer we would have
folded long ago.”
The next MISTRA grant, obtained from the David H. Koch Charitable
Foundation, was not awarded until 1985, a full three years later. That
grant was presumably made possible by the Pioneer-funded work Bouchard
had completed since 1982.
By the start of 1985, assessments had been completed on sixty-five
reared-apart twin pairs, and the study had attracted considerable atten-
tion, including that of David H. Koch, who headed the Koch Foundation.
Of course, without an offer of Pioneer Fund grants, Bouchard might have
tried harder to get money from other sources, but he did try. His friend,
businessman and developer William (Bill) I. Fine, who helped found the
Fine Theoretical Physics Institute at the University of Minnesota, intro-
duced Bouchard to many foundations in the Twin Cities, but nothing
came of these efforts.

The David H. Koch Charitable Foundation

Twins are intensely interested in twin research, making them good donors
as well as participants. David H. Koch, DZT twin and executive vice-
president of Koch Industries,57 a diversified company that operates oil
refineries, manufactures chemicals, and owns cattle ranches, was one of
them. According to Bouchard:

I had never heard of David Koch until he wrote me a letter in 1984 or 1985
asking about Cyril Burt. I wrote him a two-page reply and received a letter
back from him, asking me if I wanted to apply for a grant from his foundation.
I should have looked into his background because I would have discovered
318 B O R N TO G E T H E R — R E A R E D A PA RT

that he was a billionaire!58 I might have written a bigger grant [larger than
$10,000 to 50,000 and for more than one or two years], but I never had the
sense that a bigger grant was possible.

The Koch Charitable Foundation supports projects involving science,

education, medicine, and the arts.59 As shown in Appendix A, Table A-3,
Koch made several grants to Bouchard between 1985 and 1996.
Correspondence between grantors and grantees reveals a project’s
future funding needs while highlighting its successes. In January 1992,
Bouchard wrote to Koch inquiring about “emergency interim support”
for a reared-apart twin follow-up study of aging that was “far superior to
virtually all aging studies.” Twenty pairs had been reassessed by that time,
but the grants from the Seaver Institute (discussed in the next section) and
the study’s dental and nutrition colleagues would not let Bouchard keep
to the ten-year follow-up schedule. He requested $73,895 from Koch to
study twelve former twin pairs. The funds would have covered the twins’
travel, lodging, and research expenses only; he did not ask for assistants’
salaries. Bouchard also sent several articles to Koch, among them a pub-
lication in the Executive Editor, the journal of the National Association
of School Boards. Bouchard wrote, “We are infiltrating the educational
establishment!” meaning that educators were noticing the MISTRA find-
ings. Prior to this letter, the last Koch grant was made in 1987; the next
(and final) grant was made in 1996.
Bouchard wrote another letter to Koch in May 1995. This time, he
wanted to know if he could continue to use the remaining $11,000 from
the original grant even though the deadline had passed. Bouchard also
noted that the Pioneer Fund had “over spent” and would not be funding
research for several years. He explained that his available funds would be
used up by July, and he hoped he could submit another proposal to Koch.
Bouchard provided a long list of his accomplishments (e.g., presentation
at the NATO Conference in the Netherlands; Distinguished Scientist
Address at the American Psychological Association) adding, “We feel
that our work, taken in conjunction with other behavior geneticists, has
begun to completely transform the way in which human behavior is un-
derstood.” It was a change from Bouchard’s usual modesty. And he was
able to use the remaining $11,000.

The Seaver Institute

The Seaver Institute, founded in 1955 by Frank R. Seaver, is based in
Los Angeles, California. Seaver had made his fortune as head of Hydril,
a Houston–based company that designs and manufactures products for
petroleum drilling and production. The Seaver Institute supports research
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 319

in science, medicine, education, public affairs, and the arts. It targets proj-
ects that are likely to help humanity, and it is willing to invest in projects
unlikely to receive funding from other sources. Then President of the
Seaver Institute, Dr. Richard Call, consulted with university officials to
learn about potentially fundable projects of interest.60 In the late 1980s,
Call contacted Bouchard and met with him in Minneapolis to discuss grant
support for the MISTRA. Five grants, totaling $250,000, were awarded
in support of the MISTRA between 1988 and 1993.

The Whitfield Institute

The Whitfield Institute was a British charity established in 1999 by an
anonymous donor.61 The institute raised money with which it supported
research. Bouchard was unaware of the Whitfield’s existence until a col-
league referred him to an associate of the institute who suggested that
Bouchard submit a proposal. The proposal was approved in 2004, pro-
viding the final grant ($65,884) received by the MISTRA. According to
Bouchard, the Whitfield grant enabled a great deal of data processing and
analysis to be completed after the study ended. The Whitfield Institute
closed in 2005 because the individual responsible for funding discontin-
ued doing so.

Bouchard did not have a fortune to work with, but he made the most of
the grants that came his way.

Twins Reared Apart Research: Looking Back and Looking Ahead

Because of its approach, its methods, its findings, and its timing, the
MISTRA has substantially affected the way people think about the ori-
gins of behavioral and medical traits. Many scientists acknowledge the
effects of genes on behavior.62 The twins’ reunion stories touch us deeply,
an observation that constitutes vital information about the universal im-
portance of family relationships in human social development.
The MISTRA’s biggest contribution to psychology was showing that
behaviors found to be familial, such as intelligence, personality, and reli-
giosity, have substantial genetic components. The study also showed that
shared environments have little effect on the behavioral resemblance of
relatives living together. The limited effects of shared environments on
behavior seem counterintuitive, but most people grasp this concept when
they compare their own behaviors with those of their family members.
There have been misunderstandings of the findings. There have also
been misunderstandings of behavioral genetics concepts in general. The
biggest source of confusion comes from the meaning of the term variance.
320 B O R N TO G E T H E R — R E A R E D A PA RT

Heritability, the variance in a trait due to genetic differences among

people, applies to populations, not to individuals. Heritability estimates
of 0.50 to 0.60 for alcoholism do not mean that every child born to alco-
holic parents or every MZ twin with an alcoholic co-twin will become an
alcoholic.63 Having immediate alcoholic relatives increases one’s risk of
developing that disorder, but genes do not imply destiny. Gene-behavior
relationships are probabilistic and indirect—the risk of becoming alco-
holic in individual at-risk cases may be large or small. However, the value
of knowing the relative risk of alcoholism lets individuals take preventa-
tive measures, such as avoiding or minimizing alcohol intake.
It is also important to emphasize that most behavioral genetic findings
apply to individuals living within the normal range of human environ-
ments. The MISTRA did not study individuals from impoverished or abu-
sive settings in which environmental effects can be quite salient. For ex-
ample, a study of two-year-old twins found that children raised in deprived
neighborhoods were at increased risk for behavioral problems, beyond
any genetic liability.64
It is interesting to speculate on the impact the Jim twins might have
had on psychological research had they met thirty years earlier than they
did (in 1950) or thirty years later (in 2010). In the 1950s and 1960s, the
Jim twins might have excited interest among academics questioning envi-
ronmentalist views, but relatively few people were open to a genetic per-
spective at that time. When the Jim twins were discovered in 1979, the
psychological climate had changed just enough so that many people were
interested in what these twins might reveal about human behavior.
Had the Jim twins met in 2010, it is quite likely that they would have
been of scientific interest, but they would probably not have attracted the
worldwide attention that they did in 1979. By 2010, the heritability of
most human behavioral traits had been established by countless twin, sib-
ling, and adoption studies. (For this same reason, Cyril Burt’s study might
have stirred far less controversy had Kamin exposed Burt’s suspicious
correlations later than he did.) The irony is that the MISTRA inspired
many of these studies. Regardless, the Jim twins would still have been
valued research subjects in 2010 and beyond because of the increased
focus on how environments affect behavior and the many advances made
in the molecular genetics field.

Genes, Environments, and Heritability

Since about 2000, behavioral-genetic researchers have been paying greater
attention to family and other environmental influences on behavior, and
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 321

to how they work in concert with genetic potentials. University of Edin-

burgh research fellow Wendy Johnson believes that our understanding of
gene-behavior relationships can be enhanced through twin studies of
gene-environment transactions—that is, changes in genetic and environ-
mental variation with changes in environmental effects, such as educa-
tional, occupational, and nutritional factors. Johnson’s view makes sense.
This shift away from heritability estimation to more focused environ-
mental analysis is understandable because we now know that most mea-
sured traits have a genetic component. However, the early reared-apart
twin investigators also cared enormously about how environmental fac-
tors moderated gene expression. Newman’s final 1940 reared-apart twin
case included wonderfully speculative passages, material that rarely has a
place in today’s scientific journals given the emphases on statistical find-
ings and participant confidentiality. He and his co-author Iva Gardner
wondered if there are “thresholds of influence, and that, unless a certain
threshold of difference in environment is reached, the organisms do not
respond differently.”65 Their reasoning was based on the striking behav-
ioral similarities he observed in his final pair, eighteen-year-old MZA twins
Lois and Louise, despite their residential and educational differences.
In this 1940 case study, Newman also claimed that a common criticism
of his 1937 study was that it emphasized environmental influence and
minimized genetic influence. This comment is surprising given the prevail-
ing environmentalist perspective of that time. He attributed this criticism
to the educational specialties of his two co-authors. “Hence in our work
we may have gone too far in the opposite direction.” Throughout the
course of the study, the biologist Newman was “much more impressed
with the very great intrapair similarities of these twins, after they had
been exposed to all sorts of environmental differences.”66
Some investigators, Johnson included, have asserted that there is no
longer a place for twin study estimates of heritability because of the
increased focus on gene-environment relationships.67 Clearly, the behav-
ioral genetics field, which also includes studies of siblings, half-siblings,
adoptees, and twin-family constellations has edged away from estimating
trait heritabilities to assessing genetically based behavioral changes, ge-
netic associations between different behaviors, and the genetic bases of
environmental measures. The genetics of experience—that is, how people
perceive and remember life events—promises to be a fascinating and in-
formative area of investigation.68 However, there are things we can still
learn about the genetic basis of some traits.
Only one twin study has examined the genetic basis of romantic love
styles, the tendency to fall in love quickly or slowly over time.69 A lack of
322 B O R N TO G E T H E R — R E A R E D A PA RT

genetic influence and a shared family effect were reported. The only twin
study of the heritability of materialism (the importance of owning and
acquiring material possessions), conducted in 2009, also showed a lack
of genetic effects, with the exception of a happiness dimension (content-
ment with one’s current possessions).70 Additional studies of these behav-
iors, using a variety of methods administered on several occasions, would
be important for purposes of information gathering and replication. I
wish Bouchard had administered love style and materialism question-
naires to the MISTRA twins, but it is impossible to assess every behavior
in the same study.
Reared-apart twin studies have included mostly adults, but in 2006 I
began the first prospective study of reared-apart twin children from China,
as mentioned in Chapter 2. Genetic effects on intelligence, personality, and
interests have never been studied using young twins adopted apart or
together, but they can be estimated using these Chinese twin samples. Our
understanding of traits such as fluctuating asymmetry, traditionalism, job
satisfaction, and age at menarche would also benefit from studies tracking
these separated twins as well as other unusual kinships, such as virtual
twins (same-age unrelated siblings) and superfecundated twins (twins hav-
ing the same mother, but different fathers).71 Twin studies of behaviors that
have been assessed rarely or not at all, such as adult romantic attachment,
sleep quality, and disregard for rules, have appeared in 2011 issues of the
journal Behavior Genetics,72 but more would be welcome.

New Directions
Molecular genetics explores the effects of genes at the DNA level. Its
techniques include linkage analysis and association analysis, among
others. Linkage is the tendency for some genes to be inherited together
because of their proximity on the same chromosome. Linkage analysis is
used to map genes to specific chromosomes and to look for relationships
between genotypes and phenotypes among related individuals. Associa-
tion analysis involves direct tests of correlations between genotypes and
phenotypes and can use populations. Genes or chromosomal regions
thought to be relevant to a particular trait (“candidate genes”) can be ex-
amined.73 For example, a 2007 study looked at associations between
cognition and genes associated with oxidative stress, thought to affect
aging of the brain.74
Since the completion of the Human Genome Project in 2003 and the
International HapMap project (a description of the common patterns of
DNA sequence variation) in 2005, it has been possible to quickly scan the
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 323

entire genomes of hundreds of people to search for genotype-phenotype

associations. These efforts are called genomewide association studies
(GWAS).75
Molecular genetic studies of medical disorders show that there are Men-
delian (single gene) forms of complex conditions. Examples of such condi-
tions include breast cancer, hypercholesterolemia, and Alzheimer disease. If
an individual inherits the relevant detrimental gene for one of these disor-
ders from his or her parents, the probability of developing the disorder
approaches 100 percent. However, the Mendelian forms are quite rare,
representing only about 2 percent of the cases. The overwhelming genetic
contribution to these conditions comes from the polygenic (multiple
gene) system in which each gene contributes to a probabilistic, rather
than a deterministic, association with the disorder. This is especially true
of psychopathology.
Molecular genetic advances have largely changed the complexion of
human behavioral-genetic research and reordered its agenda.76 Journal
article titles in the 1970s focused on behavior, but many in 2011 focus on
specific genes. However, some investigators believe that conventional twin
studies are still making an impact in their areas of expertise. Susan
Roberts, who headed the MISTRA’s clinical nutrition study, said she is
surprised when people “yawn” at heritability research because she be-
lieves that the reared-apart and reared-together twins she studied strength-
ened her work.
In an overview of behavioral and psychiatric twin studies delivered
at the thirteenth International Twin Congress in Seoul, Professor Dorret
Boomsma defined a significant role for classic MZ-DZ twin comparisons,
even in the era of molecular genetic research.77 She claimed that twins,
especially behavior- and disease-discordant MZ twins, help identify genes
underlying complex quantitative traits. Twins do this by controlling the
genetic and environmental confounds inherent in biological parent-child
relationships. It is also true that, despite years of study, little is known
about why the heritabilities of height and other human traits have the
values that they do. New techniques looking at genomewide gene expres-
sion promise to bring us closer to the answers.78
New areas of study have been defined by molecular genetic advances
that have occurred since 2001 with the mapping of the human genome.
The GenomeEUTwin Project is dedicated to finding genes linked to com-
plex diseases by studying the DNA of hundreds of thousands of twin
pairs.79 This collaborative project includes researchers from Finland, Swe-
den, Norway, Denmark, the Netherlands, Italy, the United Kingdom, and
Australia with ties to projects in Canada and Estonia.
324 B O R N TO G E T H E R — R E A R E D A PA RT

Expectations for what genomewide association studies can accomplish

should be reasonable. For example, single-nucleotide polymorphisms
(SNPs, DNA sequence variations involving a change in a single nucleo-
tide) associated with height explain less than 5 percent of the variation.80
However, the heritability of height is close to 90 percent! About twenty
common gene variants have been found for type 2 diabetes, and about
seven have been found for schizophrenia, yet they explain very small per-
centages of the variance in these heritable conditions. The “missing herita-
bility” may come from rare gene variants, structural gene variants, gene-
gene interactions, gene-environment interactions, unknown gene variants,
or combinations of these.81 The genetic variation underlying general intel-
ligence has also been linked to many genes with small effects.82
Interestingly, the search for molecular genetic correlates of disease and
other traits parallels the search for nonshared environmental correlates
of behavior. The environmental causes of complex human traits are “in-
numerable and essentially non-additive,” and many small associations in
both genomics research and in the social sciences cannot be replicated.83
A great deal of work remains to be done with respect to linking specific
genes to medical conditions, as well as to physical and behavioral traits.
Twin research can further activity in epigenetics, the study of changes in
gene expression associated with factors unrelated to changes in DNA se-
quencing. In 2005, Mario F. Fraga, from Madrid’s National Cancer Center,
showed that the epigenetic profiles (patterns of chemical modification of
the genes that affect gene expression) were more alike in younger than
older MZT twin pairs.84 Epigenetic patterns were also more alike in MZT
twins who spent more rather than less time together. Gottesman has also
conducted epigenetic twin studies in the area of psychopathology.85 He has
reminded everyone that the MISTRA showed that resemblance for some
behavioral traits does not differ between MZA and MZT twins, so it will
be important to reconcile this finding with those of Fraga.
Subsequent research has shown that specific environmental factors can
affect the epigenome, thereby altering genetic expression, and that some
such effects are heritable. For example, reduced male fertility associated
with toxin exposure can be transmitted to later generations.86
In 2008, researchers identified copy number variations, differences in
DNA segments, which may be linked to disease differences in MZT twins.87
An international research team compared these characteristics in nine-
teen MZ twin pairs, ten of whom were free of Parkinson disease symp-
toms and nine who were discordant for the disease. Copy number varia-
tions were detected in both twin groups, but it is likely that they can offer
insight into disease onset and progression. Studying copy number varia-
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 325

tions in disease-discordant MZA twins might tell us how environments

affect the expression of specific conditions. Had this technology been
available earlier, the MISTRA’s MZA pairs who differed in narcolepsy,
headaches, and depressive symptoms might have made contributions
along these lines.
In the coming years, we are likely to see new behavioral genetic analy-
ses of a widening array of behaviors, including dementia, obesity, autism,
learning disabilities, schizophrenia, and bipolar illness. There will be in-
creased efforts at finding endophenotypes, the internal phenotypes that
bridge the gap between genes and complex behaviors and can help pin-
point their etiologies.88 Endophenotypes, which could vary in form (e.g.,
neurophysiological, neuroanatomical, or biochemical), may offer simpler
clues to the genetic bases of disease risk.89 Gottesman and Shields popu-
larized the term in 1973, and Gottesman has since asserted its importance
in neuropsychiatric research.90 Both quantitative and molecular ap-
proaches will be utilized in future behavioral genetic research—but
twins’ behaviors in natural and experimental settings will always excite
and inspire.

Final Comment
I share stories of newly reunited twins with Bouchard, who told me, “An
accumulation of such cases cannot do anything but help us make a per-
suasive case that something interesting is going on.” He also admitted,
“When I hear about a reunited pair it tugs at my heart strings.” These are
the pairs he won’t be studying.
As a DZT twin, I knew from an early age that my twin sister Anne and
I were very different in our behaviors and appearance, and I decided that
there was “something” fundamentally different about us. We were hardly
as alike as the MZ twins I knew in elementary school. When I learned in
science classes about the effects of genes on behavior and the differential
effects of environments on family members, our developmental differences
made sense. Perhaps that is why the reared-apart twin findings seemed so
reasonable, affirming my beliefs about the way people become who they
are.
It may be surprising to learn that twins are still being separated, even
though most adoption agencies appreciate the importance of keeping them
together. During the last thirty years, advances in assisted reproductive
technology have provided new sources of reared-apart twins. Frequently
used techniques variously involve ovarian stimulation via fertility drugs,
and the conception and implantation of multiple embryos into the mother’s
326 B O R N TO G E T H E R — R E A R E D A PA RT

womb. Because some families are unprepared to manage or support

multiple-birth children, parents sometimes arrange for twins’ separate
rearing. Parental divorce, foreign adoption, and accidental switching also
keep some twins apart.91 The good news is that better methods for re-
uniting twins, especially Internet resources, are available. And physicians
working in the area of reproductive medicine are becoming better able to
achieve successful pregnancies using single embryos rather than several.92
Had the Jim twins been discovered in 2001 or later, it is likely that
studying their epigenetic profiles, copy number variations, and various
molecular genetic characteristics would have been higher priorities than
studying their shared occupations, heart conditions, and hobbies, the ge-
netic components of which have been widely demonstrated. This is why
the Jim twins would have had a different impact on thinking about be-
havior today than they had thirty years ago in 1979 when they found
each other. Trying to tie the Jim twins’ matched headaches and hobbies
to specific genetic factors would not have produced the same high level
of general fascination and debate as did the descriptions of their matched
behaviors. However, Bouchard believes that reared-apart twins show us
genetic influence on behavior in ways that no other source can, and that
they will continue to do so. “It is so up front and personal with twins
reared apart,” he said. “You can see the effects right between the eyes.”
The MISTRA’s individual findings leave a big picture, one that it shares
with past and present studies of twins and adoptees. It has “rearranged the
furniture in psychology’s house.”93 The so-called furniture or topics were
not changed by behavior genetic studies, but the concepts, approaches,
and explanations have been. Finding genetic influence on intelligence,
personality, and many other human characteristics allows more realistic
appraisal of similarities and differences among family members. The mod-
est effects of common family influences on most behaviors—the events or
experiences that make family members alike—offer greater understand-
ing of parent-child and sibling resemblance and lack of resemblance.
They also encourage efforts toward finding the nonshared environmental
influences that significantly impact behavior and make family members
differ.
About half the variation in most behaviors is explained by genetic fac-
tors, although some traits such as brain waves show greater genetic influ-
ence, and other traits such as job satisfaction show less. Environmental
events, linked to individuals’ life circumstances, can affect the expression
of heritable traits such as intelligence and height. IQ gains have been re-
ported for abused children adopted into high socioeconomic homes.94
Poor nutrition may restrict the physical growth of a child or adolescent.95
 QUESTIONS, ANSWERS, AND TWIN STUDIES OF THE FUTURE 327

Most important, the MISTRA has helped define the domains in which
answers to questions about how behaviorally relevant genetic processes
work are likely. Advances in molecular genetics and related fields are
furnishing tools that can facilitate efforts in that direction.
We are moving toward a more informed view of what twin studies can
and cannot accomplish. “Twin studies . . . refute both biological and en-
vironmental determinism. They do not negate the effect of the environ-
ment on behavior, nor do they overglorify the role of genes. They account
for the uniqueness of each of us.”96
 APPENDIX A

Funding Sources

Detailed summaries of research funds awarded to the MISTRA on a

yearly basis are included in Tables A-1, A-2, and A-3. The information in
these tables was reproduced from Bouchard’s grant records provided to
him by the University of Minnesota’s accounting office. This much finan-
cial detail is rarely provided or requested for any study, but it is presented
here to satisfy the long-standing interests, questions, and criticisms asso-
ciated with the MISTRA’s funding sources. The largest single grant made
directly to the MISTRA was for $115,455.

Table A-1. Funding for Bouchard as the MISTRA’s Principal Investigator, Listed in Order of Amount

Award Amount
Funding Source (US$)

University of Minnesota Graduate Schoola 6,500

National Science Foundation 50,000
Business and industry (other)b 65,000
Whitfield Institute 65,884
Private donors 106,485
Spencer Foundation 181,300
Koch Charitable Foundation 185,000
Seaver Institute 250,000
Pioneer Fund 1,420,551
Total 2,330,720

a. The University of Minnesota Graduate School provided $3,800 (not listed here) to
cover the Jim twins when they were discovered in March 1979.
b. Includes Harcourt Brace Jovanovich ($10,000).
 Table A-2. Other Funding Sources for the MISTRA

Award Amount
Funding Source (US$)

National Institute of Health/National 216,123

Institute of Dental Research (Pihlstrom)
National Institute of Health/National 308,730
Institute of Dental Research (Pihlstrom)
Health and Human Services (Roberts) 1,052,213

Table A-3. Yearly Funding Sources for the MISTRA (Principal Investigator: Bouchard)

Cumulative Cumulative
Source End Date Award (US$) Year Year Total (US$)

University of 1979 year 6,500.00

Minnesota
Spencer Foundation 6/30/1980 32,000.00 1980 1979 38,500
Business and 12/31/1980 15,000.00 1980 1980 15,000
industry; other
NSF Division of 6/30/1981 50,000.00 1981
Integrative Biology
and Neurosciences
(INS)
Pioneer Fund 8/26/1981 25,000.00 1981
Spencer Foundation 6/30/1981 64,300.00 1981 1981 75,000
Pioneer Fund 7/14/1982 89,415.00 1982
Spencer Foundation 11/30/1982 85,000.00 1982 1982 174,415
Pioneer Fund 8/31/1983 68,274.00 1983 1983 68,274
Pioneer Fund 4/30/1984 57,895.00 1984 1984 57,895
Koch Industries 2/13/1985 10,000.00 1985
Koch Industries 2/13/1985 50,000.00 1985
Pioneer Fund 7/31/1985 95,000.00 1985 1985 155,000
Pioneer Fund 7/31/1986 100,000.00 1986 1986 100,000
Koch Industries 6/30/1987 50,000.00 1987
Pioneer Fund 7/31/1987 112,000.00 1987
Pioneer Fund 7/31/1987 20,000.00 1987 1987 182,000
Pioneer Fund 6/30/1988 100,000.00 1988
Seaver Institute 6/30/1988 50,000.00 1988
Pioneer Fund 12/31/1988 10,000.00 1988 1988 160,000
Pioneer Fund 6/30/1989 100,000.00 1989
Seaver Institute 6/30/1989 50,000.00 1989 1989 150,000
Pioneer Fund 6/30/1990 105,000.00 1990 1990 105,000
Seaver Institute 2/28/1991 50,000.00 1991
Pioneer Fund 6/30/1991 100,000.00 1991 1991 150,000
Industrial donors 6/25/1992 50,000.00 1992
Pioneer Fund 6/30/1992 105,000.00 1992
Table A-3. (Continued)

Cumulative Cumulative
Source End Date Award (US$) Year Year Total (US$)

Seaver Institute 6/30/1992 50,000.00 1992 1992 205,000

Pioneer Fund 6/30/1993 105,000.00 1993
Seaver Institute 9/30/1993 50,000.00 1993 1993 155,000
Pioneer Fund 6/30/1994 112,512.00 1994 1994 112,512
Koch Industries 6/30/1996 75,000.00 1996 1995 0
Pioneer Fund 6/30/1997 115,455.00 1997 1996 75,000
Private donors 12/31/1999 2,985.00 1999 1997 115,455
Private donors 12/31/1999 75,000.00 1999 1998 0
Private donors 12/31/1999 25,000.00 1999 1999 106,485
Private donors 12/31/1999 3,500.00 1999
Whitfield Institute 3/31/2004 65,884.00 2004 2004 65,884
Total 2,330,720.00
MISTRA
Other funding
NIH/NIDR 8/1/1987– 216,123.00
7/31/90
NIH/NIDR 4/1/1993– 308,730.00
3/31/1997
HHS 8/1993– 1,052,213.00
7/1998

Note: Bouchard was also co-principal investigator (with Lykken) of several twin-related projects
while the MISTRA was ongoing. Two of these grants supported development of the Minnesota Twin
Registry (National Institute of Mental Health: $379,522; and Pioneer Fund: $20,000), and one grant
supported a study of risk factors for criminality (Pioneer Fund: $99,621).
 APPENDIX B

Glossary

Active genotype-environment correlation. The seeking out of environments and

experiences compatible with one’s genetic proclivities. Studious people are
likely to be found at libraries and lectures, whereas artistic people might fre-
quent museums and galleries.
Additive genetic variance. The genetic variance resulting from the independent
effects of different genes adding up. Additive genetic variance is responsible
for most of the genetic similarity between relatives in traits such as height.
Age-sex correction. Statistical procedures for removing the effects of age and sex
or for various nonlinear or interactive effects of these variables on a trait. MZ
twins and approximately two-thirds of DZ co-twins are the same age and sex,
factors that can inflate trait similarity between them, just as it would between
two unrelated same-age males or females. DZ male-female co-twins differ in
sex, but being male or female could be associated with the level of a trait such
as spatial skill, so age- and sex-correction would be appropriate for these twins
as well. Data can be adjusted for the effects of age alone or sex alone.
Alleles. Different forms of a gene, such as the eye color gene (or allele) coding for
brown, the gene (or allele) coding for green/blue, and the gene (or allele) coding
for brown/blue.
Assortative mating. The tendency for individuals in couples to match on given
traits. If these traits, such as intelligence or interests, have a genetic component
then trait similarity between relatives such as DZ twins and full siblings can
be enhanced, reducing heritability. Assortative mating does not affect MZ
twin resemblance because MZ twins share all their genes. Assortative mating
appears to be highest for social attitudes and values.
Attenuation correction. The upward adjustment of the correlation coefficient
due to measurement error.
Biometrical modeling. Includes the class of statistical procedures that enable
the simultaneous estimation of genetic, shared environmental, and nonshared
334 APPENDIX B

environmental factors affecting variation in the traits of interest. These tech-

niques can incorporate different kinships such as twins, their parents, and their
children into the same model. Biometrical methods test how well a model fits a
given set of data. For example, investigators compare the fit of several models
and can decide if a model including genetic and nonshared environmental
components fits the data better after the addition of a shared environmental
component.
Correlation coefficient (r). A statistical measure of association between two things,
such as height and weight, that varies between −1.0 and 1.0. This correlation
is squared to obtain the proportion of variance in one variable (such as
height) that is predictable from the variance in the other variable (such as
weight).
Direct heritability estimates. Derived from the intraclass correlation of MZA
twins or other relatives. The MZA intraclass correlation of 0.50 for a person-
ality trait directly estimates the heritability of the trait because MZA twins
share only their genes. Direct heritability estimates are higher than indirect
heritability estimates, although the reason for this difference in unclear. The
intraclass correlations for unrelated siblings or unrelated parent-child pairs
directly estimate environmental effects because these individuals share only
their environments.
Dominance. The interaction between two genes, or alleles, at corresponding lo-
cations on paired chromosomes. A child who inherited a dominant gene for
type A blood from one parent and a recessive gene for type O blood from the
other parent would have type A blood. A child would have type O blood only
when he or she inherited two copies of the gene for type O, in which case there
is no dominance effect.
Emergenic trait. A unique behavior or attribute resulting from interactions among
complex gene configurations. Emergenesis explains why some genetically influ-
enced traits, such as unusual musical talent or running speed, may not run in
families. Emergenic traits can be identified by a very high MZ intraclass cor-
relation and a very low DZ intraclass correlation. This is because MZ twins
share all their genes and gene combinations, whereas DZ twins share half their
genes, on average, and rarely inherit the same complex gene combinations.
Epigenetics. The study of chemical reactions that “turn on” or “turn off” certain
genes at different points in time. For example, prenatal nutrition, stress, and
toxins may affect gene expression that is controlled by the epigenome. Envi-
ronmental toxins can lower male fertility, an effect that can be transmitted to
subsequent generations.
Epigenome. A subset of genes whose functions are a consequence not only of
their DNA sequence but also of the biochemical factors affecting their
expression.
Epistasis. Interactions among genes at different chromosomal locations. Epista-
sis is similar to dominance but involves associations among genes or alleles at
more than one locus.
 APPENDIX B 335

Gene. A hereditary unit that occupies a specific place or locus on a specific chro-
mosome. A gene is a segment of DNA (deoxyribonucleic acid) that codes for a
certain chemical product that ultimately contributes to human characteristics.
Genotype-environment correlation. The co-occurrence of particular genetically
based traits and environments. A person with an inquisitive mind would be
likely to be found in libraries and museums.
Genotype × environment interaction. The different effects of the environment
depending upon the genotypes of the individuals. A very bright child would
probably enjoy an extracurricular opportunity to study math, whereas his less
bright sibling might find the class uninteresting or even stressful.
Genetic correlation. The extent to which two traits, such as disinhibition and
experience seeking, are influenced by common genes. Genetic correlation be-
tween two traits is indicated if the MZ cross-twin cross-trait correlation ex-
ceeds the DZ cross-twin cross-trait correlation. For example, if the correlation
between the height of MZ twin 1 and the weight of MZ twin 2 is greater than
the correlation between the height of DZ twin 1 and the weight of DZ twin 2,
this suggests that height and weight are affected by some of the same genes.
Genotype. The genetic makeup of an organism; also, the set of genes relevant to
a given trait.
Heritability. The proportion of trait variance in a population that is due to all
sources of genetic differences among the members of that population, also
called broad heritability. If the heritability of intelligence is estimated to be 0.70
then 70 percent of the variance, or individual differences in the population, is
explained by genetic differences among the members of the population and 30
percent is explained by environmental differences among them plus measure-
ment error. Heritability does not apply to individuals—it would be incorrect
to say that 70 percent of an individual’s intelligence is associated with genetic
factors, because genes and environments are not separable at the individual
level, only at the population level.
Indirect heritability estimate. Heritability based on the difference between two
correlations. For example, if the MZT IQ correlation is 0.86, and the DZ cor-
relation is 0.60, the difference between them is 0.26; the 0.26 value is then
multiplied by 2 to obtain the heritability estimate of 0.52. This is because MZT
twins share all their genes plus their environment, while DZT twins share half
their genes plus their environment. The difference between the two correlations
yields half the genetic effect; the difference is multiplied by 2 to obtain the full
genetic effect.
Intraclass correlation coefficient (ri). A statistical measure of association or re-
semblance between twins and other classes of relatives. The intraclass correla-
tion normally ranges from 0.0 to 1.0, indicating no resemblance or perfect
resemblance between relatives on a trait of interest. Negative intraclass cor-
relations are possible, showing less systematic resemblance between relatives
than individuals paired randomly; this could occur in the event of contrast
effects for twins. The intraclass correlation, derived from analysis of variance,
336 APPENDIX B

expresses the ratio of the between-group variation in a trait to the total variation
in a trait. Between-pair variance refers to differences among the pairs in a
sample, the within-pair variance refers to differences within each twin pair,
and the total variance refers to their sum. If the trait under study is affected by
genetic factors, the between-pair differences would be higher for MZ than DZ
twins, and the within-pair differences would be relatively smaller; however,
the total trait variance for the two types of twins is assumed to be the same.
Genetic effects are shown if the correlation for MZ or MZA twins exceeds the
correlation for DZ or DZA twins. This correlation is not squared because it
already expresses the proportion of shared variance between relatives.
Narrow heritability. The proportion of trait variance in a population that is due
to additive genetic differences among the members of that population. Like
broad heritability, narrow heritability only applies to populations, not to indi-
viduals. Narrow heritability is responsible for most of the genetic resemblance
between relatives other than MZ twins.
Nonadditive genetic variance. Genetic variance resulting from the interactive ef-
fects of genes; see Dominance and Epistasis.
Nonshared environment. Unique events that family members experience apart
from one another that make them differ from one another in a given trait. If
one child in a family took a course in contemporary politics or followed a
vegetarian diet, this child could differ from his siblings in knowledge of world
events and in body weight.
Pairwise concordance. A measure of twin resemblance for discrete (“either-or”)
traits, equal to the number of concordant pairs, divided by the number of con-
cordant pairs plus discordant pairs. If fifteen twin pairs were concordant for
schizophrenia and ten pairs were discordant, the pairwise concordance rate
would be 15/25 or 60 percent.
Passive genotype-environment correlation. The association of genetically based
behaviors and environments that arises because parents transmit both genes
and environments to their children. Bright children are likely to have inherited
their abilities from their parents, who also provide them with intellectually
stimulating experiences.
Phenotype. Observable or measurable characteristics that are a function of one’s
genes and environments. For example, a person’s height can be observed and
measured, and is an outcome of the genes inherited from parents in conjunc-
tion with diet and health care.
Probandwise concordance. A measure of twin resemblance for discrete (“either-
or”) traits, equal to the number of affected individuals in concordant pairs
divided by the total number of affected individuals. In fact, there are two types
of concordant pairs, those in which pair members are independently identified
(doubly ascertained) and those in which they were identified because they
were twins. Doubly ascertained twin pairs are counted twice. If twenty af-
fected twin individuals in ten twin pairs were found independently, five affected
twins were found in another five pairs because their co-twins were affected, and
 APPENDIX B 337

ten discordant twin pairs were found, then the probandwise concordance rate
would be (20 + 5)/35 or slightly higher than 71 percent.
Probe. A single-stranded molecule of DNA used to identify a complementary
sequence of another DNA molecule.
Reactive (evocative) genotype-environment correlation. People’s responses to
individuals based on the expression of these individuals’ genetically based
behaviors. Athletically minded children are likely to be taken to ball games
and related activities by their parents.
Shared environment. Events family members experience in common that make
them alike in a given trait. If two siblings took piano lessons in their home or
athletic training at school, they would show more similar musical ability and
sports skills than if only one sibling had had these experiences.
Standard deviation (SD). A measure of the distribution of scores around a mean,
equal to the square root of the variance. For example, the IQ test score popu-
lation distribution follows a symmetrical bell-shaped curve in which the mean is
100 and the standard deviation is 15. Ninety-five percent of the population
scores between 70 and 130 in IQ, or 2 standard deviations above and below the
mean.
Standard error. The standard deviation of the sampling distribution. If the mean
of a population is measured several times, it will differ, and the standard error
gives an index of the spread.
Statistical control. Statistically removing the effects of one variable on another.
For example, age or education might affect a person’s answers to questions in
an interest inventory. If a researcher wanted to study sex differences in per-
sonal interest categories, it would be wise to first statistically remove the effects
of age and education on the interest scores.
Stem and leaf plot. A method for displaying the frequency with which some
classes of variables occur.
Variance. The distribution or spread of scores around the mean of a sample or
population.
Notes

Introduction
Epigraph: Walter Mischel, Introduction to Personality (New York: Holt,
Rinehart & Winston, 1981), 311.
1. See, for example, Lois W. Hoffman, “The Changing Genetics/Socialization
Balance,” Journal of Social Issues 41 (1985): 127–148.
2. Laura E. Berk, Child Development, 8th ed. (Boston: Allyn & Bacon, 2008),
119. Berk emphasizes the limitations of heritability estimates of behavior, but
she included findings from recent behavioral genetic studies in her textbook.
3. Francis Galton, “The History of Twins, as a Criterion of the Relative Powers
of Nature and Nurture,” Fraser’s Magazine 12 (1875): 566–576, p. 566.
4. Richard D. Rende, Robert Plomin, and Steven G. Vandenberg, “Who Discov-
ered the Twin Method?” Behavior Genetics 20 (1990): 277–285, p. 277.
5. Thomas J. Bouchard Jr. and Peter Propping, “Twins: Nature’s Twice-Told
Tale,” in Twins as a Tool of Behavioral Genetics, ed. Thomas J. Bouchard Jr.
and Peter Propping, 1–15 (Chichester, U.K.: John Wiley & Sons, 1993).
6. Ibid.
7. Leslie B. Arey, “Direct Proof of the Monozygotic Origin of Human Identical
Twins,” Anatomical Record 23 (1922): 245–251; Leslie B. Arey, “Chorionic
Fusion and Augmented Twinning in the Human Tube,” Anatomical Record
23 (1922): 253–261; also see Amram Scheinfeld, Twins and Supertwins (Bal-
timore: Penguin Books, 1967).
8. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A
Study of Heredity and Environment (Chicago: University of Chicago Press,
1937).
9. Gordon Allen, Letter to the Editor, March 9, 1956, American Journal of
Human Genetics 8 (1956): 194–195, www.ncbi.nlm.nih.gov/pmc/articles/
PMC1716686.
340 N OT E S TO PAG E S 2 – 4

10. Nancy L. Segal, Entwined Lives: Twins and What They Tell Us about Hu-
man Behavior (New York: Plume, 2000).
11. Kaare Christensen, James W. Vaupel, Niels V. Holm, and Anatoli I. Yashin,
“Mortality among Twins after Age 6: Fetal Origins Hypothesis versus Twin
Method,” British Medical Journal 310 (1995): 432–436.
12. John C. Loehlin and Robert C. Nichols, Heredity, Environment and Person-
ality: A Study of 850 Sets of Twins (Austin: University of Texas Press, 1976).
13. Allen Morris-Yates, G. Andrews, P. Howie, and S. Henderson, “Twins: A Test
of the Equal Environments Assumption,” Acta Psychiatrica Scandinavica 81
(1990): 322–326.
14. Adam P. Matheny, Ronald S. Wilson, and Anne B. Dolan, “Relations be-
tween Twins’ Similarity of Appearance and Behavioral Similarity: Testing an
Assumption,” Behavior Genetics 6 (1976):343–352; Robert Plomin, Lee
Willerman, and John C. Loehlin, “Resemblance in Appearance and the Equal
Environments Assumption in Twin Studies of Personality Traits,” Behavior
Genetics 6 (1976): 43–52.
15. David C. Rowe, The Limits of Family Influence: Genes, Experience, and
Behavior (New York: Guilford Press, 1994).
16. Paul Popenoe, “Twins Reared Apart,” Journal of Heredity 5 (1922): 142–144.
17. A copy of this letter was provided to me by Professor Bouchard.
18. Hermann J. Muller, “Mental Traits and Heredity,” Journal of Heredity 5
(1925):142–144. Muller’s work was followed by a series of short papers on
reared-together twins in the same journal; for a review, see Sheldon C. Reed,
“A Short History of Human Genetics in the USA,” American Journal of
Medical Genetics 3 (1979): 282–295. Another reared-apart twin case study
was published by R. Saudek, “A British Pair of Identical Twins Reared
Apart,” Character and Personality 3 (1934): 17–39.
19. Niels Juel-Nielsen, “A Psychiatric-Psychological Investigation of Monozygous
Twins Reared Apart,” Acta Psychiatrica et Neurologica Scandinavica (1965),
Monograph Supplement 183. A follow-up study that included the original
material was published in 1980.
20. Robert J. Lifton, The Nazi Doctors: Medical Killing and the Psychology of
Genocide (New York: Basic Books, 1986).
21. Arthur Jensen, “How Much Can We Boost IQ and Scholastic Achievement?”
Harvard Educational Review 39 (1969): 1–123.
22. A similar controversy erupted in 1994 following the publication of Richard
Herrnstein and Charles Murray’s book, The Bell Curve: Intelligence and
Class Structure in American Life (New York: Free Press) which discussed pos-
sible genetically based group differences in general intelligence. See John C.
Loehlin, “History of Behavior Genetics,” in Handbook of Behavior Genetics,
ed. Yong-Kyu Kim, 3–11 (New York: Springer, 2009) for discussion of this
and other significant events in the behavioral genetics field.
23. Leon J. Kamin, The Science and Politics of IQ (Potomac, Md.: Lawrence
Erlbaum, 1974); Ann Clarke and Michael McAskie, “Parent-Offspring
Resemblances in Intelligence: Theories and Evidence,” British Journal of
Psychology 67 (1976): 243–273.
 N OT E S TO PAG E S 5 –7 341

24. Kamin, Science and Politics of IQ. A 1958 analysis of 42 MZA twin pairs by
Burt’s colleague Ms. J. Conway reported an IQ correlation of 0.778.
25. Irving I. Gottsman, interview, May 7, 2011. According to Gottesman, Shields
had invited Kamin to his laboratory to see additional photographs, but Shields
did not publish them to preserve the privacy of his participants. Gottesman also
believes that Shields was too modest to have his picture taken with the twins.
26. Stephen Davis, “Nice One, Cyril?” BBC TV, January 7–13, 1984, 10–11.
27. Robert B. Joynson, The Burt Affair (London: Routledge, 1989). Ronald
Fletcher, Science, Ideology, and the Media: The Cyril Burt Scandal (New
Brunswick, N.J.: Transaction, 1991).
28. Nicholas J. Mackintosh, Cyril Burt: Fraud or Framed? (Oxford: Oxford Uni-
versity Press, 1995).
29. James Shields, “MZA Twins: Their Use and Abuse,” in Twin Research: Part
A: Psychology and Methodology, ed. Walter E. Nance, 79–93 (New York: Alan
B. Liss, 1978); David Burbridge, “Burt’s Twins: A Question of Numbers,”
Journal of the History of the Behavioral Science 42: 35–52 (2006); William
Tucker, “Burt’s Separated Twins: The Larger Picture,” Journal of the History
of the Behavioral Science 43 (2007): 81–86.
30. Nancy L. Segal, “More Thoughts on the Child Development Center Twin
Study,” Twin Research and Human Genetics 8 (2005): 276–281. Interest-
ingly, Neubauer authored the foreword to Niels Juel-Nielsen’s, Individual
and Environment: Monozygotic Twins Reared Apart (New York: Interna-
tional Universities Press, 1980).
31. Lawrence D. Perlman, “Memories of the Child Development Study Center
Study of Adopted Monozygotic Twins Reared Apart: An Unfulfilled Prom-
ise,” Twin Research and Human Genetics 8: 271–281.
32. Segal, “More Thoughts,” 2005.
33. John Money and Anke Ehrhardt, Man and Woman, Boy and Girl (Balti-
more: Johns Hopkins University Press, 1975).
34. Milton Diamond and Keith Sigmundson, “Sex Reassignment at Birth: Long-
Term Review and Clinical Implications,” Archives of Pediatric and Adoles-
cent Medicine 151 (1997): 298–304.
35. Leo Kanner, “Autistic Disturbances of Affective Contact,” The Nervous
Child 2 (1943): 217–250.
36. Edward O. Wilson, Naturalist (New York: Warner Books, 1995).
37. Gordon Allen, review of Sandra Scarr’s Race, Social Class, and Individual
Differences in I.Q. (Baltimore: Lawrence Erlbaum, Associates, 1981), Social
Biology 30 (1983): 118–119.
38. Ullica Segerstråle, Defenders of the Truth: The Battle for Science in the
Sociobiology Debate and Beyond (Oxford: Oxford University Press, 2000).
39. John Garcia, Frank R. Ervin, and Robert A. Koelling, “Learning with Pro-
longed Delay of Reinforcement,” Psychonomic Science 5 (1966): 121–122.
40. Stephen R. Coleman, “Pavlov and the Equivalence of Associability in Classical
Conditioning,” Journal of Mind and Behavior 28 (2007): 115–134.
41. Steven G. Vandenberg, “Contributions of Twin Research to Psychology,”
Psychological Bulletin 66 (1966): 326–352. Jerome Lejeune and Patricia
342 N OT E S TO PAG E S 7– 1 0

Jacobs discovered that Down syndrome was caused by an extra chromo-

some 21.
42. The cognitive deficits associated with PKU can be largely, but not completely,
offset by administering a phenylalanine-free diet and monitoring the blood
levels of affected individuals.
43. Steven Pinker, The Blank Slate: The Modern Denial of Human Nature (New
York: Viking, 2002).
44. Noam Chomsky, “A Review of B. F. Skinner’s Verbal Behavior,” Language
35 (1959): 26–58.
45. Herbert A. Simon, “Information Processing Models of Cognition,” Annual
Review of Psychology 30 (1979): 363–396.
46. Lee Willerman, The Psychology of Individual and Group Differences (San
Francisco: W. H. Freeman, 1979).
47. David Rosenthal, ed., The Genain Quadruplets: A Case Study and Theoreti-
cal Analysis of Heredity and Environment in Schizophrenia (New York:
Basic Books, 1963); Leonard L. Heston, “Psychiatric Disorders in Foster
Home Reared Children of Schizophrenic Mothers,” British Journal of Psy-
chiatry 112 (1966): 819–825; Irving I. Gottesman and James Shields, Schizo-
phrenia and Genetics: A Twin Study Vantage Point (New York: Academic
Press, 1972).
48. Loehlin, “History of Behavior Genetics.”
49. Dr. Matthew McGue, interview, October 13, 2009.
50. Steven G. Vandenberg, Richard E. Stafford, and A. M. Brown, “The Louis-
ville Twin Study,” in Progress in Human Behavior Genetics: Recent Re-
ports on Genetic Syndromes, Twin Studies, and Statistical Advances, ed.
Steven G. Vandenberg, 153–204 (Baltimore: Johns Hopkins University
Press, 1968).
51. John C. DeFries, R. Plomin, S.  G. Vandenberg, and A.  R. Kuse, “Parent-
Offspring Resemblance for Cognitive Abilities in the Colorado Adoption
Project: Biological, Adoptive, and Control Parents and One-Year-Old Chil-
dren,” Intelligence 5 (1981): 245–277.
52. Shields, “MZA Twins: Their Use and Abuse.”
53. Thomas J. Bouchard Jr., “The Hereditarian Research Program: Triumphs
and Tribulations,” in Arthur Jensen: Consensus and Controversy, ed. Sohan
Modgil and Celia Modgil, 55–76 (London: Falmer International, 1987).
54. Irving I. Gottesman, “Heritability of Personality,” Psychological Monographs:
General and Applied 77, no. 572 (1963): 1–21; Gottesman and Shields, Schizo-
phrenia and Genetics.
55. See Thomas J. Bouchard Jr., “The Drifter: A Random Walk through Psychol-
ogy” (unpublished paper, November 24, 2004).
56. Irving I. Gottesman, “Introduction,” in Juel-Nielsen, Individual and Envi-
ronment, 9.
57. James L. Jenkins and Donald G. Paterson, Studies in Individual Differences:
The Search for Intelligence (New York: Appleton-Century-Crofts, 1961).
58. Jim Lewis was remarried to a third wife named Sandy when he visited Min-
nesota. I sometimes wondered if that concerned Jim Springer’s wife, Betty.
59. Comment by former graduate student assistant Susan Resnick.
 N OT E S TO PAG E S 1 0 – 1 9 343

60. Aya Ito, Yoko Honma, Emiko Inamori, Yukari Yada, Mariko Y. Momoi, and
Yoshikazu Nakamura, “Developmental Outcome of Very Low Birth Weight
Twins Conceived by Assisted Reproduction Techniques,” Journal of Perinatol-
ogy 26 (2006): 130–133.
61. Elizabeth F. DiLalla and Paula Y. Mullineaux, “The Effects of Classroom
Environment on Problem Behaviors: A Twin Study,” Journal of School Psy-
chology 46 (2007): 107–128.
62. Constance Holden, “Identical Twins Reared Apart,” Science 207 (1980):
1323–1327.
63. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962).
64. Nancy L. Segal and Yoon-Mi Hur, “Reared Apart Korean Female Twins: Ge-
netic and Cultural Influences on Life Histories, Physical and Health-Related
Measures, and Behavioral Traits,” International Journal of Behavioral Devel-
opment 32 (2008): 542–548.
65. Nancy L. Pedersen, Paul Lichtenstein, and Pia Svdberg, “The Swedish Twin
Registry in the Third Millennium,” Twin Research and Human Genetics 5
(2002): 427–432.
66. Kari Kervinen, J. Kaprio, M. Koskenvuo, J. Juntunen, and Y. A. Kesäniemi,
“Serum Lipids and Apolipoprotein E Phenotypes in Identical Twins Reared
Apart,” Clinical Genetics 53 (1998): 191–199.
67. Kazuo Hayakawa and Tadahiko Shimizu, “Blood Pressure Discordance and
Lifestyle: Japanese Identical Twins Reared Apart and Together,” Acta Gene-
ticae Medicae et Gemellologiae 36 (1987): 485–491.
68. Shields provided brief accounts of one DZA male and ten DZA female twin
pairs in his 1962 study.
69. Thomas J. Bouchard Jr., “Memo to the Twin Research Team,” March 5, 1979.
The second pair to participate in the MISTRA were DZA female twins.
70. Tina Adler, “Seeing Double? Controversial Twins Study Widely Reported, De-
bated,” The APA Monitor 22 (1991): 1, 8; Robert Karen, Becoming Attached:
First Relationships and How They Shape Our Capacity to Love (Oxford:
Oxford University Press, 1998): 307–308.
71. Barbara Gonyo, “Genetic Sexual Attraction,” Decree 4 (1987): 1, 5. It is be-
lieved that humans evolved cues to avoid mating with close relatives, and one is
associated with cohabitating as children. This is called the Westermarck Effect.

1. The Jim Twins (February–March 1979)

1. Rosemary Rawson, “Two Ohio Strangers Find They’re Twins at 39—and a
Dream to Psychologists,” People magazine, May 7, 1979, 117–120.
2. Mike Lackey, “Identical Twins, Separated as Babies in 1939, Reunited,”
Lima [Ohio] News, February 19, 1979.
3. Barbara S. Burks and Anne Roe, “Studies of Identical Twins Reared Apart,”
Psychological Monographs 63 (1949).
4. Joy Stilley, “Never Apart Again, Twins Say,” Tuscaloosa News, November
16, 1969, 22; Bard Lindeman, The Twins Who Found Each Other (New York:
Morrow, 1969).
344 N OT E S TO PAG E S 1 9 – 32

5. “Interview with Thomas J. Bouchard, Jr.,” Intelligence Online, www.isiron-

line.org, Distinguished Contributor Interview, 2004.
6. Lykken’s account of the Jim twins is described in his book Happiness: What
Studies on Twins Show Us about Nature, Nurture, and the Happiness Set Point
(New York: Golden Books, 1999). The material in this chapter was provided
by Bouchard; Lykken was deceased when my interviews occurred.
7. David Healy, The Psychopharmacologists II: Interviews by Dr. David Healey
(London: Altman, 1998).
8. Auke Tellegen, A Brief Manual for the Differential Personality Question-
naire (unpublished manuscript, University of Minnesota, 1978); Auke Telle-
gen, “Structure of Mood and Personality and Their Relevance to Assessing
Anxiety, with an Emphasis on Self-Report,” in Anxiety and the Anxiety Dis-
orders, ed. A. Hussain Tuma and Jack D. Maser, 681–706 (Hillsdale, N.J.:
Erlbaum, 1985).
9. Jack Darley’s son, John, is the Warren Professor of Psychology at Princeton
University.
10. Dr. Polesky retired in about 2000 just after the MISTRA ended.
11. Edwin Chen, “Twins Reared Apart: A Living Lab,” New York Times Maga-
zine, December 9, 1979, 112–130.
12. Donald D. Johnson, “Reunion of Identical Twins, Raised Apart, Reveals
Some Astonishing Similarities,” Smithsonian, October 1980, 48–56, p. 53.
I found notes from a Bouchard lecture, “MZAs: Are They Really Identical?”
November 10, 1982, in which the first thing Bouchard said was, “The find-
ings are hard to believe.”
13. Mike Lackey, “Jim Twins Study Spurs Detailed U-Minn. Project,” Lima
[Ohio] News, May 14, 1979.
14. Chen, “Twins Reared Apart.”
15. Katherine Kam, “His Guide to a Heart Attack: Symptoms in Men,” Febru-
ary  14, 2007, www.webmd.com/heart-disease/features/his-guide-to-a-heart
-attack.
16. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962).
17. Confidentiality regulations prevented the release of such information by the
Minnesota investigators.
18. The last pair underwent initial assessment in April 1999.
19. Constance Holden, “Identical Twins Reared Apart,” Science 207 (1980):
1323–1327.
20. Healy, The Psychopharmacologists II.
21. “Interview with Thomas J. Bouchard, Jr.”

2. 15,000 Questions × 137 Pairs

1. Nicholas G. Martin, Lindon J. Eaves, Andrew C. Heath, R. Jardine, L. M.
Feingold, and H. J. Eysenck, “Transmission of Social Attitudes,” Proceedings
of the National Academy of Sciences of the United States of America 83
(1986): 4364–4368.
 N OT E S TO PAG E S 3 5 – 47 345

2. “Separation at the Cradle 75 Years Ago for Twins,” New Zealand news-
paper, unidentified, from Bouchard’s newspaper archive. The estimated date
is 1989, the year the twins were reunited.
3. Bent Harvald, Gudrun Hauge, Kirsten Ohm Kyvik, Kaare Christensen, Axel
Skytte, and Niels V. Holm, “The Danish Twin Registry: Past and Present,”
Twin Research and Human Genetics 7 (2004): 318–335.
4. He Mingguang, “Guangzhou Twin Registry and Guangzhou Eye Study,” ab-
stract, paper presented at the 13th International Congress on Twin Studies,
Seoul, South Korea, June 4–7, 2010.
5. The number of DNA markers analyzed expanded with improvements in
laboratory techniques (Mary Mount, personal communication, November
23, 2009). In 1979, the Jim twins were compared across eight red blood cell
groups (ABO, RH, MNSs, Lewis, Kell, Duffy, Kidd, and P), seven serum pro-
teins (Gc, Hp, A, PGM, acP, EsD, and Bf), and two other factors (Gm and Inv).
In April 1999, twins in our last pair were compared across two red blood cell
systems (ABO and Rh) and five DNA fragment length polymorphisms (probe
SLi737, locus D12S11; probe SLi986, locus D17S79; probe SLi989, locus
D7S467; probe SLi1090, locus D6S132, and probe SLi106, locus D2S44).
The DNA analysis in combination with blood group analysis was introduced
in July 1994.
6. David T. Lykken, “The Diagnosis of Zygosity in Twins,” Behavior Genetics 8
(1978): 437–473.
7. Dr. Alison MacDonald is a senior lecturer at the University of East London.
8. Robert R. Race and Ruth Sanger, Blood Groups in Man (Oxford: Oxford
University Press, 1975).
9. Nancy L. Segal, “Zygosity Diagnosis: Laboratory and Investigator’s Judgment,”
Acta Geneticae Medicae et Gemellologiae, Twin Research 33 (1984): 515–520.
10. Most cases of ectodermal dysplasia are X-linked recessive, a transmission
pattern that does not fit these twins.
11. “Happiness Is a Reunited Set of Twins,” U.S. News and World Report, April
13, 1987, 63–66.
12. See Daniel Hanson, “The Gene Illusion Confusion,” review of Joseph, The
Gene Illusion (2004), PsycCRITIQUES 50, no. 52, article 14. Hanson likened
Joseph to an “adrenalized boxer” when it comes to behavioral genetic
research.
13. Jay Joseph, The Gene Illusion: Genetic Research in Psychiatry and Psychol-
ogy under the Microscope (New York: Algora, 2004).
14. Matt McGue and Thomas J. Bouchard Jr., “Adjustment of Twin Data for the
Effects of Age and Sex,” Behavior Genetics 14 (1984): 325–343.
15. Nancy L. Segal, “Oskar and Jack,” in Indivisible by Two, 50–78 (Cambridge,
Mass.: Harvard University Press, 2007).
16. “Telling Them Why,” obituary, Guardian, September 9, 1989.
17. Jill Todd, “Finding My Twin Brought Happiness Beyond Belief,” Woman,
September 1985, 40–41.
18. John Stroud, “Last Word” (estimated date 1989). Great Britain, from Boucha-
rd’s newspaper archive.
346 N OT E S TO PAG E S 4 9 – 6 1

19. Nancy L. Segal, “More Thoughts on the Child Development Center Twin
Study,” Twin Research and Human Genetics 8 (2005): 276–281.
20. William H. Tucker, “Re-Reconsidering Burt: Beyond a Reasonable Doubt,”
Journal of the History of the Behavioral Sciences 33 (1997): 145–162, p. 152.
21. Rebecca L. Hegar, “Sibling Placement in Foster Care and Adoption: An Over-
view of International Research,” Children and Youth Services Review 27
(2005): 717–739; Ilene Staff and Edith Fein, “Together or Separate: A Study
of Siblings in Foster Care,” Child Welfare 41 (1992): 257–270.
22. Jeffrey Kluger, The Sibling Effect: What the Bonds among Brothers and Sis-
ters Reveal about Us (New York: Riverhead Books, 2011).
23. Nancy L. Segal, “Implications of Twin Research for Legal Issues Involving
Young Twins,” Law and Human Behavior 17 (1993): 43–58.
24. Nancy L. Segal, Kevin A. Chavarria, and Joanne Hoven, “Twin Research:
Evolutionary Perspective on Social Relations,” in Family Relationships: An
Evolutionary Perspective, ed. Catherine A. Salmon and Todd K. Shackelford,
312–335 (New York: Oxford University Press, 2007): 312–333; Adam Brookes,
“China’s Unwanted Girls,” BBC News Online, August 23, 2001, http://news
.bbc.co.uk/2/hi/asia-pacific/1506469.stm.
25. Nancy L. Segal and Iris Blandón-Gitlin, “Twins Switched at Birth: A Case
from the Canary Islands,” Twin Research and Human Genetics 13 (2010):
115–119; Nancy L. Segal, Someone Else’s Twin: The True Story of Babies
Switched at Birth (New York: Prometheus Books, 2011).
26. Antonio Garrido-Lestache, “Identification of New-Born Babies by Finger-
prints,” International Criminal Police Review 481 (2000): 19–24.
27. James Shields, “Twins Brought Up Apart,” Eugenics Review 50 (1958): 115–
123, p. 117.
28. Charles W. Mueller and Toby L. Parcel, “Measures of Socioeconomic Status:
Alternatives and Recommendations,” Child Development 52 (1981): 13–21.
29. “Body Watch,” WGBH, Boston in conjunction with American Health Maga-
zine, June 31, 1987, available on video.
30. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962), 94.
31. “Epistasis: Puzzling Inheritance Patterns Explained,” www.answers.com/
topic/epistasis (accessed 7/6/11).
32. Robert Plomin, John C. DeFries, Gerald E. McClearn, and Peter McGuffin,
Behavioral Genetics, 5th ed. (New York: Worth Publishers, 2008).
33. Michael C. Neale and Lon R. Cordon, Methodology for Genetic Studies of
Twins and Families (Dordrecht, the Netherlands: Springer, 2010).
34. It is assumed that the overall variance of the trait does not differ between the
two types of twins.
35. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A Study
of Heredity and Environment (Chicago: University of Chicago Press, 1937).
36. Hermine H. M. Maes, Gaston P. Beunen, Robert F. Vlietinck, et al., “Inheri-
tance of Physical Fitness in 10-Year-Old Twins and Their Parents,” Medicine
and Science in Sports and Exercise 28 (1996): 1479–1491.
37. The statistical reliability of heritability estimates is twenty times higher if
based on MZA twins rather than MZT twins. David T. Lykken, Seymour
 N OT E S TO PAG E S 62 – 6 8 347

Geisser, and Auke Tellegen, “Heritability Estimates from Twin Studies: The Ef-
ficiency of the MZA Design” (unpublished manuscript, Department of Psy-
chology, University of Minnesota, Minneapolis, 1981).
38. Douglas S. Falconer and Trudy F. C. Mackay, Introduction to Quantitative
Genetics, 4th ed. (New York: Longman, 1996).
39. Thomas J. Bouchard Jr. and Matthew McGue, “Familial Studies of Intelli-
gence: A Review,” Science 212 (1981): 1055–1059.
40. Thomas J. Bouchard Jr., Nancy L. Segal, Auke Tellegen, Matt McGue, Marga-
ret Keyes, and Robert Krueger, “Genetic Influence on Social Attitudes: An-
other Challenge to Psychology from Behavior Genetics,” in Behavior Genetic
Principles: Development, Personality and Psychopathology, ed. Lisabeth F.
DiLalla, 89–104 (Washington, D.C.: American Psychological Association,
2004). If a trait shows nonadditive genetic variance, then heritability is over-
estimated by doubling the difference in the MZT and DZT correlations.
That is because MZ twins share all sources of genetic variance, both additive
and nonadditive.
41. McGue and Bouchard, “Adjustment of Twin Data.”
42. Plomin, Behavioral Genetics.

3. Early Findings (1979–1983)

Epigraph: David T. Lykken, “Research with Twins: The Concept of Emergen-
esis,” Psychophysiology 19 (1981): 361–373, p. 361.
1. Thomas J. Bouchard Jr., Leonard Heston, Elke Eckert, Margaret Keyes, and
Susan Resnick, “The Minnesota Study of Twins Reared Apart: Project De-
scription and Sample Results in the Development Domain,” in Twin Research
3: Part B. Intelligence, Personality and Development, ed. Luigi Gedda, Paolo
Parisi, and Walter Nance, 227–333 (New York: Alan R. Liss, Inc., 1981).
2. The original paper lists the twins’ age as twenty-four, but they were twenty-
three at the time of their assessment.
3. These twins had had varying degrees of contact over the years, but had been
apart for a total of fifty-four years.
4. Bouchard et al., “Minnesota Study of Twins Reared Apart,” p. 223.
5. David Rosenthal, ed., The Genain Quadruplets: Case Study and Theoretical
Analysis of Heredity and Environment in Schizophrenia (New York: Basic
Books, 1963).
6. Elke D. Eckert, Leonard L. Heston, and Thomas J. Bouchard Jr. (1981), “MZ
Twins Reared Apart: Preliminary Findings of Psychiatric Disturbances and
Traits . . . ,” in Twin Research 3: Part B. Intelligence, Personality and Devel-
opment, ed. Luigi Gedda, Paolo Parisi, and Walter Nance, 179–188 (New
York: Alan R. Liss, Inc., 1981).
7. Frank C. Verhulst and Herma J.  M. Verluis-Den Bieman, “Developmental
Course of Problem Behaviors in Adolescent Adoptees,” Journal of the Ameri-
can Academy of Child and Adolescent Psychiatry 34 (1995): 151–158.
8. Margaret A. Keyes, Anu Sharma, Irene J. Elkins, William G. Iacono, and
Matt McGue, “The Mental Health of US Adolescents Adopted in Infancy,”
Archives of Pediatric and Adolescent Medicine 162 (2008): 419–425; Anu R.
348 N OT E S TO PAG E S 6 8 –7 1

Sharma, Matthew K. McGue, Peter L. Benson, “The Psychological Adjust-

ment of United States Adopted Adolescents and Their Nonadopted Siblings,”
Child Development 69 (1998): 791–802.
9. Thomas J. Bouchard Jr., “The Study of Mental Ability Using Twin and Adop-
tion Designs,” in Twin Research 3: Part B. Intelligence, Personality, and De-
velopment, ed. Luigi Gedda, Paolo Parisi, and Walter Nance, 21–23 (New
York: Alan R. Liss, Inc., 1981).
10. The objectors agreed to let Bouchard turn on a tape recorder during his lec-
ture, but they never protested again nor appeared in his class.
11. “Freedom of Expression,” editorial, Minnesota Daily, April 4, 1974. A cam-
pus debate on academic freedom was held several weeks later.
12. Constance Holden, “Behavioral Geneticist Celebrates Twins Scorns PC Sci-
ence,” Science 325 (2009): 27. Constance Holden, who covered the MISTRA
on many occasions, was killed in a car accident in 2010.
13. Tina Adler, “Seeing Double? Controversial Twins Study Widely Reported,
Debated,” The APA Monitor 22 (1991): 1, 8.
14. Sam Fulwood III, “Scholars Gather to Debate Genetics-Crime Research:
Conference: Meeting in Maryland Will Explore the Implications of Study
into Whether Some People are Inherently Predisposed toward Criminal Be-
havior,” http://articles.latimes.com/1995-09-93/news/mn-49167_1_criminal
-behavior, September 23, 1995.
15. David Lykken, “Statistical Significance in Psychological Research,” Psycho-
logical Bulletin 70 (1968): 151–159.
16. Howard F. Taylor, The IQ Game: Methodological Inquiry into the Heredity/
Environment Controversy (New Brunswick, N.J.: Rutgers University Press,
1980).
17. Taylor’s book was viewed favorably by critics of twin research. See, for ex-
ample, a review by Jerry Hirsch in Social Biology 30 (1983): 116–118.
18. Thomas J. Bouchard Jr., “Do Environmental Similarities Explain the Simi-
larity in Intelligence of Identical Twins Reared Apart?” Intelligence 7 (1983):
175–184.
19. Susan L. Farber, Twins Reared Apart: A Reanalysis (New York: Basic Books,
1981).
20. Positive reviews: John D. Rainer, Social Biology 28 (1981): 163–165;
J. Thomas Dalby, Journal of School Psychology 20 (1982): 256–258; James
R. Stabenau, Psychoanalytic Quarterly 53 (1984): 121–129; Howard E. Gru-
ber, “Nature vs. Nurture: A Natural Experiment,” New York Times, March 1,
1981. Interestingly, in that same review, Gruber offered a positive appraisal of
Niels Juel-Nielsen’s book Individual and Environment: Monozygotic Twins
Reared Apart (New York: International Universities Press, 1980). See also
John Rosegrant, Journal of Personality Assessment 46 (1982): 659–660.
Rosegrant asserted that the previous reared-apart twin data were “too poor
to be more than impressionistic” and that the MISTRA promised “more
trustworthy data.”
Negative reviews: Richard J. Rose, “Separated Twins: Data and Their
Limits,” Science 215 (1982): 959–960; Irving I Gottesman, American Journal
 N OT E S TO PAG E S 7 1 –7 3 349

of Psychology 95 (1982): 350–352; Gregory Carey, American Journal of

Psychiatry 139 (1982): 377; Peter McGuffin, Psychiatric Annals 12 (1982):
358. A review by Sandra Scarr, Journal of Social and Biological Structures 6
(1983): 89–91, applauded Farber’s efforts at organizing the material but be-
lieved Farber fell into “Kamin’s (1974) trap of wanting to divide and con-
quer the intellectual results for seemingly political reasons” (p. 90); Scarr’s
comment was in reference to Leon J. Kamin, The Science and Politics of IQ
(Potomac, Md.: Lawrence Erlbaum, 1974).
21. W. Joseph Wyatt and Donna M. Midkiff, “Biological Psychiatry: A Practice
in Search of a Science,” Behavior and Social Issues 15 (2006): 132–151.
22. Jack Kaplan, “How to Inherit IQ: An Exchange,” letter to the editor, New
York Review of Books, March 15, 2007, 56.
23. Frank Sulloway, “How to Inherit IQ: An Exchange,” reply to the editor, New
York Review of Books, March 15, 2007, 56.
24. Richard E. Nisbett, Intelligence and How to Get It: Why Schools and Cul-
tures Count (New York: W. W. Norton, 2009).
25. Ibid., 26.
26. Dan Agin, “How to Inherit IQ: The Fetal Question,” letter to the editor, New
York Review of Books, October 25, 2007, 82.
27. Thomas J. Bouchard Jr. and Matt McGue, “Genetic and Environmental In-
fluences on Human Psychological Differences,” Journal of Neurobiology 54
(2003): 4–45.
28. Greater personality similarity has been found between later-splitting MZT
twins (those more likely to be subject to fetal transfusion syndrome) than
earlier-splitting MZT twins, although the reasons are unclear. Deborah   K.
Sokol, Cynthia A. Moore, Richard J. Rose, et al., “Intrapair Differences in
Personality and Cognitive Ability among Young Monozygotic Twins Distin-
guished by Chorion Type,” Behavior Genetics 25 (1995): 457–466.
29. Nancy L. Segal, Shirley A. McGuire, June Havlena, Patricia Gill, and Scott L.
Hershberger, “Intellectual Similarity of Virtual Twin Pairs: Developmental
Trends,” Personality and Individual Differences 42 (2007): 1209–1219.
30. Sandra Scarr, Richard A. Weinberg, and Irwin D. Waldman, “IQ Correlations
in Transracial Adoptive Families,” Intelligence 17 (1993): 541–555.
31. Matt McGue et al., “A Comparison of Identical Twins Reared Apart on a
Battery of Information Processing Measures,” paper presented at the 11th
Annual Behavior Genetics Association, Purchase, N.Y., 1981.
32. Thomas J. Bouchard Jr. and David T. Lykken, “Similarity in Intelligence in
Monozygotic Twins Reared Apart” (abstract), and Irving I. Gottesman,
Gregory C. Carey, and Thomas J. Bouchard Jr., “MMPI Personality Scale
Similarity in 26 Pairs of Identical Twins Reared Apart,” abstract, Behavior
Genetics 12 (1982): 578, 585–586.
33. David T. Lykken, “Research with Twins: The Concept of Emergenesis,” Psy-
chophysiology 19 (1981): 361–373.
34. Thomas J. Bouchard Jr., “Twins Reared Together and Apart: What They Tell
Us about Human Diversity,” in Individuality and Determinism: Chemical and
Biological Bases, ed. Sidney W. Fox, 147–178 (New York: Plenum, 1984).
350 N OT E S TO PAG E S 7 3 –78

35. John M. Hettema, Michael C. Neale, and Kenneth S. Kendler, “Physical Simi-
larity and the Equal-Environment Assumption in Twin Studies of Psychiatric
Disorders,” Behavior Genetics 25 (1995): 327–335.
36. Thomas J. Bouchard Jr., David T. Lykken, Nancy L. Segal, and Kimerly J.
Wilcox, “Development in Twins Reared Apart: A Test of the Chronogenetic
Hypothesis,” in Human Growth: A Multidisciplinary Review, ed. Andrea
Demirjian, 299–310 (London: Taylor & Francis, 1986), 301.
37. David T. Lykken, Auke Tellegen, and William G. Iacono, “EEG Spectra in
Twins: Evidence for a Neglected Mechanism of Genetic Determination,”
Physiological Psychology 10 (1982): 60–65.
38. David T. Lykken, Auke Tellegen, and Karen Thorkelson, “Genetic Determina-
tion of EEG Frequency Spectra,” Biological Psychology 1 (1974): 245–259.
39. Niels Juel-Nielsen, Individual and Environment: Monozygotic Twins Reared
Apart (New York: International Universities Press), 68.
40. The sixth channel was the electrooculogram.
41. The Shafer event is an EEG task in which the multiple EEG epochs at about
250 msec prestimulus and 750 msec poststimulus are averaged. Noise in the
EEG averages to baseline, and the remaining variation is the brain’s response
to the stimulus.
42. Lykken was an early advocate of using silver/silver chloride electrodes in
order to obtain unbiased recordings. Don Fowles, interview, May 20, 2011.
43. The two age- and sex-corrected scores were transformed to a mean of 50
and a standard deviation of 10 before analysis.
44. The composite IQ (R + M) was calculated by converting the Raven and Mill-
Hill to z-scores and adding them.
45. Gerrit J. S. Wilde, “Inheritance of Personality Traits,” Acta Psychologia 22
(1964): 37–51.
46. Gordon Claridge, Sandra Canter, and W.  I. Hume, Personality Differences
and Biological Variations: A Study of Twins (Oxford: Pergamon Press,
[1974]).
47. Daniel Hankins, Charles Drage, Noe Zambel, and Richard Kronenberg,
“Pulmonary Function in Identical Twins Raised Apart,” American Review of
Respiratory Disease 125 (1982): 119–121.
48. This is called the forced expiratory flow rate (25–75 percent of the total
forced vital capacity, or maximum volume of gas exhaled from full
inhalation).
49. This is called the instantaneous expiratory flow rate (75 percent of the total
forced vital capacity).
50. This is called the forced expiratory volume (amount of air expelled during
the first second of the forced vital capacity maneuver).
51. A recent large study found that twin concordance for smoking or nonsmok-
ing does not affect the heritability of pulmonary measures; see Truls S. Inge-
brigtsen, Simon F. Thomsen, Sophie van der Sluis, et al., “Genetic Influences
on Pulmonary Function: A Large Sample Twin Study,” Lung 189 (2011):
323–330.
 N OT E S TO PAG E S 8 3 – 8 7 351

4. Sexual Orientation, Cognition, and Medical Traits (1984–1987)

1. Some researchers thought that this was because processing speed was a ma-
jor component of general intelligence, but others thought it reflected variable
associations between information processing and special mental abilities.
2. Matt McGue, Thomas J. Bouchard Jr., David T. Lykken, and Dale Feuer, “In-
formation Processing Abilities in Twins Reared Apart,” Intelligence 8 (1984):
239–258.
3. Michael I. Posner, Stephen J. Boies, William H. Eichelman, and Richard L.
Taylor, “Retention of Visual and Name Codes of Single Letters,” Journal of
Experimental Psychology 79 (1969): 1–16.
4. Saul Sternberg, “The Discovery of Processing Stages: Extension of Donders’
Method,” Acta Psychologia 30 (1969): 276–315.
5. Roger Shepard and Jacqueline Metzler, “Mental Rotation of Three Dimen-
sional Objects,” Science 171 (1971): 701–703.
6. The twins’ mean correct response times were regressed against the size of the
memory sets.
7. The twins’ median reaction time for correct (“same”) responses was re-
gressed against the angular displacement of the figures.
8. These results contradict the idea that biological differences have their great-
est effect on rate of cognitive processing.
9. Rudolf Barcal, Josef Sova, Miloslava Křižanouská, J. Levy, and Jiři Matousěk,
“Genetic Background of Circadian Rhythms,” Nature 220 (1968): 1128–1131.
10. Bruce R. Hanson, Franz Halberg, Naip Tuna, et al., “Rhythmometry Reveals
Heritability of Circadian Characteristics of Heart Rate of Human Twins
Reared Apart,” Cardiologia (Bulletin of the Italian Society of Cardiology) 29
(1984): 267–282, p. 268.
11. This device was an ambulatory electrocardiocorder (model 445B) manufac-
tured by Delmar Avionics in Irvine, California.
12. Franz Halberg, Eugene A. Johnson, Walter Nelson, Walter Runge, and Robert
Sothern, “Autorhythmometry Procedures for Physiologic Self-Measurements
and Their Analysis,” Physiology Teacher 1 (1972): 1–11.
13. Hanson et al., “Rhythmometry Reveals Heritability,” 279.
14. Franz Halberg, Germaine Cornélissen, George Katinas, et al., “Transdisci-
plinary Unifying Implications of Circadian Findings in the 1950s,” Journal
of Circadian Rhythms 1 (2003), doi:10.1186/1740-3391-1-2.
15. The twins’ difference in sexual activity during heart rate monitoring could
be considered a nongenetic effect on their heart functioning. However, one
twin’s decision to have sex and to let the researchers know could reflect the
sisters’ genetically based differences in extraversion or risk-taking.
16. Mathea R. Allensmith, Barbara McClellan, and Michael Butterworth, “The
Influence of Heredity and Environment on Human Immunoglobulin Levels,”
Journal of Immunology 102 (1969): 1504–1510.
17. Antibody titers measure the level of antibodies in the blood.
18. We assessed antibodies in response to type 14 pneumococcal capsular
polysaccharides.
352 N OT E S TO PAG E S 8 7– 9 1

19. Peter F. Kohler, Victor J. Rivera, Elke D. Eckert, Thomas J. Bouchard Jr., and
Leonard L. Heston, “Genetic Regulation of Immunoglobulin and Specific
Antibody Levels in Twins Reared Apart,” Journal of Clinical Investigation 75
(1985): 883–888, p. 887.
20. William H. Knobloch, Nancy M. Leavenworth, Thomas J. Bouchard Jr., and
Elke D. Eckert, “Eye Findings in Twins Reared Apart,” Ophthalmic Paediat-
rics and Genetics 5 (1985): 59–66.
21. Not all features of the retina, such as the pattern of blood vessels, are identical
for MZ twins. “Retinal Biometrics,” October 10, 2011, http://360biometrics
.com/faq/Retinal-Biometrics.php.
22. Hanne Nødgaard, Helle Hansen, Helle Andreasen, and Henrik Toft Sø-
rensen, Hanne Norggaard, “Risk Factors Associated with Retinopathy of
Prematurity (ROP) in Northern Jutland, Denmark 1990–1993,” Acta Oph-
thmologica Scandinavica, 74 (1996): 306–310.
23. Elke D. Eckert, Thomas J. Bouchard Jr., Joseph Bohlen, and Leonard L. Heston,
“Homosexuality in Twins Reared Apart,” British Journal of Psychiatry 148
(1986): 421–425, p. 421.
24. Leonard L. Heston and James Shields, “Homosexuality in Twins: A Family
Study and a Registry Study,” Archives of General Psychiatry 18 (1968):
149–160.
25. J. Michael Bailey, “Texas, John Loehlin, and the Genetics of Sexual Orienta-
tion,” paper presented at the Festschrift Honoring John Loehlin, Newport,
Rhode Island, June 6, 2011.
26. J. Michael Bailey, Michael P. Dunne, and Nicholas G. Martin, “Genetic and
Environmental Influences on Sexual Orientation and Its Correlates in an
Australian Twin Sample,” Journal of Personality and Social Psychology 78
(2000): 524–536.
27. Niklas Långström, Qazi Rahman, Eva Carlström, and Paul Lichtenstein, “Ge-
netic and Environmental Effects on Same-Sex Sexual Behavior: A Population
Study of Twins in Sweden,” Archives of Sexual Behavior 39 (2010): 75–80.
28. Scott L. Hershberger and Nancy L. Segal, “The Cognitive, Behavioral, and
Personality Profiles of a Male Monozygotic Triplet Set Discordant for Sex-
ual Orientation,” Archives of General Psychiatry 33 (2004): 497–514;
Nancy L. Segal, “Female to Male: Two Monozygotic Twin Pairs Discordant
for Transsexualism,” Archives of Sexual Behavior 35 (2006): 347–358; Ray
Blanchard, Kenneth J. Zucker, Ana Cavacas, Sara Allin, Susan J. Bradley, and
Debbie C. Schachter, “Birth Order and Sibling Sex Ratio in Homosexual
Male Adolescents and Probably Prehomosexual Feminine Boys,” Develop-
mental Psychology 31 (1995): 22–30.
29. Niels Juel-Nielsen, Individual and Environment: Monozygotic Twins Reared
Apart (New York: International Universities Press, 1980); James Shields,
Monozygotic Twins Brought Up Apart and Brought Up Together (London:
Oxford University Press, 1962).
30. Benjamin Radford, “2 Percent of Americans Identify as Gay,” Discovery
News, April 11, 2011, http://news.discovery.com/human/about-2-of-ameri-
cans-are-gay-110411.html.
 N OT E S TO PAG E S 9 1 – 9 6 353

31. Elke D. Eckert, Thomas J. Bouchard Jr., Joseph Bohlen, and Leonard L. Heston,
“Homosexuality in Twins Reared Apart,” British Journal of Psychiatry 148
(1986): 421–425.
32. Ibid.
33. See Gardner Lindzey, “Some Remarks Concerning Incest, the Incest Taboo,
and Psychoanalytic Theory,” American Psychologist 22 (1967): 1051–1059,
for a fascinating treatment of this topic.
34. Barbara Gonyo, “Genetic Sexual Attraction,” American Adoption Congress,
Decree 4 (Winter 1987): 1, 5.
35. Karri Silventoinen, Jaakko Kaprio, Eero Lahelma, et al., “Assortative Mating
by Body Weight and BMI: Finnish Twins and Their Spouses,” American Jour-
nal of Human Biology 15 (2003): 620–627; Robert Plomin, John C. DeFries,
Gerald E. McClearn, and Peter McGuffin, Behavioral Genetics: A Primer,
5th ed. (New York: Worth Publishers, 2008).
36. Suma Jacob, Martha K. McClintock, Bethanne Zelano, and Carole Ober,
“Paternally Inherited HLA Alleles Are Associated with Women’s Choice of
Male Odor,” Nature Genetics 30 (2002): 175–179.
37. Diane Amery, “Couple Wed 20 Years Discover They’re Twins,” The Sun
[London], December 17, 1985, 19.
38. Niels Juel-Nielsen, 1980.
39. Eckert et al., “Homosexuality in Twins,” 424.
40. Ibid., 425.
41. Hillary Lips, Sex and Gender: An Introduction (New York: McGraw-Hill,
2005).
42. Thomas J. Bouchard Jr., “Twins Reared Together and Apart: What They Tell
Us about Human Diversity,” in Individuality and Determinism: Chemical and
Biological Bases, ed. Sidney W. Fox, 147–184 (New York: Plenum, 1984).
43. Thomas J. Bouchard Jr., David T. Lykken, Nancy L. Segal, and Kimerly J.
Wilcox, “Development in Twins Reared Apart: A Test of the Chronogenetic
Hypothesis,” Human Growth: A Multidisciplinary Review, ed. Andrea Demir-
jian, 299–310 (London: Taylor & Francis, 1986).
44. Linda K. Dixon and Ronald C. Johnson, The Roots of Individuality (Bel-
mont, Calif.: Wadsworth, 1980), 96.
45. Carole A. Roberts and Charles B. Johansson, “The Inheritance of Cognitive
Interest Styles among Twins,” Journal of Vocational Behavior 4 (1974): 237–
243. The Holland scales can be derived from subjects’ responses to the Strong
items, although the content is not strictly comparable between the two instru-
ments. In addition, the 1974 intraclass correlations may be inflated because
they were not age- and sex-corrected.
46. Harold D. Grotevant, Sandra Scarr, and Richard Weinberg, “Patterns of In-
terest Similarity in Adoptive and Biological Families,” Journal of Personality
and Social Psychology 35 (1977): 667–676.
47. JoAnn C. Boraas, Louise B. Messer, and Michael J. Till, “Dental Characteris-
tics of Twins Reared Apart,” 65th Annual Meeting of the International Soci-
ety for Dental Research, Chicago, Journal of Dental Research 66 (1987),
abstract no. 27.
354 N OT E S TO PAG E S 9 6 – 1 0 1

48. In 2007, Stephen Rich became Director of the Center for Public Health Ge-
nomics at the University of Virginia, Charlottesville.

5. Pivotal Papers: Personality and IQ (1988 and 1990)

1. Daniel Goleman, “Major Personality Study Finds That Traits Are Mostly
Inherited,” New York Times, December 2, 1986, 17–18, p. 17.
2. Auke Tellegen, David T. Lykken, Thomas J. Bouchard Jr., Kimerly J. Wilcox,
Nancy L. Segal, and Stephen Rich, “Personality Similarity in Twins Reared
Apart and Together,” Journal of Personality and Social Psychology 54 (1988):
1031–1039. This paper has had over 800 citations in Google Scholar as of
June 2011 (at 100 citations, a paper is considered a “Social Science Citation
Classic”).
3. Lisa Lynch, “Twins’ Personality Study Made Big Media Splash before Jour-
nal Approval,” Science Writer, Winter 1990, 15–19, p. 19.
4. See, for example, Lois W. Hoffman, “The Influence of the Family Environ-
ment on Personality: Accounting for Sibling Differences,” Psychological Bul-
letin 110 (1991): 187–203; John S. Lang, “How Genes Shape Personality,”
U.S. News and World Report, April 13, 1987, 58–66.
5. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A Study
of Heredity and Environment (Chicago: University of Chicago Press, 1937).
6. Incomplete data for the separated twins precluded further personality com-
parisons between the MZA and MZT pairs.
7. Ibid., 332 p. 332.
8. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962). Shields administered the
Self-Rating Questionnaire (SRQ), which was designed for his study by Pro-
fessor Hans J. Eysenck. The SRQ was the forerunner of the Eysenck Person-
ality Questionnaire.
9. Robert Plomin, John C. DeFries, Gerald E. McClearn, and Peter McGuf-
fin, Behavioral Genetics: A Primer, 5th ed. (New York: Worth Publishers,
2008).
10. Shields conducted brief assessments of eleven DZA pairs.
11. David T. Lykken, Thomas J. Bouchard Jr., Matt McGue, and Auke Tellegen,
“The Minnesota Twin Family Registry: Some Initial Findings,” Acta Gene-
tice Medicae et Gemellologiae, Twin Research 39 (1990): 35–70.
12. Auke Tellegen, “Structure of Mood and Personality and Their Relevance to
Assessing Anxiety, with an Emphasis on Self-Report,” in Anxiety and the
Anxiety Disorders, ed. A. Hussain Tuma and Jack D. Maser, 681–706 (Hills-
dale, N.J.: Erlbaum, 1985).
13. A fifth inventory booklet was added toward the end of the study.
14. Quoted in Constance Holden, “The Genetics of Personality” Science 237
(1987): 598–599.
15. Ibid., 600.
16. John Leo and Elizabeth Taylor, “Exploring the Traits of Twins,” Time maga-
zine, January 12, 1987, 63.
 N OT E S TO PAG E S 1 0 1 – 1 0 5 355

17. Goleman, “Major Personality Study.”

18. Richard J. Rose, Markku Koskenvuo, Jaakko Kaprio, Seppo Sarna, and
Heimo Langinvainio, “Shared Genes, Shared Experiences, and Similarity of
Personality: Data from 14,288 Adult Finnish Co-Twins,” Journal of Person-
ality and Social Psychology 54 (1988): 161–171.
19. David T. Lykken, Matt McGue, Thomas J. Bouchard Jr., and Auke Tellegen,
“Does Contact Lead to Similarity or Similarity to Contact?” Behavior Gene-
tics 20 (1990): 547–561.
20. Richard J. Rose, Jaakko Kaprio, Christopher J. Williams, Richard Viken, and
Karen Obremski, “Social Contact and Sibling Similarity: Facts, Issues, and
Red Herrings,” Behavior Genetics 20 (1990): 763–778.
21. Richard J. Rose was Lykken’s doctoral student at the University of Minne-
sota, and received his Ph.D. in 1964.
22. Dr. Richard J. Rose, personal correspondence.
23. Michael V. Ellis and Erica S. Robbins, “In Celebration of Nature: A Dialogue
with Jerome Kagan,” Journal of Counseling and Development 68 (1990):
623–627.
24. Interestingly, Aristotle (384–322 b.c.) came up with the suggestion that dis-
tributive justice consists of treating equals equally and unequals unequally
(Nicomachean Ethics, Book V, Chapter 3, “Distributive Justice, in Accor-
dance with Geometrical Proportion”). Of course, treating children differ-
ently (or unequally) does not imply, or condone, treating one more favorably
than another.
25. This was expected based on the DZA and DZT twins’ intraclass correlations
being less than half as large as those of the MZA and MZT twins.
26. The higher order factor of Positive Emotionality, which includes Social
Closeness, also showed within-family effects.
27. Nightline, ABC, October 1989.
28. Thomas J. Bouchard Jr., David T. Lykken, Matthew McGue, Nancy L. Segal,
and Auke Tellegen, “Sources of Human Psychological Differences: The Min-
nesota Study of Twins Reared Apart,” Science 250 (1990): 223–228, p. 228.
(This paper has had over 900 citations in Google Scholar as of June 2011.)
29. This paper underwent many drafts, so the sample probably included fewer
than the full number of pairs that had been assessed at the time of its submis-
sion. A young set of MZA twins was also excluded from the analysis.
30. See Thomas J. Bouchard Jr., “Twins Reared Together and Apart: What
They Tell Us about Human Diversity,” in Individuality and Determinism:
Chemical and Biological Bases, ed. Sidney W. Fox, 147–184 (New York:
Plenum, 1984). The conference, “The Chemical and Biological Bases for In-
dividuality,” was sponsored by the Liberty Fund and held in Key Biscayne,
Florida.
31. Ibid.
32. David Wechsler, Manual for the Wechsler Adult Intelligence Scale (New
York: Psychological Corporation, 1955).
33. John C. Raven, Manual for Raven’s Progressive Matrices and Vocabulary
Scales (London: Lewis, 1986).
356 N OT E S TO PAG E S 1 0 7– 1 1 3

34. David T. Lykken, “Research with Twins: The Concept of Emergenesis,” Psy-
chophysiology 19 (1982): 361–373.
35. Gerald E. McClearn, Boo Johansson, Stig Berg, et al., “Substantial Genetic
Influence on Cognitive Abilities in Twins 80 or More Years Old,” Science
276 (1997): 1560–1563.
36. Nancy L. Segal, Shirley A. McGuire, June Havlena, Patricia Gill, and Scott L.
Hershberger, “Intellectual Similarity of Virtual Twin Pairs: Developmental
Trends,” Personality and Individual Differences 42 (2007): 1209–1219.
37. Christiane Capron and Michel Duyme, “Assessment of Effects of Socio-
economic Status in a Full Cross-Fostering Study,” Nature 340 (1989): 552.
38. Newman et al., Twins, 1937. Newman et al.’s IQ data were recalculated to
yield Stanford-Binet and Otis IQ intraclass correlations of 0.68 (originally
0.67) and 0.74 (originally 0.73). Note that upon omitting four “extreme
cases” (four sets of co-twins with substantial educational differences), New-
man et al. found that educational and social differences accounted for just
20 percent of the IQ variance.
39. The results were 0.06 ± 0.15 (time together before separation), 0.08 ± 0.15
(time apart to first reunion), −0.14 ± 0.15 (total contact time), and 0.17 ± 0.15
(percentage of lifetime apart).
40. A 2003 study found that heritability estimates of general intelligence were
near zero among seven-year-old twin children from impoverished families,
with 60 percent of the variance associated with shared environmental factors.
The reverse was true for twins from affluent families. The investigators indi-
cated that it would be inappropriate to claim that outcomes among children
from poor environments are more closely tied to their environments than are
outcomes among children from favorable environments. It is possible that
some genetic influences on general intelligence are common to genetic influ-
ences on socioeconomic status. Eric Turkheimer, Andreana Haley, Mary Wal-
dron, Brian D’Onofrio, and Irving I. Gottesman, “Socioeconomic Status Mod-
ifies Heritability of IQ in Young Children,” Psychological Science 14 (2003):
623–628; Nancy L. Segal and Wendy Johnson, “Twin Studies of General
Mental Ability,” in Handbook of Behavior Genetics, ed. Yong-Kyu Kim, 81–
99 (New York: Springer, 2009).
41. Hermine H. H. Maes, Gaston P. Beunen, Robert F. Vlietinck, et al. “Inheri-
tance of Physical Fitness in 10-Year-Old Twins and Their Parents,” Medicine
and Science in Sports and Exercise 28 (1996): 1479–1491.
42. Nicholas G. Martin, Lindon J. Eaves, Andrew C. Heath, Rosemary Jardine,
Lynn M. Feingold, and Hans J. Eysenck, “Transmission of Social Attitudes,”
Proceedings of the National Academy of Sciences of the United States of
America 83 (1986): 4364–4368, p. 4368.
43. David M. Buss, Evolutionary Psychology: The New Science of the Mind, 2nd
ed. (Boston: Allyn & Bacon, 2004).
44. Thomas H. Maugh, “Major Study Says Personality Mostly a Matter of
Genes,” Philadelphia Inquirer, October 12, 1990, 1A, 14A.
45. Richard M. Dudley, “IQ and Heredity,” letter to the editor, Science 252
(1991): 191; Jonathan Beckwith, Lisa Geller, and Sahotra Sarkar, “IQ and
Heredity,” letter to the editor, Science 252 (1991): 191.
 N OT E S TO PAG E S 1 1 3 – 1 1 7 357

46. Thomas J. Bouchard Jr., David T. Lykken, Matthew McGue, Nancy L. Segal,
and Auke Tellegen, “IQ and Heredity: Response,” Science 252 (1991): 191–
192. The phrase in the quote taken from this article was also emphasized in
the original text.
47. Ibid.
48. Thomas J. Bouchard Jr., David T. Lykken, Matt McGue, Nancy L. Segal, and
Auke Tellegen, “When Kin Correlations Are Not Squared,” Science 250
(1990): 1998.
49. Arthur R. Jensen, “Note on Why Genetic Correlations Are Not Squared,”
Psychological Bulletin 75 (1971): 223–224.
50. Marij Gielen, Catharina E.  M. van Beijsterveldt, Catherine Derom, et al.,
“Secular Trends in Gestational Age and Birthweight in Twins,” Human Repro-
duction 25 (2010): 2346–2353; Geoffrey A. Machin and Louis G. Keith, An
Atlas of Multiple Pregnancy: Biology and Pathology (New York: Parthenon,
1999).
51. Nancy L. Segal, Entwined Lives: Twins and What They Tell Us about Hu-
man Behavior (New York: Plume, 2000).
52. Ronald S. Wilson, “Twin Growth: Initial Deficit, Recovery and Trends in
Concordance from Birth to Nine Years,” Annals of Human Biology 6 (1979):
205–220.
53. Nancy L. Pedersen, Robert Plomin, John R. Nesselroade, and Gerald E. Mc-
Clearn, “A Quantitative Genetic Analysis of Cognitive Abilities during the
Second Half of the Life Span,” Psychological Sciences 3 (1992): 346–353.
54. Matt McGue, Thomas J. Bouchard Jr., William G. Iacono, and David T.
Lykken, “Behavioral Genetics of Cognitive Ability: A Life-Span Perspective,”
in Nature, Nurture, and Psychology, ed. Robert Plomin and Gerald E. Mc-
Clearn, 59–76 (Washington, D.C.: American Psychological Association, 1993).
55. “Sources of Human Psychological Differences,” tags: “Heredity Vs. Environ-
ment” and “Racial Differences,” David Duke (Web site), November 11, 2002,
www.davidduke.com/race-information-library/racial-differences/sources -of
-human-psychological-differences_66.html.
56. Hans H. Stassen, David T. Lykken, Peter Propping, and Gianni Bomben,
“Genetic Determination of the Human EEG: Survey of Recent Results on
Twins Reared Together and Apart,” Human Genetics 80 (1988): 165–176.

6. Job Satisfaction, Cardiac Characteristics, and More (1989–1990)

1. Richard D. Arvey, Thomas J. Bouchard Jr., Nancy L. Segal, and Lauren M.
Abraham, “Job Satisfaction: Environmental and Genetic Components,”
Journal of Applied Psychology 74 (1989): 187–192.
2. “The Biology of Business: Homo Administrans,” The Economist, September
23, 2010, www.economist.com/node/17090697.
3. Barry M. Staw, Nancy E. Bell, and John A. Clausen, “The Dispositional Ap-
proach to Job Attitudes: A Lifetime Longitudinal Test,” American Science
Quarterly 31 (1986): 56–77.
4. David J. Weiss, René V. Dawis, George W. England, and Lloyd H. Lofquist,
Manual for the Minnesota Satisfaction Questionnaire, Minnesota Studies in
358 N OT E S TO PAG E S 1 1 9 – 1 2 9

Vocational Rehabilitation No. 45 (Minneapolis: Industrial Relations Center,

University of Minnesota, 1967).
5. Russell Cropanzano and Keith James, “Some Methodological Consider-
ations for the Behavioral Genetic Analysis of Work Attitudes,” Journal of
Applied Psychology 75 (1990): 433–439.
6. Thomas J. Bouchard Jr., Richard D. Arvey, Lauren M. Keller, and Nancy L.
Segal, “Genetic Influences on Job Satisfaction: A Reply to Cropanzano and
James,” Journal of Applied Psychology 77 (1992): 89–93.
7. William G. Iaocono and David T. Lykken, “The Orienting Response: Impor-
tance of Instructions,” Schizophrenia Bulletin 5 (1979): 11–14.
8. Malcolm H. Lader and Lorna Wing, Physiological Measures, Sedative Drugs,
and Morbid Anxiety (London: Oxford University Press, 1966).
9. David T. Lykken, William G. Iacono, Keith Haroian, Matthew McGue, and
Thomas J. Bouchard Jr., “Habituation of the Skin Conductance Response
to Strong Stimuli: A Twin Study,” Psychophysiology 25 (1988): 4–15. Four
reared-apart same-sex sibling sets were included among the DZA twin
pairs.
10. David T. Lykken, “Autobiography” (unpublished manuscript, University of
Minnesota, 2006).
11. The electrodes were connected to separate Beckman Type 9844 skin conduc-
tance couplers that drove two channels of a Beckman RM Dynograph.
12. Velia Trejo, Catherine Derom, Robert Vlietinck, et al., “X Chromosome In-
activation Patterns Correlate with Fetal-Placental Anatomy in Monozygotic
Twin Pairs: Implications for Immune Relatedness and Concordance for Au-
toimmunity,” Molecular Medicine 1 (1994): 62–70.
13. Allen H. Kline, James B. Sidbury Jr., and Curt P. Richter, “The Occurrence of
Ectodermal Dysplasia and Corneal Dysplasia in One Family,” Journal of Pe-
diatrics 55 (1959): 355–366.
14. Bruce Hanson, Naip Tuna, Thomas J. Bouchard Jr., et al., “Genetic Factors in
the Electrocardiogram and Heart Rate of Twins Reared Apart and Together,”
American Journal of Cardiology 63 (1989): 606–609.
15. Hugh W. Simpson, Carmen Kelsey, Richard A. Gatti, et al., “Autorhythmom-
etry in Myasthenia Gravis: Detection of Chronopathology and Assessment
of Condition by Rhythm-Adjusted Level of Grip Strength,” Journal of Inter-
disciplinary Cycle Research 2 (1971): 397–416.
16. Hanson et al., “Genetic Factors in the Electrocardiogram.”
17. Peter Watson, Twins: An Uncanny Relationship (Chicago: Contemporary
Books, 1981).
18. Thomas J. Bouchard Jr., Nancy L. Segal, and David T. Lykken, “Genetic and
Environmental Influences on Special Mental Abilities in a Sample of Twins
Reared Apart,” Acta Geneticae Medicae et Gemellologiae, Twin Studies 39
(1990): 193–206.
19. Matt McGue and Thomas J. Bouchard Jr., “Genetic and Environmental De-
terminants of Information Processing and Special Mental Abilities: A Twin
Analysis,” in Advances in the Psychology of Human Intelligence, vol. 5, ed.
Robert J. Sternberg, 7–45 (Hillsdale, N.J.: Lawrence Erlbaum, 1989).
 N OT E S TO PAG E S 1 2 9 – 1 3 5 359

20. In 1983, psychologist Earl Hunt and others hypothesized that cognitive pro-
cessing measures reflect basic neurological mechanisms. Earl B. Hunt, “On
the Nature of Intelligence,” Science 219 (1983): 141–146.
21. Philip A. Vernon, “Speed of Information Processing and General Intelli-
gence,” Intelligence 7 (1983): 53–70.
22. Keith J. Hayes, “Genes, Drives, and Intellect,” Psychological Reports 10 (1962):
299–342.
23. Thomas J. Bouchard Jr., David T. Lykken, Auke Tellegen, and Matthew Mc-
Gue, “Genes, Drives, Environment and Experience: EPD Theory-Revised,”
in Intellectual Talent: Psychometric and Social Issues, ed. Camilla P. Ben-
bow and David Lubinski, 5–43 (Baltimore: Johns Hopkins University Press,
1996).
24. Ibid., p. 31.
25. John C. DeFries, Geoffrey C. Ashton, Ronald C. Johnseon, et al., “Parent-
Offspring Resemblance for Specific Cognitive Abilities in Two Ethnic Groups,”
Nature 261 (1986): 131–133.
26. McGue and Bouchard, “Genetic and Environmental Determinants,
pp. 41–42.
27. Matt McGue, Thomas J. Bouchard Jr., David T. Lykken, and Dale Feuer,
“Information Processing Abilities in Twins Reared Apart,” Intelligence 8
(1984): 239–258; Thomas J. Bouchard Jr., David T. Lykken, Nancy L. Segal,
and Kimerly J. Wilcox, “Development in Twins Reared Apart: A Test of the
Chronogenetic Hypothesis,” Human Growth: A Multidisciplinary Review,
ed. Andrea Demirjian, 299–310 (London: Taylor & Francis, 1986).
28. Nancy L. Segal, “Twins: The Finest Natural Experiment,” Personality and
Individual Differences 49 (2010): 317–323.
29. Constance Holden, “Offbeat Twins,” Science 288 (2000): 1735; Dateline
NBC, “The Ties That Bind,” March 30, 1999.
30. Thomas J. Bouchard Jr. and Kimerly J. Wilcox, “Behavior Genetics,” in
McGraw-Hill Encyclopedia of Science and Technology, 6th ed. (New York:
McGraw-Hill, 1989).
31. Thomas J. Bouchard Jr. and Nancy L. Segal, “Advanced Mathematical Rea-
soning Ability: A Behavioral Genetic Perspective,” Behavioral and Brain Sci-
ences 13 (1990): 191–192.
32. Paul M. Brinich et al., The Adoption Bibliography (Washington, D.C.: Amer-
ican Adoption Congress, 1990).
33. Le Journal International de Médecine 178 (1990): 27–36.
34. David T. Lykken, Thomas J. Bouchard Jr., Matt McGue, and Auke Tellegen,
“The Minnesota Twin Family Registry: Some Initial Findings,” Acta Gene-
ticae Medicae et Gemellologiae, Twin Studies, 39 (1990): 35–70.
35. Allison Kelly, “Wedding Gift of a Long Lost Brother,” unlabeled newspaper
article from Bouchard’s archive, September 1989.
36. Elizabeth Mullener, “Nature or Nurture? Ask Twins,” The Times-Picayune
[New Orleans], February 16, 1990, A1–A8.
37. Melvin Konner, “Under the Influence,” Omni magazine, January 1989, 62–64,
90–91.
360 N OT E S TO PAG E S 1 3 6 – 1 39

7. Psychopathology and Religiosity (1990)

1. Gilles de la Tourette, “Étude sur une affection nerveuse characterisée par de
l’incoordination motrice accompagnée d’écholalie et de coprolalie,” Archives
de Neurologie (Paris) 9 (1884): 19–42, 158–200.
2. See Niels Juel-Nielsen, Individual and Environment: Monozygotic Twins
Reared Apart (New York: International Universities Press, 1980); Horatio H.
Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A Study of Hered-
ity and Environment (Chicago: University of Chicago Press, 1937).
3. R. Arlen Price, Kenneth K. Kidd, Donald J. Cohen, David L. Pauls, and James
F. Leckman, “A Twin Study of Tourette Syndrome,” Archives of General Psy-
chiatry 42 (1985): 815–820.
4. Peter Vieregge, Cornelia Schäfer, and Johannes Jörg, “Concordant Gilles de
la Tourette’s Syndrome in Monozygotic Twins: A Clinical, Neurophysiologi-
cal and CT Study,” Journal of Neurology 235 (1988): 366–367; Ann Le Cou-
teur, Anthony Bailey, Susan Goode, et al., “A Broader Phenotype of Autism:
The Clinical Spectrum in Twins,” Journal of Child Psychology and Psychiatry
37 (1996): 785–801.
5. Jesse F. Abelson, Kenneth Y. Kwan, Brian J. O’Roak, et al., “Sequence Vari-
ants in SLITRK1 Are Associated with Tourette’s Syndrome,” Science 310
(2005): 317–320.
6. Revised versions of the Diagnostic Statistical Manual (DSM; e.g., DSM-IV)
call the condition Tourette disorder, while earlier versions and other sources
call it Tourette syndrome.
7. Nancy L. Segal, Maurice W. Dysken, Thomas J. Bouchard Jr., Nancy L. Peder-
sen, and Leonard L. Heston, “Tourette’s Disorder in a Set of Reared-Apart
Triplets: Genetic and Environmental Influences,” American Journal of Psychia-
try 147 (1990): 196–199; Darlene C. Ifill-Taylor, “Diagnosis of Tourette’s Dis-
order,” letter to the editor, American Journal of Psychiatry 147 (1990): 1386.
8. David Rosenthal, The Genain Quadruplets: A Case Study and Theoretical
Analysis of Heredity and Environment in Schizophrenia (New York: Basic
Books, 1963).
9. A number of twins shared their life stories with the media and, with their
consent, I published detailed essays on three of our separated sets in my
book Indivisible by Two: Lives of Extraordinary Twins (Cambridge, Mass.:
Harvard University Press, 2007).
10. William M. Grove, Elke D. Eckert, Leonard Heston, Thomas J. Bouchard Jr.,
Nancy L. Segal, and David T. Lykken, “Heritability of Substance Abuse and
Antisocial Behavior: A Study of Monozygotic Twins Reared Apart,” Biologi-
cal Psychiatry 27 (1990): 1293–1304.
11. Irving I. Gottesman and James Shields, Schizophrenia: The Epigenetic Puzzle
(Cambridge, U.K.: Cambridge University Press, 1982).
12. Susan L. Farber, Identical Twins Reared Apart: A Reanalysis (New York:
Basic Books, 1980).
13. The Diagnostic Statistical Manual-III (1980) introduced a multiaxial system
to recognize different aspects of behavioral disorders such as clinical and
personality.
 N OT E S TO PAG E S 1 39 – 1 42 361

14. The Research Diagnostic Criteria (RDC) and St. Louis Criteria specify symp-
toms of psychiatric disorders. The RDC criteria were developed in the 1970s
to increase diagnostic consistency of American and European patients.
15. Probandwise concordance is calculated as the number of affected individuals
in concordant pairs divided by the total number of affected individuals. Ge-
neticists prefer this method to pairwise concordance, or number of concor-
dant pairs divided by the number of concordant pairs plus discordant pairs.
The probandwise rate is preferred for several reasons, such as enabling direct
comparison with the population prevalence and rates for other relatives.
Matt McGue, “When Assessing Twin Concordance, Use the Probandwise,
Not the Pairwise Rate,” Schizophrenia Bulletin 18 (1992): 171–176.
16. Elke D. Eckert, Thomas J. Bouchard Jr., Joseph Bohlen, and Leonard L. Heston,
“Homosexuality in Twins Reared Apart,” British Journal of Psychiatry 148
(1986): 421–425.
17. Nancy L. Segal, Elke D. Eckert, William M. Grove, Thomas J. Bouchard Jr.,
and Leonard L. Heston, “A Summary of Psychiatric and Psychological Find-
ings from the Minnesota Study of Twins Reared Apart,” Etiology of Mental
Disorder, ed. Einar Kringlen, Nils Johan Lavik, and Svenn Torgersen (World
Psychiatric Association, Regional Symposium, August 23–26, Oslo, Norway),
183–200 (Vindern, Norway: University of Oslo, Department of Psychiatry,
1990).
18. Robert Plomin, John C. DeFries, Gerald E. McClearn, and Peter McGuffin,
Behavioral Genetics, 5th ed. (New York: Worth Publishers, 2008).
19. Robert Plomin, John C. DeFries, and Gerald E. McClearn, Behavioral Gene-
tics: A Primer (San Francisco: W. H. Freeman, 1980).
20. Robert Plomin, John C. DeFries, and Gerald E. McClearn, Behavioral Gene-
tics: A Primer, 2nd ed. (New York: W. H. Freeman, 1990), pp. 386–387. The
work referenced in this passage is John C. Loehlin and Robert C. Nichols,
Heredity, Environment, and Personality: A Study of 850 Sets of Twins (Aus-
tin: University of Texas Press, 1976).
21. Niels G. Waller, Brian A. Kojetin, Thomas J. Bouchard Jr., David T. Lykken,
and Auke Tellegen, “Genetic and Environmental Influences on Religious In-
terests, Attitudes, and Values: A Study of Twins Reared Apart and Together,”
Psychological Science 1 (1990): 138–142.
22. Laura B. Koenig, Matt McGue, Robert F. Krueger, and Thomas J. Bouchard
Jr., “Genetic and Environmental Influences on Religiousness: Findings for
Retrospective and Current Religiousness Ratings,” Journal of Personality 73
(2005): 471–488, p. 251. The article in the special issue refers to Lyndon I.
Eaves, Brian D’Onofrio, and Robert Russell, “Transmission of Religion and
Attitudes,” Twin Research 2 (1999): 59–61.
23. Nicholas G. Martin, Lindon J. Eaves, Andrew C. Heath, Rosemary Jardine,
Lynn M. Feingold, and Hans J. Eysenck, “Transmission of Social Attitudes,”
Proceedings of the National Academy of Sciences of the United States of
America 83 (1986): 4364–4368.
24. Waller et al., “Genetic and Environmental Influences.”
25. Jerry S. Wiggins, “Substantive Dimensions of Self-Report in the MMPI Pool,”
Psychological Monographs 80, no. 630 (1966): 1–42.
362 N OT E S TO PAG E S 1 42 – 1 4 8

26. Niels G. Waller, David T. Lykken, and Auke Tellegen, “Occupational Inter-
ests, Leisure Time Interests, and Personality: Three Domains or One? Find-
ings from the Minnesota Twin Registry,” in Assessing Individual Differences
in Human Behavior: New Concepts, Methods, and Findings, ed. David Lu-
binski and René V. Dawis, chapter 9 (Palo Alto, Calif.: Davies-Black, 1995).
The scale is derived from factor analysis.
27. Jo-Ida C. Hansen and Donald P. Campbell, Manual for the SVIB-SCII (Stan-
ford, Calif.: Stanford University Press, 1985).
28. Waller et al., “Occupational Interests.” Waller chose a different thesis topic—
“Genetic Tobit Factor Analysis: Quantitative Genetic Modeling with Cen-
sored Data”—and was hired by the University of California–Davis.
29. Gordon W. Allport, Philip E. Vernon, and Gardner Lindzey, Manual for the
Study of Values: A Scale for Measuring the Dominant Interests in Personal-
ity, 3rd ed. (Boston: Houghton-Mifflin, 1960).
30. The estimated shared environmental component for the Religious Leisure
Time Interests scale of 0.11 did not differ significantly from zero.
31. Only two of the fifteen correlations were statistically significant, but they
were modest in size. Wiggins—Performance IQ: r = −0.26; Wiggins—Full Scale
IQ: r = −0.21.
32. Jeffrey Kluger, Jeff Chu, Boward Liston, Maggie Sieger, and Daniel Williams,
“Is God in Our Genes?” Time, October 25, 2004, p. 67.
33. Ellen Robinson-Hayes, “Beliefs Flow Out of Genes?” Sacramento Bee, Feb-
ruary 17, 1990, A-1.
34. Julie Begley, “An Impossible Christmas Surprise,” The Voice 39, December
24, 1997.
35. The twins’ November 1997 reunion was taped by local Channel 3 News.
36. Dorine Leogrande, “A Tale of Two Sisters,” Connecticut Jewish Ledger, De-
cember 12, 1997, 1, 27. At the end of one of our evenings together, the twins
laughed and crossed their eyes—they had independently discovered this trick
and enjoyed entertaining (or horrifying) people with it.
37. Ibid.
38. Mary E. Holmes, Being You (London: Austin & Macauley, 2008), 156–157.
39. Segal, Indivisible by Two.
40. Thomas J. Bouchard Jr., David T. Lykken, Matt McGue, and Auke Tellegen,
“Intrinsic and Extrinsic Religiousness: Genetic and Environmental Influ-
ences and Personality Correlates,” Twin Research 2 (1999): 88–98.
41. Robert Plomin and Denise Daniels, “Why Are Children in the Same Family
So Different from Each Other?” Behavioral and Brain Sciences 10 (1987):
1–16.
42. “Theodore Wachs,” Purdue University Psychological Sciences, Develop-
mental Faculty, n.d., www.psych.purdue.edu/index.php/faculty/12-faculty/
39-wachs-theodore-.html.
43. Thomas J. Bouchard Jr. and Matthew McGue, “Genetic and Rearing Envi-
ronmental Influences on Adult Personality: An Analysis of Adopted Twins
Reared Apart,” Journal of Personality 58 (1990): 263–292.
44. Harrison G. Gough, “An Interpreter’s Syllabus for the California Psychologi-
cal Inventory,” in Advances in Psychological Assessment, vol. 1, ed. Paul
 N OT E S TO PAG E S 1 4 8 – 1 57 363

McReynolds, 55–79 (Palo Alto, Calif.: Consulting Psychologists Press,

1968), p. 57.
45. Rudolph H. Moos and Bernice S. Moos, Manual: Family Environment Scale
(Palo Alto, Calif.: Consulting Psychologists Press, 1986).
46. When scores do not differ greatly across individuals, it is not possible to find
meaningful relationships between these scores and other measures. If most
adoptees were adopted on their day of birth, it would be impossible to detect
a relationship between age at adoption and birth weight, which varies con-
siderably across newborns.
47. Note that attenuation is expected for correlations based on difference scores.
48. Constance Holden, “Double Features,” The 1990 World Book Year Book:
The Annual Supplement to the World Book Encyclopedia (Chicago: World
Book, 1990), 140–153.
49. “Zweites Ich,” Der Spiegel, April 1990, 248–250.
50. Thomas H. Maugh II, “Major Study Says Personality Mostly a Matter of
Genes,” Philadelphia Inquirer, October 12, 1990, 1A, 14A.

8. Dental Traits, Allergies, and Vocational Interests (1991–1992)

Epigraph: JoAnn C. Boraas, Louise B. Messer, and Michael J. Till, “A Genetic
Contribution to Dental Caries, Occlusion, and Morphology as Demonstrated
by Twins Reared Apart,” Journal of Dental Research 67 (1988): 1150–1155,
p. 1150.
1. Sidney B. Finn and Robert C. Caldwell, “Dental Caries in Twins I: A Com-
parison of the Caries Experience of Monozygotic Twins, Dizygotic Twins
and Unrelated Children,” Archives of Oral Biology 8 (1963): 571–585; Rob-
ert S. Corrucini and Rosario H. Potter, “Genetic Analysis of Occlusional
Variation in Twins,” American Journal of Orthodontics 78 (1980): 140–154.
2. Susan L. Farber, Identical Twins Reared Apart: A Reanalysis (New York:
Basic Books, 1980).
3. Malocclusion refers to a difference in size between the upper and lower jaws
or jaw and tooth size, leading to overcrowding or an abnormal bite. Maloc-
clusion is most often hereditary. Jack D. Rosenberg, “Malocclusion of Teeth,”
MedlinePlus Medical Encyclopedia, updated February 22, 2010, www.nlm
.nih.gov/medlineplus/ency/article/001058.htm.
4. Overbite occurs when the upper teeth are slightly longer than normal; overjet
occurs when the upper teeth protrude visibly (buck teeth). “Difference between
Overbite and Overjet,” Difference Between, updated January 23, 2011, www
.differencebetween.net/science/health/difference-between-overbite-and-overjet/.
5. Douglas  S. Falconer and Trudy F. C. Mackay, Introduction to Quantitative
Genetics, 4th ed. (New York: Longman, 1996).
6. John P. Conry, Louise B. Messer, Joann C. Boraas, Dorothy P. Aeppli, and
Thomas J. Bouchard Jr., “Dental Caries and Treatment Characteristics in Hu-
man Twins Reared Apart,” Archives of Oral Biology 38 (1993): 937–943.
7. Lisa Adams, “Periodontal Disease—Gingivitis and Periodontitis,” Suite101,
updated September 19, 2010, www.suite101.com/content/peridontal-disease
—gingivitis-and-periodontitis-a287303.
364 N OT E S TO PAG E S 1 57– 1 6 3

8. The National Institute of Dental Research is now the National Institute of

Dental and Craniofacial Research.
9. Hans-Erich Reiser and Friedrich Vogel, “Uber die Erblichkeit der Zahnsteinbil-
dung Beim Menschen,” Deutsche Zahuntz Zeitschrift 13 (1958): 1355–1358.
10. Sebastian G. Ciancio, Stanley P. Hazen, and John J. Cunat, “Periodontal Ob-
servations in Twins,” Journal of Periodontal Research 4 (1969): 42–45.
11. Bryan S. Michalowicz, Dorothy Aeppli, John G. Virag, et al., “Periodontal
Findings in Adult Twins,” Journal of Periodontology 62 (1991): 293–299.
12. Information was obtained from the mesiobuccal or mesiolingual tooth surface.
13. Ginny G. Lane, “The Beginner’s Guide to the Bootstrap Method of Resam-
pling,” presented at the Annual Meeting of the Southwest Educational Re-
search Association, Dallas, Texas, January 27–29, 2000.
14. Bryan S. Michalowicz, Scott R. Diehl, John C. Gunsolley, et al., “Evidence of
a Substantial Genetic Component for Risk of Adult Periodontitis,” Journal
of Periodontology 71 (2000): 1699–1707.
15. Bryan S. Michalowicz, Dorothy P. Aeppli, Ramesh K. Kuba, et al., “A Twin
Study of Genetic Variation in Proportional Radiographic Alveolar Bone
Height,” Journal of Dental Research 70 (1991): 1431–1435.
16. The device is called a cephalostat head positioner. Testers used seven ana-
tomical measures to position co-twins similarly within the apparatus.
17. Minoru Nakata, Pao-Lo Yu, Bailey Davis, and W. E. Nance, “Genetic Deter-
minants of Craino-Facial Morphology: A Twin Study,” Annals of Human
Genetics 37 (1974): 431–443.
18. Kenneth S. Kornman, Allison Crane, Hwa-Ying Wang, et al., “The Interleu-
kin-1 Genotype as a Severity Factor in Adult Periodontal Disease,” Journal
of Clinical Periodontology 24 (1997): 72–77.
19. Bryan M. Michalowicz and Bruce L. Pihlstrom. “Genetic Factors Associated
with Periodontal Disease,” in Carranza’s Clinical Periodontology, 10th ed.,
ed. Michael G Newman et al., 193–208 (Philadelphia: W. B. Saunders, 2006).
This chapter first appeared in the 2002 edition.
20. Gary C. Armitage and Paul B. Robertson, “The Biology, Prevention, Diagno-
sis and Treatment of Periodontal Diseases: Scientific Advances in the United
States,” Journal of the American Dental Association 140 (2009): 36S–43S.
21. “Allergies,” The Free Dictionary, n.d., http://medical-dictionary.thefreedic-
tionary.com/allergies.
22. Susan L. Farber. Identical Twins Reared Apart: A Reanalysis (New York;
Basic Books, 1981). The symptoms listed here were recorded for twins in
studies conducted between 1937 and 1980. Physicians currently do not
believe such symptoms necessarily represent allergic reactions. For exam-
ple, allergy is only one of several causes of asthma. Malcolm Blumenthal,
personal communication, February 20, 2011.
23. Ibid.
24. Malcolm N. Blumenthal and D. Bernard Amos, “Genetic and Immunologic
Basis of Atopic Responses,” Chest 9 (1987): 176S–184S.
25. Peter F. Kohler, Victor J. Rivera, Elke D. Eckert, Thomas J. Bouchard Jr., and
Leonard L. Heston, “Genetic Regulation of Immunoglobulin and Specific
 N OT E S TO PAG E S 1 6 3 – 1 7 1 365

Antibody Levels in Twins Reared Apart,” Journal of Clinical Investigation 75

(1985): 883–888.
26. Bruce Hanson, Matthew McGue, Beatrice Roitman-Johnson, Nancy L. Segal,
Thomas J. Bouchard Jr., and Malcolm N. Blumenthal, “Atopic Disease and
Immunoglobulin E in Twins Reared Apart and Together,” American Journal
of Human Genetics 48 (1991): 873–879.
27. John B. Davis, “Asthma and Wheezy Bronchitis in Children,” Clinical Allergy
6 (1976): 329–338; Jeanne M. Smith, “Epidemiology and Natural History of
Asthma, Allergic Rhinitis, and Atopic Dermatitis (Eczema),” in Allergy: Prin-
ciples and Practice, 3rd ed., ed. Elliott Middleton Jr., Charles F. Reed, Elliot
F. Ellis, N. Franklin Adkinson Jr., and Jon W. Yunginger, 891–929 (St. Louis,
Mo.: Mosby, 1988).
28. Malcolm N. Blumenthal and Sergio Bonini, “Immunogenetics of Specific Im-
mune Responses to Allergens in Twins and Families,” in Genetic and Envi-
ronmental Factors in Clinical Allergy, ed. David G. March and Malcolm N.
Blumenthal, 132–142 (Minneapolis: University of Minnesota Press, 1990).
29. Correlations can be calculated in the absence of a normal distribution, but
should be interpreted cautiously. In the study, high values precluded normal-
ization even after log transformation.
30. Daniel P. Moloney, Thomas J. Bouchard Jr., and Nancy L. Segal, “A Genetic and
Environmental Analysis of the Vocational Interests of Monozygotic and Dizy-
gotic Twins Reared Apart,” Journal of Vocational Behavior 39 (1991): 76–109.
31. See, for example, John C. Loehlin and Robert C. Nichols, Heredity, Environ-
ment, and Personality: A Study of 850 Sets of Twins (Austin: University of
Texas Press, 1976).
32. Jo-Ida C. Hansen and Donald P. Campbell, Manual for the SVIB-SCII (Stan-
ford, Calif.: Stanford University Press, 1985); Donald J. Jackson, Jackson
Vocational Interest Survey Manual (Port Huron, Mich.: Research Psycholo-
gists Press, 1977).
33. Thomas J. Bouchard Jr. and Matthew McGue, “Genetic and Rearing Environ-
mental Influences on Adult Personality: An Analysis of Adopted Twins Reared
Apart,” Journal of Personality 58 (1990): 263–292.
34. The correlations for Achievement Orientation were: MZA (ri = 0.73) and
DZA (ri = 0.19). The general model assumes an absence of dominance.
35. Xi-zhi Wu and Mao-zai Tian, “A Longitudinal Study of the Effects of Family
Background Factors on Mathematics Achievements Using Quantile Regres-
sion,” Acta Mathematicae Applicatae Sinica 24 (2000): 85–98.
36. Eric Turkheimer, “Three Laws of Behavior Genetics and What They Mean,”
Current Directions in Psychological Science 9 (2000): 160–164.

9. Creativity, Work Values, and Evolution (1992–1993)

1. David T. Lykken, “Trial by Polygraph,” Behavioral Sciences and the Law 2
(1984): 75–92.
2. David T. Lykken, “The Case for Parental Licensure,” in Psychopathology: An-
tisocial, Criminal, and Violent Behavior, ed. Theodore Milton, Erik Simonsen,
366 N OT E S TO PAG E S 1 7 1 – 1 7 5

Morten Birket-Smith, and Roger D. Davis, 122–143 (New York: Guilford

Press, 2002).
3. David T. Lykken, Happiness: What Studies of Twins Tell Us about Nature,
Nurture, and the Happiness Set Point (New York: Golden Books, 1999).
4. David T. Lykken et al., “Heritability Estimates from Twin Studies: The Efficiency
of the MZA Design” (unpublished manuscript, University of Minnesota).
5. David T. Lykken, Matthew McGue, Auke Tellegen, and Thomas J. Bouchard
Jr., “Emergenesis: Genetic Traits That May Not Run in Families,” American
Psychologist 47 (1992): 1565–1577.
6. Polymorphic genes are those having multiple allelic forms.
7. Lykken et al., “Emergenesis” (1992), 1566.
8. Twin studies might miss emergenic traits when the underlying factors mimic
additive transmission and traits appear familial.
9. Lykken et al., “Emergenesis.”
10. Ibid., 1571.
11. Ibid., 1572.
12. Robert Plomin, John C. DeFries, and Gerald E. McClearn, Behavioral Gene-
tics: A Primer, 4th ed. (New York: Worth Publishers, 2001), 172.
13. Carrie M. Nielson, “The Power of Epistasis,” Science Translational Medicine
3 (2011): 79.
14. Niels G. Waller, Thomas J. Bouchard Jr., David T. Lykken, Auke Tellegen, and
Dawn M. Blacker, “Creativity, Heritability, Familiality: Which Word Does
Not Belong?” Psychological Inquiry 4 (1993): 235–237.
15. Hans J. Eysenck, “Creativity and Personality: Suggestions for a Theory,”
Psychological Inquiry 4 (1993): 147–178.
16. Larry Johnson and Annette Lamb, “Critical and Creative Thinking—Bloom’s
Taxonomy,” Teacher Tap, n.d., http://eduscapes.com/tap/topic69.htm.
17. Mihalyi Csikszentmihalyi, Creativity: Flow and the Psychology of Discovery
and Invention (New York: HarperPerennial, 1997).
18. The dimensionality of the thirty items was assessed by nonlinear factor analy-
sis, a procedure that assesses relations and interactions among groups of re-
lated items or factors. The Creative Personality Scale was treated as a unidi-
mensional construct because scale scores based on the unit-weighted sum of
responses reflect general factor variance; Waller et al., “Creativity, Heritabil-
ity, Familiality.”
19. David T. Lykken, Thomas J. Bouchard Jr., Matt McGue, and Auke Tellegen,
“The Heritability of Interests: A Twin Study,” Journal of Applied Psychology
78 (1993): 649–661.
20. Itamar Gati, “The Structure of Vocational Interests,” Psychological Bulletin
109 (1991): 309–324.
21. Daniel P. Moloney, Thomas J. Bouchard Jr., and Nancy L. Segal, “A Genetic and
Environmental Analysis of the Vocational Interests of Monozygotic and Dizy-
gotic Twins Reared Apart,” Journal of Vocational Behavior 39 (1991): 76–109.
22. Lykken et al., “Heritability of Interests.”
23. David T. Lykken, Matt McGue, and Auke Tellegen, “Recruitment Bias in
Twin Research: The Rule of Two-Thirds Reconsidered,” Behavior Genetics
17 (1987): 343–362.
 N OT E S TO PAG E S 1 7 5 – 1 8 4 367

24. Ibid.
25. Inventory items were scaled to remove individual differences in the twins’
use of item-response categories (see the appendix to Lykken et al., “Herita-
bility of Interests.” For example, some twins might have consistently chosen
1s or 2s in the Minnesota Leisure-Time Interest Test.).
26. The mean heritability was 0.32 for items, but it increased to 0.48 for factors
and to 0.53 for superfactors.
27. Lauren M. Keller, Thomas J. Bouchard Jr., Richard D. Arvey, Nancy L. Segal,
and René V. Dawis, “Work Values: Genetic and Environmental Influences,”
Journal of Applied Psychology 77 (1992): 79–88.
28. Evan G. Gay, David J. Weiss, Darwin D. Hendel, René V. Dawis, and Lloyd
H. Lofquist, Manual for the Minnesota Importance Questionnaire, Minne-
sota Studies in Vocational Rehabilitation 28 (Minneapolis: Work Adjust-
ment Project, Industrial Relations Center, University of Minnesota, 1971).
29. Matt McGue and Thomas J. Bouchard Jr., “Genetic and Environmental De-
terminants of Information Processing and Special Mental Abilities: A Twin
Analysis,” in Advances in the Psychology of Human Intelligence, vol. 5, ed.
Robert J. Sternberg, 7–45 (Hillsdale, N.J.: Lawrence Erlbaum, 1989).
30. J. Philippe Rushton, David W. Fulker, Michael C. Neale, David K. B. Nias, and
Hans J. Eysenck, “Altruism and Aggression: The Heritability of Individual Dif-
ferences,” Journal of Personality and Social Psychology 6 (1986): 1192–1198.
31. Nancy L. Segal, Indivisible by Two: Lives of Extraordinary Twins (Cam-
bridge, Mass.: Harvard University Press, 2007).
32. Linda Mealey, “Kinship: The Ties That Bind (Disciplines),” in Conceptual
Challenges in Evolutionary Psychology: Innovative Research Strategies, ed.
Harmon R. Holcomb III, 19–38 (Dordrecht, the Netherlands: Kluwer, 2001).
33. Nancy L. Segal, “Twin, Adoption and Family Methods as Approaches to
Evolution of Individual Differences,” in The Evolution of Personality and
Individual Differences, ed. David M. Buss, 303–337 (Oxford: Oxford Uni-
versity Press, 2011).
34. Steven G. Gangestad, “Evolutionary Biology Looks at Behavior Genetics,”
Personality and Individual Differences 49 (2010): 289–295.
35. Thomas J. Bouchard Jr. and John C. Loehlin, “Genes, Evolution, and Person-
ality,” Behavior Genetics 31 (2001): 243–273.
36. Daniel G. Freedman, Human Sociobiology: A Holistic Approach (New York:
Free Press, 1979).
37. Edward O. Wilson, Sociobiology: The New Synthesis (Cambridge, Mass.:
Harvard University Press, 1975).
38. William D. Hamilton, “The Genetical Evolution of Human Behaviour,” Jour-
nal of Theoretical Biology 7 (1964): 1–52.
39. Nancy L. Segal, “Twin, Sibling and Adoption Methods: Tests of Evolution-
ary Hypotheses,” American Psychologist 48 (1993): 943–956.
40. Charles B. Crawford and Judith L. Anderson, “Sociobiology: An Environ-
mentalist Discipline?” American Psychologist 44 (1989): 1449–1459.
41. Bruce L. Ellis, “Timing of Pubertal Maturation in Girls: An Integrated Life
History Approach,” Psychological Bulletin 130 (2004): 920–954.
42. Mealey, “Kinship,” 23.
368 N OT E S TO PAG E S 1 8 4 – 1 9 0

43. Linda Mealey and Nancy L. Segal, “Heritable and Environmental Variables
Affect Reproduction-Related Behaviors, but Not Ultimate Reproductive Suc-
cess,” Personality and Individual Differences 14 (1993): 783–794.
44. Peter F. Briggs, M-B History Record: Self-Administered Form (Brandon, Vt.:
Clinical Psychology, [1969]). M-B stands for Minnesota-Briggs.
45. Joseph L. Rodgers, Hans-Peter Kohler, Kirsten Ohm Kyvik, and Kaare Chris-
tensen, “Behavior Genetic Modeling of Human Fertility: Findings from a
Contemporary Danish Study,” Demography 38 (2001): 29–42.
46. Lykken et al., “Emergenesis.”
47. Thomas J. Bouchard Jr., David T. Lykken, Matthew McGue, Nancy L. Segal,
and Auke Tellegen, “Sources of Human Psychological Differences: The Min-
nesota Study of Twins Reared Apart,” Science 250 (1990): 223–228.
48. Thomas J. Bouchard Jr. and Peter Propping, eds., Twins as a Tool of Behav-
ior Genetics (Chichester, U.K.: Wiley & Sons, 1993).
49. Thomas J. Bouchard Jr., “Genetic and Environmental Influences on Adult
Personality: Evaluating the Evidence,” in Foundations of Personality, ed.
P. Joop Hettema and Ian J. Deary, 15–44 (Dordrecht: Kluwer, 1993).
50. John C. Loehlin, Genes and Environment in Personality Development (New-
bury Park, N.J.: Sage, 1992).
51. David C. Rowe, The Limits of Family Influence: Genes, Experience, and
Behavior (New York: Guilford Press, 1994), 44.
52. Rainer Riemann, Alois Angleitner, and Jan Strelau, “Genetic and Environ-
mental Influences on Personality: A Study of Twins Reared Together Using
the Self- and Peer Report NEO-FFI Scales,” Journal of Personality 65 (1997):
449–475. The correlation between the twins’ and peers’ reports was .55; the
correlation between the two peers who rated each twin was .61.
53. Robert Plomin and Gerald E. McClearn, eds., Nature, Nurture, and Psychol-
ogy (Washington, D.C.: American Psychological Association, 1993).
54. Robert Plomin, Michael J. Owen, and Peter McGuffin, “The Genetic Basis of
Complex Human Behavior,” Science 264 (1994): 1733–1739.
55. Matt McGue, Thomas J. Bouchard Jr., William G. Iacono, and David T.
Lykken, “Behavioral Genetics of Cognitive Ability: A Life-Span Perspective,”
in Nature, Nurture, and Psychology, ed. Robert Plomin and Gerald E. Mc-
Clearn, chapter 3; see p. xiii. Interestingly, just two years earlier, in 1991, the
APA Monitor published an article concerning the controversial reception of
the MISTRA’s IQ findings.
56. The original data from Newman et al. and Juel-Nielsen were reanalyzed to
derive intraclass correlations by our procedures.
57. Nancy L. Segal, “Twins: The Finest Natural Experiment,” Personality and
Individual Differences 49 (2010): 317–323.
58. Nancy L. Segal, Shirley A. McGuire, June Havlena, Patricia Gill, and Scott L.
Hershberger, “Intellectual Similarity of Virtual Twin Pairs: Developmental
Trends,” Personality and Individual Differences 42 (2007): 1209–1219.
59. John C. Loehlin, Joseph M. Horn, and Lee Willerman, “Modeling IQ
Change: Evidence from the Texas Adoption Project,” Child Development 60
(1989): 993–1004.
60. Plomin et al., Behavioral Genetics.
 N OT E S TO PAG E S 1 9 0 – 1 9 9 369

61. Thomas J. Bouchard Jr., “The Genetic Architecture of Human Intelligence,”

in Biological Approaches to the Study of Human Intelligence, ed. Philip E.
Vernon, 33–93 (New York: Ablex, 1993).
62. Francis Galton, “The History of Twins as a Criterion of the Relative Powers
of Nature and Nurture,” Journal of the Anthropological Institute 5 (1885):
391–406.
63. Richard L. Doty, Paul Shaman, and Michael Dann, “Development of the
University of Pennsylvania Smell Identification Test: A Standardized Micro-
encapsulated Test of Olfactory Function,” Physiology and Behavior 32 (1984):
489–502.
64. Kathryn M. Abbe and Frances M. Gill, Twins on Twins (New York: Clark-
son N. Potter, 1980).

10. Family Environments, Happiness, Sensation Seeking,

and the MMPI (1994–1997)
1. Yoon-Mi Hur and Thomas J. Bouchard Jr., “Genetic Influences on Percep-
tions of Childhood Family Environment: A Reared Apart Twin Study,” Child
Development 66 (1995): 330–345.
2. David C. Rowe, “Environmental and Genetic Influence on Dimensions of
Perceived Parenting: A Twin Study,” Developmental Psychology 17 (1981):
203–208; “A Biometrical Analysis of Perceptions of Family Environment:
A Twin Study of Twin and Singleton Sibling Kinships,” Child Development
54 (1983): 416–423.
3. Robert Plomin, Gerald E. McClearn, Nancy L. Pedersen, John R. Nesselroade,
and Cindy S. Bergeman, “Genetic Influence on Adults’ Ratings of Their Current
Family Environment,” Journal of Marriage and Family 52 (1989): 791–803.
4. Jean Block, Lives through Time (Berkeley, Calif.: Bancroft, 1971); Rudolf H.
Moos and Bernice S. Moos, Manual: Family Environment Scale (Palo Alto,
Calif.: Consulting Psychologists Press, 1986).
5. David T. Lykken, Happiness: What Studies of Twins Tell Us about Nature,
Nurture, and the Happiness Set Point (New York: Golden Books, 1999).
6. David G. Myers and Edward Diener, “Who Is Happy?” Psychological Sci-
ence 6 (1995): 10–19.
7. David T. Lykken and Auke Tellegen, “Happiness Is a Stochastic Phenome-
non,” Psychological Science 7 (1996): 186–189.
8. In an effort to find evidence compatible with the idea that natural selection
favored happy people, one could compare the life history and reproductive
outcomes of people who vary along given happiness dimensions.
9. Jerome H. Barkow, “Happiness in Evolutionary Perspective,” in Uniting Psy-
chology and Biology: Integrative Perspectives on Human Development, ed.
Nancy L. Segal et al., 397–418 (Washington, D.C.: American Psychological
Association, 1997).
10. Cecilia Tomassini, Knud Juel, Niels V. Holm, Axel Skytthe, and Kaare Chris-
tensen, “Risk of Suicide in Twins: 51 Year Follow Up Study,” British Medical
Journal 327 (2003): 373–374. Twins in this study were not organized by
zygosity.
370 N OT E S TO PAG E S 1 9 9 – 2 0 2

11. Alexander Weiss, Timothy C. Bates, and Michelle Luciano, “Happiness Is

a  Personal(ity) Thing: The Genetics of Personality and Well-Being in a
Representative Sample,” Psychological Science 19 (2008): 205–210; Richard
E. Lucas, A. E. Clark, Y. Georgellis, and E. Diener, “Reexamining Adaptation
and the Set Point Model of Happiness: Reactions to Changes in Marital Sta-
tus,” Journal of Personality and Social Psychology 84 (2003): 527–539.
12. Auke Tellegen, David T. Lykken, Thomas J. Bouchard Jr., Kimerly J. Wilcox,
Nancy L. Segal, and Stephen Rich, “Personality Similarity in Twins Reared
Apart and Together,” Journal of Personality and Social Psychology 54 (1988):
1031–1039.
13. Lykken and Tellegen, “Happiness Is a Stochastic Phenomenon,” 189.
14. Lykken, Happiness, 2. In fact, an online version of Lykken and Tellegen’s
article shows that the comment concerning the futility of trying to be hap-
pier is crossed out and the words “Not True” are scribbled in the margin. I
wondered if Lykken did that. cogprints.org/767/3/167.pdf.
15. See David T. Lykken, “The Heritability of Happiness,” Harvard Mental
Health Newsletter, 1998.
16. Daniel Goleman, “Forget Money; Nothing Can Buy Happiness, Some Re-
searchers Say,” New York Times, July 16, 1996, http://www.nytimes.com/1996
/07/16/science/forget-money-nothing-can-buy-happiness-some-researchers
-say.html.
17. Ibid.
18. Ibid., C-9. For example, Pavarotti was described as tired and emotional at
the May 2002 concert due to the recent death of his father and his having the
flu. “A Tired and Emotional Pavarotti Keeps Annual Charity Date in His
Hometown,” Hello!, May 29, 2002, www.hellomagazine.com/music/2002
/05/29/pavarotti/.
19. Lykken, “The Heritability of Happiness.”
20. Ibid.
21. Ibid.
22. Ibid.
23. Malcolm Gladwell, Outliers: The Story of Success (Boston: Little, Brown,
2008).
24. Francis Galton, Hereditary Genius: An Inquiry into Its Laws and Conse-
quences (New York: D. Appleton, 1891).
25. Quinn McNemar, “Twin Resemblances in Motor Skills and the Effect of
Practice Thereon,” Journal of Genetic Psychology 42 (1933): 70–97; Daniel
Q. Marisi, “Genetic and Extragenic Variance in Motor Performance,” Acta
Geneticae Medicae et Gemellologiae, Twin Studies 26 (1977): 197–204;
Nicholas G. Martin, John B. Gibson, John G. Oakeshott, et al., “A Twin Study
of Psychomotor Performance during Alcohol Intoxication: Early Results,”
Twin Research 3: Epidemiological and Clinical Studies, ed. Luigi Gedda,
Paolo Parisi, and Walter E. Nance, 89–96 (New York: Alan Liss, 1981).
26. Paul W. Fox, Scott L. Hershberger, and Thomas J. Bouchard Jr., “Genetic
and Environmental Contributions to the Acquisition of a Motor Skill,”
Nature 384 (1996): 356–358. The other psychomotor tasks have not been
analyzed.
 N OT E S TO PAG E S 2 0 4 – 2 0 7 371

27. “Practice Not Perfect without Genes,” Reuters news service, November 27,
1996.
28. Olympic gymnasts Paul and Morgan Hamm are DZ twins with nearly
matching elite talent and, thus, are exceptional.
29. Erinn R. Johnson, “Delaware Girl Learns a New Word the Hard Way at Bee,”
Scripps Howard Foundation Wire, June 10, 2004, www.shfwire.com/node
/2711.
30. David DiLalla, Gregory Carey, Irving I. Gottesman, and Thomas J.
Bouchard Jr., “Heritability of MMPI Personality Indicators of Psychopa-
thology in Twins Reared Apart,” Journal of Abnormal Psychology 105
(1996): 491–499.
31. Starke R. Hathaway and J. Charnley McKinley, “A Multiphasic Personality
Schedule (Minnesota): I. Construction of the Schedule,” Journal of Personal-
ity 10 (1940): 249–254.
32. Interview with the author, July 2010.
33. Hal H. Goldsmith and Irving I. Gottesman, “An Extension of Construct Va-
lidity for Personality Scales Using Twin-based Criteria,” Journal of Research
in Personality 11 (1977): 381–397.
34. T-scores are standard scores that have a mean of 50 and standard deviation
of 10.
35. Nancy L. Segal and Yoon-Mi Hur, “Reared Apart Korean Female Twins: Ge-
netic and Cultural Influences on Life Histories, Physical and Health-related
Measures, and Behavioral Traits,” International Journal of Behavioral Devel-
opment 32 (2008): 542–548.
36. Yoon-Mi Hur and Thomas J. Bouchard Jr., “The Genetic Influence be-
tween Impulsivity and Sensation-Seeking,” Behavior Genetics 27 (1997):
455–463.
37. Benjamin B. Wolman, Dictionary of Behavioral Science, 2nd ed. (New York:
Academic Press, 1989).
38. Marvin Zuckerman, Behavioral Expressions and Biosocial Bases of Sensa-
tion Seeking (New York: Cambridge University Press, 1994).
39. Nancy L. Pedersen, Robert Plomin, Gerald E. McClearn, and Lars Friberg,
“Neuroticism, Extraversion and Related Traits in Adult Twins Reared
Apart,” Journal of Personality and Social Psychology 55 (1988): 950–957.
40. Lindon J. Eaves, Nicholas G. Martin, and Sibyl B. G. Eysenck, “An Applica-
tion of the Analysis of Covariance Structure to the Psychological Study of
Impulsiveness,” British Journal of Mathematical and Statistical Psychology
30 (1977): 185–197.
41. Daisy Schalling, Gunnar Edman, and Marie Åsberg, “Impulsive Cognitive
Style and Inability to Tolerate Boredom: Psychobiological Studies of Tem-
peramental Vulnerabilities,” in Biological Bases of Sensation Seeking: Impul-
sivity and Anxiety, ed. Marvin Zuckerman, 123–145 (Hillsdale, N.J.: Law-
rence Erlbaum, 1983); Daisy Schalling, Marie Åsberg, Gunnar Edman, and
Lars Oreland, “Markers for Vulnerability to Psychopathology: Temperament
Traits Associated with Platelet MAO Activity,” Acta Psychiatrica Scandi-
navica 76 (1987): 172–182. Monoamine oxidase (MAO) is an enzyme in-
volved in activating neurotransmitters so may affect mood.
372 N OT E S TO PAG E S 2 0 8 – 2 1 4

42. Deborah G. Betsworth, “An Investigation of the Genetic and Environmental

Influences on Vocational Interests,” (PhD diss., University of Minnesota, 1992).
43. Daniel P. Moloney, Thomas J. Bouchard Jr., and Nancy L. Segal, “A Genetic and
Environmental Analysis of the Vocational Interests of Monozygotic and Dizy-
gotic Twins Reared Apart,” Journal of Vocational Behavior 39 (1991): 76–109.
44. Deborah G. Betsworth, Thomas J. Bouchard Jr., Catherine R. Cooper, et al.,
“Genetic and Environmental Influences on Vocational Interests Assessed
Using Adoptive and Biological Families and Twins Reared Apart and To-
gether,” Journal of Vocational Behavior 44 (1994): 263–278.
45. Richard D. Arvey, Brian P. McCall, Thomas J. Bouchard Jr., Paul Taubman,
and Marcie A. Cavanaugh, “Genetic Influences on Job Satisfaction and Work
Values,” Personality and Individual Differences 17 (1994): 21–33.
46. Richard D. Arvey, Thomas J. Bouchard Jr., Nancy L. Segal, and Lauren M.
Abraham, “Job Satisfaction: Environmental and Genetic Components,”
Journal of Applied Psychology 74 (1989): 187–192.
47. Arvey et al., “Genetic Influence.”
48. Abraham Tesser, “On the Importance of Heritability in Psychological Re-
search: The Case of Attitudes,” Psychological Review 100 (1993): 129–142.
49. Jere R. Behrman, Mark R. Rosenzweig, and Paul Taubman, “College Choice
and Wages: Estimates Using Data on Female Twins,” Review of Economics
and Statistics 78 (1996): 672–685.
50. Lindon Eaves, Brian D’Onofrio, and Robert Russell, “Transmission of Reli-
gion and Attitudes,” Twin Research 2 (1999): 59–61.
51. Genomics.energy.gov, “History of the Human Genome Project,” Human Ge-
nome Project Information, updated May 12, 2011, www.ornl.gov/sci/tech
resources/Human_Genome/project/hgp.shtml.
52. Associated Press, “Scientists Finish First Phase in Mapping of Human Genes,”
New York Times, March 19, 1996, www.nytimes.com/1996/03/19/science
/scientists-finish-first-phase-in-mapping-of-human-genes.html; Nicholas Wade,
“Research Team Takes Big Stride in the Mapping of Human Genes,” New
York Times, March 15, 1997, www.nytimes.com/1997/03/15/us/research-team
-takes-big-stride-in-the-mapping-of-human-genes.html.
53. Thomas J. Bouchard Jr., “Genes, Environment and Personality,” Science 264
(1994): 1700–1701. Bouchard’s paper, listed under “Perspectives,” followed
J. H. Thomas’s “Mind of the Worm,” a fascinating piece on how Caenorhab-
ditis elegans could inform us about behavioral complexity and plasticity.
54. Ibid., 1701.
55. Frank J. Sulloway, Born to Rebel: Birth Order, Family Dynamics, and Cre-
ative Lives (New York: Pantheon, 1996).
56. Aurelio José Figueredo, Jon A. Sefcek, Geneva Vasquez, Barbara H. Brum-
bach, James E. King, and W. Jake Jacobs, “Evolutionary Personality Psychol-
ogy,” The Handbook of Evolutionary Psychology, ed. David M. Buss, 851–
877 (New York: Wiley, 2005).
57. Richard D. Arvey and Thomas J. Bouchard Jr., “Genetics, Twins, and Organ-
izational Behavior,” in Research in Organizational Behavior 16, ed. Barry
Staw and Larry L. Cummings, 47–82 (Greenwich, Conn.: JAI Press, 1994).
 N OT E S TO PAG E S 2 1 4 – 2 1 7 373

58. Thomas J. Bouchard Jr., “Longitudinal Studies of Personality and Intelligence:

A Behavior Genetic and Evolutionary Psychology Perspective,” in Interna-
tional Handbook of Personality and Intelligence, ed. Donald H. Saklofske and
Moshe Zeidner, 81–106 (New York: Plenum, 1995).
59. Ronald S. Wilson, “The Louisville Twin Study: Developmental Synchronies
in Behavior,” Child Development 54 (1983): 298–316.
60. Robert Altemeyer, Enemies of Freedom: Understanding Right-Wing Author-
itarianism (San Francisco: Jossey-Bass, 1988).
61. Thomas J. Bouchard Jr., “Genetics and Evolution: Implications for Psychologi-
cal Theories,” in Measures of the Five Factor Model and Psychological Type: A
Major Convergence of Research and Theory, ed. James Newman, 20–42
(Gainesville, FL: Center for the Applications of Psychological Type, 1995), p. 42.
62. Thomas J. Bouchard Jr., David T. Lykken, Matthew McGue, Nancy L. Segal,
and Auke Tellegen, “Sources of Human Psychological Differences: The Min-
nesota Study of Twins Reared Apart,” Science 250 (1990): 223–228.
63. Thomas J. Bouchard Jr., “The Galton Lecture: Behavior Genetic Studies of
Intelligence, Yesterday and Today: The Long Journey from Plausibility to
Proof,” Journal of Biosocial Science 28 (1996): 527–555, p. 529.
64. Thomas J. Bouchard Jr., “Whenever the Twain Shall Meet,” The Sciences 37
(1997): 52–57.
65. Dolly was cloned in 1996, but the event was announced in 1997.
66. Nancy L. Segal, “Behavioral Aspects of Intergenerational Cloning: What
Twins Tell Us,” Jurimetrics 38 (1997): 57–67.
67. Donors and clones would share their parents in one sense because the donor’s
parents conceived the donor and the donor conceived a genetically identical
child. However, in another sense they would not because the donor, and not
the donor’s parents, actually created and possibly gestated the child. Further-
more, mitochondrial (extracellular) DNA is transmitted intact from mothers
to children. A cloned child’s mitochondrial DNA would typically come from
an unrelated individual who provided the enucleated egg with which the do-
nor’s cell nucleus was fused; in this case the mitochondrial DNA of the clone
and the donor would differ. However, if the donor also provided the enucle-
ated egg, then the mitochondrial DNA would be the same. See Matthew J.
Evans, Cagan Gurer, John D. Loike, et al., “Mitochondrial DNA Genotypes in
Nuclear Transfer-Derived Cloned Sheep,” Nature Genetics 23 (1999): 90–93.
68. Laura E. Berk, Child Development (Needham Heights, Mass.: Allyn &
Bacon, 1997).
69. The most recent edition is Robert Plomin, John C. DeFries, Gerald E. Mc-
Clearn, and Peter McGuffin, Behavioral Genetics, 5th ed. (New York: Worth
Publishers, 2008).
70. Thomas J. Bouchard Jr. and Matthew McGue, “Familial Studies of Intelli-
gence: A Review,” Science 212 (1981): 1055–1059.
71. John C. Loehlin, “Partitioning Genetic and Environmental Contributions to
Behavioral Development,” American Psychologist 44 (1989): 1285–1,292.
72. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A
Study of Heredity and Environment (Chicago: University of Chicago Press,
374 N OT E S TO PAG E S 2 1 7– 2 2 3

1937); James Shields, Monozygotic Twins Brought Up Apart and Brought

Up Together (London: Oxford University Press, 1962); Niels Juel-Nielsen,
Individual and Environment: Monozygotic Twins Reared Apart (New York:
International Universities Press, 1980).
73. Thomas J. Bouchard Jr., David T. Lykken, Matthew McGue, Nancy L. Segal,
and Auke Tellegen, “Sources of Human Psychological Difference: The Min-
nesota Study of Twins Reared Apart,” Science 250 (1990): 223–228; Nancy
L. Pedersen, Robert Plomin, John R. Nesselroade, and Gerald E. McClearn,
“A Quantitative Genetic Analysis of Cognitive Abilities during the Second
Half of the Life Span,” Psychological Science 3 (1992): 346–353.
74. Gerald E. McClearn, Boo Johansson, Stig Berg, et al., “Substantial Genetic
Influence on Cognitive Abilities in Twins 80 or More Years Old,” Science
276 (1997): 1560–1563.
75. Denise L. Newman, Auke Tellegen, and Thomas J. Bouchard Jr., “Individual
Differences in Adult Ego Development: Sources of Influence in Twins Reared
Apart,” Journal of Personality and Social Psychology 74 (1998): 985–995;
Kathryn McCourt, Thomas J. Bouchard Jr., David T. Lykken, Auke Tellegen,
and Margaret Keyes, “Authoritarianism Revisited: Genetic and Environmen-
tal Influences Examined in Twins Reared Apart and Together,” Personality
and Individual Differences 27 (1999): 985–1014.
76. Lawrence Wright, “Double Mystery,” The New Yorker, August 7, 1995,
45–62.
77. Roderic Angle, “Nature’s Clones,” Psychology Today, August 1997, 38–69.
78. Lawrence Wright, Twins: And What They Tell Us about Who We Are (New
York): John Wiley & Sons, 1997.

11. “Larks” and “Owls,” Ego Development,

and Authoritarianism (1998–2002)
1. Yoon-Mi Hur, Thomas J. Bouchard Jr., and David T. Lykken, “Genetic and
Environmental Influences on Morningness-Eveningness,” Personality and In-
dividual Differences 25 (1998): 917–925.
2. M. D. Drennan, J. Selby, G. A. Kerkhof, and D. I. Boomsma, “Morningness/
Eveningness Is Heritable,” Society for Neuroscience Abstracts 18 (1992): 196.
3. J. A. Horne and O. Ostberg, “A Self-Assessment Questionnaire to Determine
Morningness-Eveningness in Human Circadian Rhythms,” International
Journal of Chronobiology 4 (1976): 97–110. Our form was adapted from
the published one.
4. Jane Loevinger, Ego Development: Conceptions and Theories (San Fran-
cisco: Jossey-Bass, 1976).
5. Jane Loevinger and Ruth Wessler, Measuring Ego Development, vol. 1 (San
Francisco: Jossey-Bass, 1970), 8.
6. Denise L. Newman, Auke Tellegen, and Thomas J. Bouchard Jr., “Individual
Differences in Adult Ego Development: Sources of Influence in Twins Reared
Apart,” Journal of Personality and Social Psychology 74 (1998): 985–995.
7. Ibid.
 N OT E S TO PAG E S 2 2 3 – 2 2 8 375

8. Loevinger and Wessler, Measuring Ego Development; Jane Loevinger and

L. X. Hy Le, “Measuring Ego Development: Supplementary Manual and Sen-
tences for Form 8 of the Washington University Sentence Completion Test”
(unpublished manuscript, Washington University, 1989).
9. Jane Loevinger, “The Construct Validity of the Sentence Completion Test of
Ego Development,” Applied Psychological Measures 3 (1979): 281–311.
10. Loevinger and Wessler, Measuring Ego Development.
11. Ego development scores were based on a cumulative frequency or “ogive” of
item-level scores. Scoring rules were provided in the test manual.
12. We found ego development heritabilities of 0.54 (without controlling for
cognition), 0.56 (controlling for verbal skill), and 0.47 (controlling for IQ).
13. Newman et al., “Individual Differences,” p. 993.
14. Robert R. McCrae and Paul T. Costa, “Openness to Experience and Ego
Level in Loevinger’s Sentence Completion Test: Dispositional Contributions
to Developmental Models of Personality,” Journal of Personality and Social
Psychology 39 (1980): 1179–1190.
15. Lawrence D. Cohen and P. Michiel Westenberg, “Intelligence and Maturity:
Meta-Analytic Evidence for the Incremental and Discriminative Validity of
Loevinger’s Measure of Ego Development,” Journal of Personality and Social
Psychology 86 (2004): 760–772. Loevinger notes that word count is correlated
with complexity of expression, but not all lengthy responses are complex.
16. Dewey K. Ziegler, Yoon-Mi Hur, Thomas J. Bouchard Jr., Ruth S. Hassanein,
and Ruth Barter, “Migraine in Twins Raised Together and Apart,” Headache
38 (1998): 417–422.
17. Harvald Bent and Mogens Hauge, “A Catamnestic Investigation of Danish
Twins,” Danish Medical Bulletin 3 (1956): 150–158; Kathleen R. Merikan-
gas, C. Tierney, Nicholas G. Martin, Andrew C. Heath, and Neil Risch, “Ge-
netics of Migraine in the Australian Twin Registry,” in New Advances in
Headache Research, vol. 4, ed. F. Clifford Rose, 27–28 (London: Smith-
Gordon, 1994).
18. Dewey K. Ziegler, Ruth S. Hassanein, David Harris, and Roberta Stewart,
“Headache in a Non-clinic Twin Population,” Headache 14 (1975): 213–218.
19. Susan L. Farber, Identical Twins Reared Apart: A Reanalysis (New York:
Basic Books, 1980).
20. Brian S. Schwartz, Walter F. Stewart, David Simon, and Richard Lipton.
“Epidemiology of Tension Type Headache,” Journal of the American Medi-
cal Association 279 (1998): 381–383.
21. In 2003, the Swedish Twin and Adoption Study of Aging reported a 0.48
heritability for migraine headaches, based on 413 pairs of MZA, DZA, MZT,
and DZT female twin pairs. See Dan A. Svensson, Bo Larsson, Elisabet
Waldenlind, and Nancy L. Pedersen, “Shared Rearing Environment in Mi-
graine: Results from Twins Reared Apart and Twins Reared Together,” Head-
ache 43 (2003): 235–244.
22. Yoon-Mi Hur, Thomas J. Bouchard Jr., Elke Eckert, “Genetic and Environ-
mental Influences on Self-Reported Diet: A Reared-Apart Twin Study,” Phys-
iology and Behavior 64 (1998): 629–636.
376 N OT E S TO PAG E S 2 2 8 – 2 32

23. Richard R. Fabsitz, Robert J. Garrison, Manning Feinleib, and Marthana

Hjortland, “A Twin Analysis of Dietary Intake: Evidence for a Need to Con-
trol for Possible Environmental Differences in MZ and DZ Twins,” Behavior
Genetics 8 (1978): 15–25.
24. Vegemite is a yeast abstract made from leftover brewers yeast with vegetable
and spice additives. Marmite is a yeast abstract, namely, a byproduct of
brewing beer.
25. Claude Bouchard, Angelo Tremblay, Jean-Pierre Després, et al., “The Re-
sponse to Long-Term Overfeeding in Identical Twins,” New England Journal
of Medicine 322 (1990): 1477–1482.
26. Richard Luschinger and Geoffrey E. Arnold, Genetics of the Voice, Voice-
Speech-Language (Belmont, Calif.: Wadsworth, 1965); György Forrai and Géza
Gordos, “A New Acoustical Method for the Discrimination of Monozygotic
and Dizygotic Twins,” Acta Paediatrica Hungarica 24 (1983): 315–322.
27. Farber, Identical Twins Reared Apart.
28. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962).
29. Robert T. Sataloff, “Genetics of the Voice,” Journal of Voice 8 (1995): 16–19.
30. Michael J. Hammer, “Comparison of Within-Pair Similarities of Monozy-
gotic and Dizygotic Twins for Mean F0 and F0 Range” (M.A. thesis, Univer-
sity of Kansas, 1998).
31. Grant Fairbanks, Voice and Articulation Drillbook, 2nd ed. (New York:
Harper & Row, 1960), p. 127.
32. Farber, Identical Twins Reared Apart.
33. Theodor W. Adorno, Else Frenkel-Brunswik, Daniel Levinson, and Nevitt
Sanford, The Authoritarian Personality (New York: Harper & Row, 1950).
34. Robert Altemeyer, Enemies of Freedom: Understanding Right-Wing Author-
itarianism (San Francisco: Jossey-Bass, 1988).
35. Merrill Ring was a member of the philosophy faculty at California State
University–Fullerton from 1973 until his 2010 retirement.
36. John C. Loehlin and Robert C. Nichols, Heredity, Environment and Person-
ality: A Study of 850 Sets of Twins (Austin: University of Texas Press, 1976).
37. Nicholas G. Martin, Lindon J. Eaves, Andrew C. Heath, Rosemary Jardine,
Lynn M. Feingold, and Hans J. Eysenck, “Transmission of Social Attitudes,”
Proceedings of the National Academy of Sciences of the United States of
America 83 (1986): 4364–4368.
38. Altemeyer, Enemies of Freedom.
39. Glenn D. Wilson and John R. Patterson, “A New Measure of Conservatism,”
British Journal of Social and Clinical Psychology 7 (1968): 264–269.
40. Kathryn McCourt, Thomas J. Bouchard Jr., David T. Lykken, Auke Tellegen,
and Margaret Keyes, “Authoritarianism Revisited: Genetic and Environmen-
tal Influences Examined in Twins Reared Apart and Together,” Personality
and Individual Differences 27 (1999): 985–1014.
41. Kathryn Corson, “A Genetic and Environmental Analysis of Right-Wing
Authoritarianism in Twins Reared Apart and Together” (PhD diss., Univer-
sity of Minnesota, 1996).
 N OT E S TO PAG E S 2 32 – 2 37 377

42. An assortative mating parameter was included in this model.

43. The only significant MISTRA correlation showed that high authoritarianism
was associated with lower rearing socioeconomic status (−0.19). The only
significant reared-together twin correlations linked higher authoritarianism
with fewer years of father’s education (−0.23) and lower rearing socioeco-
nomic status (−0.17). These relationships were too small to suggest practical
importance.
44. Thomas J. Bouchard Jr., Matt McGue, David Lykken, and Auke Tellegen,
“Intrinsic and Extrinsic Religiousness: Genetic and Environmental Influences
and Personality Correlates,” Twin Research 2 (1999): 88–98.
45. Lindon Eaves, Brian D’Onofrio, and Robert Russell, “Transmission of Reli-
gion and Attitudes,” Twin Research 2 (1999): 59–61, p. 59.
46. Niels G. Waller, Brian A. Kojetin, Thomas J. Bouchard Jr., David T. Lykken,
and Auke Tellegen, “Genetic and Environmental Influences on Religious In-
terests, Attitudes, and Values: A Study of Twins Reared Apart and Together,”
Psychological Science 1 (1990): 138–142.
47. Allen E. Bergin, “Religiosity and Mental Health: A Critical Reevaluation and
Meta-Analysis,” Professional Psychology: Research and Practice 14 (1983):
170–184.
48. Wade C. Rowatt, Jordan LaBouff, Megan Johnson, Paul Froese, and Jo-Ann
Tsang, “Associations among Religiousness, Social Attitudes, and Prejudice in
a National Random Sample of American Adults,” Psychology of Religion
and Spirituality 1 (2009): 14–24.
49. Gordon W. Allport and J. Michael Ross, “Personal Religious Orientation and
Prejudice,” Journal of Personality and Social Psychology 5: (1967): 432–443,
p. 172.
50. Edward O. Wilson, On Human Nature (Cambridge, Mass.: Harvard Univer-
sity Press, 1978).
51. Edward O. Wilson, Consilience: The Unity of Knowledge (New York: Alfred
A. Knopf, 1998).
52. Bouchard et al., “Intrinsic and Extrinsic Religiousness,” p. 89.
53. Richard L. Gorsuch and G. Daniel Venable, “Development of an ‘Age Univer-
sal’ I-E Scale,” Journal of the Scientific Study of Religion 22 (1983): 181–187.
54. The Public Opinion Inventory also used this format (−4 to +4).
55. David Cesarini, Christopher T. Dawes, James H. Fowler, Magnus Johannes-
son, Paul Lichtenstein, and Björn Wallace, “Heritability of Cooperative Be-
havior in the Trust Game,” Proceedings of the National Academy of Sciences
of the United States of America 105 (2008): 3721–3726; James H. Fowler,
Laura A. Baker, and Christopher T. Dawes, “Genetic Variation in Political
Participation,” American Political Science Review 102 (2008): 233–248; Alan
G. Sanfey, “Social Decision-Making: Insights from Game Theory and Neuro-
science,” Science 318 (2007): 598–602.
56. Robert Plomin, John C. DeFries, Gerald E. McClearn, and Peter McGuffin,
Behavioral Genetics, 5th ed. (New York: Worth Publishers, 2008).
57. Thomas J. Bouchard and David T. Lykken, “Life Achievement in a Sample
of Twins Reared Apart: Estimating the Role of Genetic and Environmental
378 N OT E S TO PAG E S 2 37– 2 4 0

Influences,” in Talent Development III. Proceedings from the 1995 Henry B.

and Jocelyn Wallace National Symposium on Talent Development, ed. Nicho-
las Colangelo and Susan G. Assouline, 81–97 (Scottsdale, Ariz.: Gifted Psy-
chology Press, 1999).
58. Frank Barron, The Psychology of Creativity (New York: Holt, Rinehart &
Winston, 1965).
59. Frank Barron and Paolo Parisi, “Twin Resemblances in Creativity and in
Esthetic and Emotional Response,” Acta Geneticae Medicae et Gemellolo-
giae 25 (1976): 213–217.
60. Niels G. Waller, Thomas J. Bouchard Jr., David T. Lykken, Auke Tellegen,
and Dawn M. Blacker, “Creativity, Heritability, Familiality: Which Word
Does Not Belong?” Psychological Inquiry 4 (1993): 235–237.
61. Laird D. McLean, “Organizational Culture’s Influence on Creativity and Inno-
vation: A Review of the Literature and implications for Human Resources De-
velopment,” Advances in Developing Human Resources 7 (2005): 226–246.
62. Bouchard and Lykken, “Life Achievement,” p. 90.
63. Francis Galton, Hereditary Genius: An Inquiry into Its Law and Conse-
quences (London: Macmillan, 1891).
64. Karl Pearson, The Life, Letters and Labours of Francis Galton, vol. 1 (Cam-
bridge, U.K.: Cambridge University Press, 1924).
65. Isabel B. Myers and Mary H. McCaulley, Manual: A Guide to the Develop-
ment and Use of the Myers-Briggs Type Indicator (Palo Alto, Calif.: Consult-
ing Psychologists Press, 1985).
66. Thomas J. Bouchard Jr. and Yoon-Mi Hur, “Genetic and Environmental
Influences on the Continuous Scales of the Myers-Briggs Type Indicator: An
Analysis Based on Twins Reared Apart,” Journal of Personality 66 (1998):
135–149.
67. Thomas J. Bouchard Jr., Matt McGue, Yoon-Mi Hur, and Joseph M. Horn,
“A Genetic and Environmental Analysis of the California Psychological In-
ventory Using Adult Twins Reared Apart and Together,” European Journal of
Personality 12 (1998): 307–320.
68. Thomas J. Bouchard Jr. and Nancy L. Pedersen, “Twins Reared Apart: Na-
tures Double Experiment,” in On the Way to Individuality: Current Method-
ological Issues in Behavioral Genetics, ed. Michele C. LaBuda and Elena L.
Grigorenko, 71–93 (Commack, N.Y.: Nova Scientific, 1999).
69. Lindon J. Eaves, Hans J. Eysenck, Nicholas G. Martin, Genes, Culture and
Personality: An Empirical Approach (New York: Academic Press, 1989);
Kerry L. Jang et al., “Heritability of the Big Five Personality Dimensions and
Their Facets: A Twin Study,” Journal of Personality 64 (1996): 577–591;
John C. Loehlin, Genes and Environment in Personality Development (New-
bury Park, Calif.: Sage, 1992).
70. David L. DiLilla, Irving I. Gottesman, Gregory J. Carey, and Thomas J.
Bouchard Jr., “Heritability of MMPI Harris-Lingoes and Subtle-Obvious
Subscales in Twins Reared Apart,” Assessment 6 (1999): 353–366.
71. Daniel N. Wiener, “Subtle and Obvious Keys for the MMPI,” Journal of
Consulting Psychology 12 (1948): 164–170.
 N OT E S TO PAG E S 2 4 0 – 2 47 379

72. Alan F. Friedman, Richard Lewak, David S. Nichols, and James T. Webb,
Psychological Assessment with the MMPI-2 (Mahwah, N.J.: Lawrence
Erlbaum, 2001).
73. David L. DiLalla, Irving I. Gottesman, Gregory Carey, and Thomas J.
Bouchard Jr., “Heritability of MMPI Harris-Lingoes and Subtle-Obvious
Subscales in Twins Reared Apart,” Assessment 6 (1999): 353–366.
74. David L. DiLalla, Gregory Carey, Irving I. Gottesman, and Thomas J. Bouchard
Jr., “Heritability of MMPI Personality Indicators of Psychopathology in Twins
Reared Apart,” Journal of Abnormal Psychology 105 (1996): 491–499.
75. Seymour S. Kety, “From Rationalization to Reason,” American Journal of
Psychiatry 131 (1974): 957–963.
76. Kristian E. Markon, Robert F. Krueger, Thomas J. Bouchard Jr., and Irving I.
Gottesman, “Normal and Abnormal Personality Traits: Evidence for Genetic
and Environmental Relationships in the Minnesota Study of Twins Reared
Apart,” Journal of Personality 70 (2002): 661–693.
77. Richard L. Doty, Paul Shaman, and Michael Dann, “Development of the
University of Pennsylvania Smell Identification Test: Standardized Microen-
capsulated Test for Olfactory Function,” Physiology and Behavior 32 (1984):
489–502.
78. Bryan S. Michalowicz, Dorothee P. Aeppli, John G. Virag, et al., “Periodontal
Findings in Adult Twins,” Journal of Periodontology 62 (1991): 293–299.
79. Bryan S. Michalowicz, Dorothee P. Aeppli, Ramesh K. Kuba, et al., “A Twin
Study of Genetic Variation in Proportional Radiographic Alveolar Bone
Height,” Journal of Dental Research 70 (1991): 1431–1435.
80. Bryan S. Michalowicz, Larry F. Wolff, David Klump, et al., “Periodontal Bac-
teria in Adult Twins,” Journal of Periodontology 70 (1999): 263–273.
81. Bryan S. Michalowicz, Bruce L. Pihlstrom, J.  S. Hodges, and Thomas J.
Bouchard Jr., “No Heritability of Temporomandibular Joint Signs and Symp-
toms,” Journal of Dental Research 79 (2000): 1573–1578.
82. See, for example, Nancy L. Segal, Scott L. Hershberger, and Sara Arad,
“Meeting One’s Twin: Perceived Social Closeness and Familiarity,” Evolution-
ary Psychology 1 (2003): 70–95; Wendy Johnson, Thomas J. Bouchard Jr.,
Nancy L. Segal, Margaret Keyes, and Jay Samuels, “The Stroop Color-Word
Test: Genetic and Environmental Influences, Reading, Mental Ability, and Per-
sonality Correlates,” Journal of Educational Psychology 95 (2003): 58–65.

12. Twin Relationships, Social Attitudes, and

Mental Abilities (2003–2005)
1. Nightline, ABC News, October 29, 1989.
2. Nancy L. Segal, “Twin, Adoption and Family Methods as Approaches to the
Evolution of Individual Differences,” in The Evolution of Personality and
Individual Differences, ed. David M. Buss and Patricia Hawley, 303–337
(Oxford: Oxford University Press, 2011).
3. Nancy L. Segal, “Education Issues,” Twin Research and Human Genetics 8
(2005): 409–414.
380 N OT E S TO PAG E S 2 47– 2 5 3

4. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A

Study of Heredity and Environment (Chicago: University of Chicago Press,
1937), 153.
5. Ibid, 195.
6. Ibid., 239.
7. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962), 51.
8. Niels Juel-Nielsen, Individual and Environment: Monozygotic Twins Reared
Apart (New York: International Universities Press, 1980), 16.
9. Shields, Monozygotic Twins, 21. The majority of Shields’s twins were identi-
fied following a 1953 BBC television program, Twin Sister, Twin Brother, in
which twins were asked to complete a questionnaire published in the then-
current Radio Times. A special appeal was made for MZA twins, although
nine of the eleven DZA twins for whom Shields obtained some behavioral
data were identified via the BBC.
10. Nancy L. Segal, Scott L. Hershberger, and Sara Arad, “Meeting One’s Twin:
Perceived Social Closeness and Familiarity,” Evolutionary Psychology 1 (2003):
70–95.
11. Ellen A. Stewart, Exploring Twins: Towards a Social Analysis of Twinship
(London: Macmillan Press, 2000).
12. Maria Leek and Peter K. Smith, “Cooperation and Conflict in Three-
Generation Families,” in The Psychology of Grandparenthood: An Interna-
tional Perspective, ed. Peter K. Smith, 177–194 (London: Routledge, 1991).
13. Daniel G. Freedman, Human Sociobiology: A Holistic Approach (New York:
Free Press, 1979), 129.
14. Nancy L. Segal, Someone Else’s Twin: The True Story of Babies Switched at
Birth (Amherst, N.Y.: Prometheus Books, 2011).
15. Robert F. Krueger, Kristian E. Markon, and Thomas J. Bouchard Jr., “The
Extended Genotype: The Heritability of Personality Accounts for the Herita-
bility of Recalled Family Environments in Twins Reared Apart,” Journal of
Personality 71 (2003): 809–833.
16. David C. Rowe, “Environmental and Genetic Influences on Dimensions of
Perceived Parenting: A Twin Study,” Developmental Psychology 17 (1981):
203–208.
17. David T. Lykken and Auke Tellegen, “Happiness Is a Stochastic Phenome-
non,” Psychological Science 7 (1996): 186–189.
18. Robert F. Krueger, Susan South, Wendy Johnson, and William Iacono, “The
Heritability of Personality Is Not Always 50%: Gene-Environment Inter-
actions and Correlations between Personality and Parenting,” Journal of
Personality 76 (2008): 1485–1521.
19. Thomas J. Bouchard Jr., Nancy L. Segal, Auke Tellegen, Matt McGue, Mar-
garet Keyes, and Robert Krueger, “Genetic Influence on Social Attitudes:
Another Challenge to Psychology from Behavior Genetics,” in Behavior Ge-
netic Principles: Development, Personality and Psychopathology, ed. Lisa-
beth F. DiLalla, 89–104 (Washington, D.C.: American Psychological Associa-
tion, 2004).
 N OT E S TO PAG E S 2 5 3 – 2 5 6 381

20. Ibid., 95. He is referring to Nicholas G. Martin, Lindon J. Eaves, Andrew C.

Heath, Rosemary Jardine, Lynn M. Feingold, and Hans J. Eysenck, “Trans-
mission of Social Attitudes,” Proceedings of the National Academy of Sci-
ences of the United States of America 83 (1986): 4364–4368.
21. Auke Tellegen, David T. Lykken, Thomas J. Bouchard Jr., Kimerly J. Wilcox,
Nancy L. Segal, and Stephen Rich, “Personality Similarity in Twins Reared
Apart and Together,” Journal of Personality and Social Psychology 54 (1988):
1031–1039.
22. Thomas J. Bouchard and Matt McGue, “Genetic and Environmental Influ-
ences on Human Psychological Differences” Journal of Neurobiology 54
(2003): 4–45; Bouchard Jr. et al., “Genetic Influence on Social Attitudes.”
23. Deborah Finkel and Matt McGue, “Sex Differences and Non-Additivity in
Heritability of the Multidimensional Personality Questionnaire Scales,”
Journal of Personality and Social Psychology 72 (1997): 929–938.
24. Glenn D. Wilson and John R. Patterson, “A New Measure of Conservatism,”
British Journal of Social and Clinical Psychology 7 (1968): 264–269.
25. John C. Loehlin and Robert C. Nichols, Heredity, Environment and Person-
ality: A Study of 850 Sets of Twins (Austin: University of Texas Press, 1976).
26. Lindon J. Eaves, Andrew Heath, Nicholas Martin, et al., “Comparing the
Biological and Cultural Inheritance of Personality and Social Attitudes in
the Virginia 30,000 Study of Twins and Their Relatives,” Twin Research 2
(1999): 62–80.
27. Thomas J. Bouchard Jr., Nancy L. Segal, Auke Tellegen, Matt McGue, Mar-
garet Keyes, and Robert Krueger, “Evidence for the Construct Validity and
Heritability of the Wilson-Patterson Conservatism Scale: A Reared-Apart
Twins Study of Social Attitudes,” Personality and Individual Differences 34
(2003): 959–969.
28. Peter Watson, Twins: An Uncanny Relationship? (Chicago: Contemporary
Books, 1981).
29. Thomas J. Bouchard, David T. Lykken, Matt McGue, and Auke Tellegen,
“Intrinsic and Extrinsic Religiousness: Genetic and Environmental Influ-
ences and Personality Correlates,” Twin Research 2 (1999): 88–98.
30. Laura B. Koenig, Matt McGue, Robert F. Krueger, and Thomas J. Bouchard
Jr., “Genetic and Environmental Influences on Religiousness: Findings for
Retrospective and Current Religiousness Ratings,” Journal of Personality 73
(2005): 471–488.
31. Attitudes vary as a function of cohort as well as age. When in vitro fertiliza-
tion was introduced in 1978, people worried that it would alter family dynam-
ics and structure in negative ways. However, in vitro fertilization has been an
effective reproductive technology, helping many infertile couples have the
families they desired. The longitudinal MISTRA data on both twins and non-
twins should shed light on factors affecting attitudinal change, both more
conservative and less conservative.
32. In 2010 e-mail correspondence, Koenig wrote, “When I went on job talks the
twins reared-apart were the first thing people mentioned if they had any
knowledge at all about behavior genetics.” I was sure that I saw Koenig at
382 N OT E S TO PAG E S 2 5 6 – 2 6 1

the 2009 Western Psychological Association meeting in Portland, Oregon,

but I was wrong: it was Anne, her identical twin sister who teaches psychol-
ogy at the University of San Diego. I did not know Koenig was a twin. Also
see Chapter 11; religiosity-personality associations were generally not found
by the MISTRA.
33. Bouchard et al., “Genetic Influence on Social Attitudes,” p. 95.
34. “It’s Always Better with Twins”: Remembering the Life and Research of
David T. Lykken,” special session, International Congress on Twin Studies,
Ghent, Belgium, June 7–10, 2007.
35. Jürgen R. Stroop, “Studies of Interference in Serial Verbal Reactions,” Jour-
nal of Experimental Psychology 18 (1935): 643–662.
36. Raymond De Young, “Stroop Task: A Test of Capacity to Direct Attention,”
EPLab OnLine Measures—Stroop Task—Version: 4.2, updated January 5,
2010, http://snre.umich.edu/eplab/demos/st0/stroopdesc.html.
37. Charles J. Golden, Stroop Color and Word Test (Chicago: Stoelting, 1978).
38. Lissy F. Jarvik, “Genetic Components and Intellectual Functioning during Senes-
cence: A 20-Year Study of Aging Twins,” Behavior Genetics 2 (1972): 159–171.
39. Richard L. Slossen, Slossen Intelligence Test and Oral Reading Test for Chil-
dren and Adults (East Aurora, N.Y.: Slossen Educational, 1984); Richard W.
Woodcock, Woodcock Reading Mastery Tests-Revised Forms G & H-Exam-
iner’s Manual (Circle Pines, Minn.: American Guidance Services, 1987).
40. Wendy Johnson, Thomas J. Bouchard Jr., Nancy L. Segal, Margaret Keyes,
and Jay Samuels, “The Stroop Color-Word Test: Genetic and Environmental
Influences, Reading, Mental Ability, and Personality Correlates,” Journal of
Educational Psychology 95 (2003): 58–65.
41. Wendy Johnson, Thomas J. Bouchard Jr., Nancy L. Segal, and Jay Samuels,
“General Intelligence and Reading Performance in Adults: Is the Genetic
Factor Structure the Same as for Children?” Personality and Individual Dif-
ferences 38 (2005): 1413–1428.
42. Fluid ability refers to individuals’ genetically based abilities and are not
learned. Crystallized ability is intelligence based on culture and learning.
43. John C. DeFries, David W. Fulker, and Michelle C. Labuda, “Evidence for a
Genetic Aetiology in Reading Disability of Twins,” letter to the editor, Na-
ture 329 (1987): 537–539.
44. Wendy Johnson, Thomas J. Bouchard Jr., Robert F. Krueger, Matt McGue,
and Irving I. Gottesman, “Just One g: Consistent Results from Three Test
Batteries,” Intelligence 32 (2004): 95–107.
45. William Kessen, The Child (New York: John Wiley & Sons, 1965).
46. Arthur R. Jensen, The g Factor: The Science of Mental Ability (Westport,
Conn.: Praeger, 1998).
47. A model recognizing separate mental abilities is Howard Gardner’s theory of
multiple intelligences, described in Frames of Mind (New York: Basic Books,
2011). Gardner’s model has not been subjected to the same rigorous exami-
nations as the IQ test and other psychometric measures. A more recent evo-
lutionary based theory of intelligence refers to mental modules or content-
independent mechanisms; see Leda Cosmides and John Tooby, “The Modular
Nature of Human Intelligence,” in The Origin and Evolution of Intelligence,
 N OT E S TO PAG E S 2 6 1 – 2 6 8 383

ed. Arnold B. Scheibel and J. William Schopf, 71–101 (Toronto: John Barlett,
1997).
48. Some participants did not complete all tests due to time constraints. When
possible, missing data were dealt with by imputation of scores according to
strict guidelines.
49. Johnson et al., “Just One g,” 104.
50. Thomas J. Bouchard Jr., “Genetic Influence on Human Intelligence (Spearman’s
g): How Much?” Annals of Human Biology 36 (2009): 527–544, p. 532.
51. Wendy Johnson and Thomas J. Bouchard Jr., “The Structure of Human Intel-
ligence: It’s Verbal, Perceptual, and Image Rotation (VPR) Not Fluid and
Crystallized,” Intelligence 33 (2005): 393–416.
52. Wendy Johnson and Thomas J. Bouchard Jr., “Constructive Replication of
the Visual-Perceptual-Image Rotation (VPR) Model in Thurstone’s (1941)
Battery of 60 Tests of Mental Ability,” Intelligence 33 (2005): 417–430.
53. Wendy Johnson, Jan te Nijenhuis, and Thomas J. Bouchard Jr., “Replication
of the Hierarchical Visual-Perceptual-Image Rotation Model in de Wolff and
Buiten’s (1963) Battery of 46 Tests of Mental Ability,” Intelligence 35 (2007):
69–81.
54. Wendy Johnson, Jan te Nijenhuis, and Thomas J. Bouchard Jr., “Still Just 1
g: Consistent Results from Five Test Batteries,” Intelligence 36 (2008):
81–95.
55. Professor Tony Vernon, personal communication, July 8, 2010. British psy-
chologist P. (Philip) E. Vernon is Tony Vernon’s late father.
56. Edward Novitski, Human Genetics, 2nd ed. (New York: Macmillan, 1982).
57. Thomas J. Bouchard Jr., “Genetic Influence on Human Psychological Traits:
A Survey,” Current Directions in Psychological Science 13 (2004): 148–151.
58. Ibid., 151.
59. Ibid., 148.

13. Sexual Development, Fluctuating Asymmetry, Body Size,

and the Structure of Intelligence (2006 and Beyond)
1. I hope to complete a brief assessment of this pair at a later date.
2. Nancy L. Segal and Joanne Hoven Stohs, “Resemblance for Age at Menarche
in Female Twins Reared Apart and Together,” Human Biology 79 (2007):
623–635.
3. Hanan reached menarche at thirteen years, two years later than her twin.
4. Susan L. Farber, Identical Twins Reared Apart: A Reanalysis (New York:
Basic Books, 1980).
5. Martha K. McClintock, “Menstrual Synchrony and Suppression,” Nature 229
(1971): 244–245. Some researchers have challenged these findings on method-
ological grounds; see H. Clyde Wilson, “A Critical Review of Menstrual Syn-
chrony Research,” Psychoneuroendocrinology 17 (1992): 565–591.
6. Mean age at menarche has remained stable for the past fifty years; see Angela
Diaz, Marc Laufer, and Lesley L. Breech, “Menstruation in Girls and Ado-
lescents: Using the Menstrual Cycle as a Vital Sign,” Pediatrics 118 (2006):
2245–2250.
384 N OT E S TO PAG E S 2 6 8 – 2 7 3

7. Jay Belsky, Laurence Steinberg, and Patricia Draper, “Childhood Experience,

Interpersonal Development, and Reproductive Strategy: An Evolutionary
Theory of Socialization,” Child Development 62 (1991): 647–670.
8. Segal and Stohs, “Resemblance for Age at Menarche.”
9. Federico J. Soriguer, Stella González-Romero, Isabel Esteva, et al., “Does the
Intake of Nuts and Seeds Alter the Appearance of Menarche?” Acta Obstet-
rica Gynecologia Scandinavica 74 (1995): 455–461.
10. Nancy L. Segal and Joanne Hoven Stohs, “Age at First Intercourse in Twins
Reared Apart: Genetic Influence and Life History Events,” Personality and
Individual Differences 47 (2009): 127–132.
11. Steven W. Gangestad and Randy Thornhill, “The Evolutionary Psychology
of Extrapair Sex: The Role of Fluctuating Asymmetry,” Evolution and Hu-
man Behavior 18 (1996): 69–88.
12. Anders P. Møller and Andrew Pomiankowski, “Fluctuating Asymmetry and
Sexual Selection,” Genetica 89 (1993): 267–279.
13. Randy Thornhill and Steven W. Gangestad, “The Scent of Symmetry: A Hu-
man Sex Pheromone That Signals Fitness?” Evolution and Human Behavior
20 (1999): 175–201.
14. Theresa Ann Markow and Irving I. Gottesman, “Fluctuating Dermatoglyphic
Asymmetry in Psychotic Twins,” Psychiatry Research 29 (1989): 37–43.
15. Fifty-four studies of human fluctuating asymmetry have been published in
Evolution and Human Behavior since 1997.
16. Steven W. Gangestad, personal communication, February 20, 2011. A study
linking increased fluctuating asymmetry and schizophrenia predisposition
was published in the mid-1980s; see Therese A. Markow and Kevin Wandler,
“Fluctuating Dermatoglyphic Asymmetry and the Genetics of Liability to
Schizophrenia,” Psychiatry Research 19 (1986): 323–328.
17. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A Study
of Heredity and Environment (Chicago: University of Chicago Press, 1937).
18. Ibid., 141.
19. Sonya J. Elder, Susan B. Roberts, Megan A. McCrory, et al., “Effect of Body
Composition Methodology on Heritability Estimates of Body Fatness” (2011,
under review).
20. The reliability of the anthropometric measurements was not assessed. Unreli-
ability would, however, lead to more conservative estimates of FA heritability.
21. Anders Paper Møller, “A Review of Developmental Instability, Parasitism
and Disease,” Infection, Genetics and Evolution 6 (2006): 133–140.
22. Gregory Livshits and Eugene Kobyliansky, “Study of Genetic Variance in the
Fluctuating Asymmetry of Anthropometrical Traits,” Annals of Human Biol-
ogy 16 (1989): 121–129.
23. Wendy Johnson, Steven W. Gangestad, Nancy L. Segal, and Thomas J. Bouchard
Jr., “Heritability of Fluctuating Asymmetry in a Human Twin Sample: The
Effect of Trait Aggregation,” American Journal of Human Biology 20 (2008):
651–658.
24. Stefan Van Dongen, “What Do We Know about the Heritability of Develop-
mental Stability? Answers from a Bayesian Model,” Evolution 61 (2007):
1022–1042, p. 1041.
 N OT E S TO PAG E S 2 7 3 – 2 8 3 385

25. Wendy Johnson, Nancy L. Segal, and Thomas J. Bouchard Jr., “Fluctuating
Asymmetry and General Intelligence: No Genetic or Phenotypic Associa-
tion,” Intelligence 36 (2008): 279–288.
26. Linda Mealey et al., “Symmetry and Perceived Facial Attractiveness: A
Monozygotic Co-Twin Comparison,” Journal of Personality and Social Psy-
chology 76 (1999): 157–165.
27. Linda Mealey, “Do Parents Show Favoritism for Their Symmetric Children?”
paper presented at the Human Behavior and Evolutionary Society, New
Brunswick, N.J., June 2002.
28. Thornhill and Gangestad, “Scent of Symmetry.”
29. Nancy L. Segal, Indivisible by Two: Lives of Extraordinary Twins (Cam-
bridge, Mass.: Harvard University Press, 2007).
30. Elizabeth Quinn, “What Is Hydrostatic Underwater Weighing?” About.com
Sports Medicine, updated March 30, 2011, http://sportsmedicine.about.com
/od/fitnessevalandassessment/g/UnderwaterWeigh.htm.
31. Mary E. Holmes, Being You (London: Austin & Macauley, 2008), 161–162.
32. Ibid., 165.
33. Ibid., 167–168.
34. Sonya J. Elder, Alice H. Lichtenstein, Anastassios G. Pittas, et al., “Genetic
and Environmental Influences on Factors Associated with Cardiovascular
Disease and the Metabolic Syndrome,” Journal of Lipid Research 50 (2009):
1917–1926.
35. Sonya J. Elder, Michael C. Neale, Paul J. Fuss, et al., “Genetic and Environ-
mental Influences on Eating Behavior—A Study of Twin Reared Apart,”
FASEB Journal 23 (2009): abstract 545.7. The published version of this re-
search is a conference abstract, so full details of the study are unavailable.
36. Albert J. Stunkard and Samuel Messick. “The Three-Factor Eating Question-
naire to Measure Dietary Restraint, Disinhibition and Hunger,” Journal of
Psychosomatic Research 29 (1985): 71–83.
37. Nancy L. Pedersen, Robert Plomin, Gerald E. McClearn, and Lars Friberg,
“Neuroticism, Extraversion and Related Traits in Adult Twins Reared
Apart,” Journal of Personality and Social Psychology 55 (1988): 950–957.
38. A 2003 reared-together study found heritabilities of 0.45 for disinhibition,
0.08 for hunger, and 0.00 for restraint. The different results from those of the
MISTRA could be due to the all-female sample and abbreviated and modified
version of the Eating Questionnaire. See Benjamin M. Neale, Suzanne E.
Mazzeo, and Cynthia M. Bulik, “A Twin Study of Dietary Restraint, Disinhibi-
tion and Hunger: An Examination of the Eating Inventory (Three Factor Eat-
ing Questionnaire),” Twin Research and Human Genetics 8 (2003): 471–478.
39. Longitudinal studies track physical and behavioral traits of the same individ-
uals continuously over time to assess age changes. Cross-sectional studies si-
multaneously compare physical and behavioral traits of two or more different
age groups to assess age differences.
40. Kenneth S. Kendler, Lisa J. Halberstadt, Frank Butera, John Myers, Thomas
J. Bouchard Jr., and P. Ekman, “The Similarity of Facial Expressions in Re-
sponse to Emotion-Inducing Films in Reared-Apart Twins,” Psychological
Medicine 38 (2008): 1475–1483.
386 N OT E S TO PAG E S 2 8 4 – 2 8 7

41. Segal, Indivisible by Two.

42. Oskar and Jack, directed by Frauke Sandig (New York: First Run/Icarus Films,
1996), VHS, was originally produced in Germany. A play by Mexican author
Andrés Roemer, Oskar y Jack (Mexico City: Miguel Ángel Porrúa, 2011),
based on “Oskar and Jack” in Segal, Indivisible by Two, has been written.
43. Thomas J. Bouchard Jr., “Genes and Human Psychological Traits,” in The
Innate Mind: Foundations for the Future, ed. Peter Carruthers, Stephen Lau-
rence, and Stephen Stich, vol. 3, 69–89 (Oxford: Oxford University Press,
2007). This chapter summarizes a great deal of MISTRA research on intelli-
gence, personality, and social attitudes.
44. Wendy Johnson, Jan te Nijenhuis, and Thomas J. Bouchard Jr., “Still Just 1 g:
Consistent Results from Five Test Batteries,” Intelligence 36 (2008): 81–95.
45. Wendy Johnson, Jan te Nijenhuis, and Thomas J. Bouchard Jr., “Replication
of the Hierarchical Visual-Perceptual-Image Rotation Model in de Wolff and
Buiten’s (1963) Battery of 46 Tests of Mental Ability,” Intelligence 35 (2007):
69–81.
46. Wendy Johnson, Thomas J. Bouchard Jr., Matt McGue, et al., “Genetic and
Environmental Influences on the Verbal-Perceptual-Image Rotation (VPR)
Model of the Structure of Mental Abilities in the Minnesota Study of Twins
Reared Apart,” Intelligence 35 (2007): 542–562.
47. In order to have unique subject data, only two members of two MZA triplet
sets and two DZA triplets were included, eliminating four MZA pairs and
four DZA opposite-sex pairs.
48. Lylas G. Mogk and Marja Mogk, Macular Degeneration: The Complete Guide
to Saving and Maximizing Your Sight (New York: Ballantine Books, 1999).
49. Wendy Johnson, Eric Turkheimer, Irving I. Gottesman, and Thomas J.
Bouchard Jr., “Beyond Heritability: Twin Studies in Behavioral Research,”
Current Directions in Psychological Science 19 (2009): 217–220; Wendy
Johnson, “Understanding the Genetics of Intelligence: Can Height Help? Can
Corn Oil?” Current Directions in Psychological Science 19 (2010): 177–182;
Wendy Johnson, Lars Penke, and Frank M. Spinath, “Heritability in the Era of
Molecular Genetics: Some Thoughts for Understanding Genetic Influences on
Behavioural Traits,” European Journal of Personality 25 (2011): 254–266.
50. Robert Plomin and John C. DeFries, “Genetics and Intelligence: Recent Data,”
Intelligence 4 (1980): 15–24; Nancy L. Segal and Jean Russell, “IQ Similarity
in Monozygotic and Dizygotic Twin Children: Effects of the Same versus Sepa-
rate Examiners: A Research Note,” Journal of Child Psychology and Psychia-
try 32 (1991): 703–708.
51. Irving I. Gottesman, “Biogenetics of Race and Class,” in Social Class, Race
and Psychological Development, ed. Martin Deutsch, Irwin Katz, and Ar-
thur R. Jensen, 11–51 (New York: Holt, Rinehart & Winston, 1968). The IQ
differences from Shields’s study were IQ equivalents of verbal and nonverbal
test scores. The MISTRA within-pair IQ scores excluded twins from outside
the United States.
52. Thomas J. Bouchard Jr., “Genetic Influence on Human Intelligence (Spear-
man’s g): How Much?” Annals of Human Biology 36 (2009): 527–544.
 N OT E S TO PAG E S 2 8 7– 2 9 2 387

53. Steven Pinker, The Blank Slate: The Modern Denial of Human Nature (New
York: Viking, 2002).
54. Wendy Johnson and Thomas J. Bouchard Jr., “Sex Differences in Mental
Abilities: g Masks the Dimensions on Which They Lie,” Intelligence 35
(2007a): 23–39.
55. Wendy Johnson and Thomas J. Bouchard Jr., “Sex Differences in Mental
Ability: A Proposed Means to Link Them to Brain Structure and Function,”
Intelligence 35 (2007b): 197–209.
56. Sally P. Springer and Georg Deutsh, Left Brain, Right Brain: Perspectives
from Cognitive Neuroscience (San Francisco: W. H. Freeman, 2001).
57. Elizabeth Hampson, “Variations in Sex-Related Differences across the Men-
strual Cycle,” Brain and Cognition 14 (1990): 26–43.
58. Richard J. Haier, Rex E. Jung, Ronald A. Yeo, Kevin Head, and Michael T.
Alkire, “Structural Brain Variation and General Intelligence,” Neuroimage
23 (2004): 425–433.
59. Johnson and Bouchard, “Sex Differences in Mental Abilities” (2007b).
60. Wendy Johnson, Rex E. Jung, Roberto Colom, and Richard J. Haier, “Cogni-
tive Abilities Independent of IQ Correlate with Regional Brain Structure,”
Intelligence 36 (2008): 18–28.
61. “Twins of Fate,” 48 Hours: Mystery, CBS Television, June 20, 2002.
62. Wendy Johnson and Thomas J. Bouchard Jr., “Linking Abilities, Interests and
Gender via Latent Class Analysis,” Journal of Career Assessment 17 (2009):
3–38.
63. These percentages refer to the subsets of individuals indicating these interests
and occupations.
64. Jennifer Connellan, “Sex Differences in Human Neonatal Social Perception,”
Infant and Behavior Development 23 (2000): 113–118.
65. David S. Moore and Scott P. Johnson, “Mental Rotation in Human Infants:
A Sex Difference,” Psychological Science 19 (2008): 1063–1066.
66. Laura N. Koenig and Thomas J. Bouchard Jr., “Genetic and Environmental In-
fluences on the Traditional Moral Values Triad—Authoritarianism, Conserva-
tism and Religiousness—as Assessed by Quantitative Behavior Genetic Meth-
ods,” in Where God and Science Meet: How Brain and Evolutionary Studies
Alter Our Understanding of Religion, vol. 1: Evolution, Genes, and the Reli-
gious Brain, ed. Patrick McNamara, 31–60 (Westport, Conn.: Praeger, 2006).
67. Thomas J. Bouchard Jr., “Authoritarianism, Religiousness, and Conserva-
tism: Is ‘Obedience to Authority’ the Explanation for Their Clustering, Uni-
versality and Evolution?” in The Biological Evolution of Religious Mind and
Behavior, ed. Eckart Voland and Wulf Schiefenhövel, 165–180 (Berlin:
Springer, 2009). The Traditional Moral Virtues Triad was first explicated in
Koenig and Bouchard’s 2006 book chapter (see previous note).
68. Michael Hout, Andrew Greeley, and Melissa J. Wilde, “The Demographic
Imperative in Religious Change in the United States,” American Journal of
Sociology 107 (2001): 468–500.
69. Thomas J. Bouchard Jr., Nancy L. Segal, Auke Tellegen, Matt McGue, Mar-
garet Keyes, and Robert Krueger, “Genetic Influence on Social Attitudes:
388 N OT E S TO PAG E S 2 9 2 – 2 9 4

Another Challenge to Psychology from Behavior Genetics,” in Behavior Ge-

netic Principles: Development, Personality and Psychopathology, ed. Lisa-
beth F. DiLalla, 89–104 (Washington, D.C.: American Psychological Associa-
tion, 2004).
70. Remus Ilies, Richard D. Arvey, and Thomas J. Bouchard Jr., “Darwinism,
Behavioral Genetics and Organizational Behavior: A Review and Agenda for
Future Research,” in “Darwinian Perspectives on Behavior in Organizations,”
special issue, Journal of Organizational Behavior 27 (2006): 121–141.
71. Thomas J. Bouchard Jr., “Strong Inference: A Strategy for Advancing Psycho-
logical Science,” in Experience and Development: A Festschrift in Honor of
Sandra Wood Scarr, ed. Kathleen McCartney and Richard Weinberg, 39–59
(London: Taylor & Francis, 2009).
72. John R. Platt, “Strong Inference,” Science 146 (1964): 347–353.
73. Nancy L. Segal, “Personality and Birth Order in Monozygotic Twins Adopted
Apart: A Test of Sulloway’s Theory,” Twin Research and Human Genetics 11
(2008): 103–107.
74. Frank J. Sulloway, “Birth Order, Sibling Competition, and Human Behavior,”
in Conceptual Challenges in Evolutionary Psychology: Innovative Research
Strategies, ed. Harmon R. Holcomb III, 39–84 (Dordrecht: Kluwer Academic,
2001).
75. Ibid.
76. Stanley Coren, “Twinning Is Associated with an Increased Risk of Left-
Handedness and Inverted Writing Posture,” Early Human Development 40
(1994): 23–17.
77. Nancy L. Sicotte, Roger P. Woods, and John C. Mazziota, “Handedness in
Twins: A Meta-Analysis,” Laterality 4 (1999): 265–286; Marian Annett, “Cere-
bral Asymmetry in Twins: Predictions of the Right Shift Theory,” Neuropsycholo-
gia 41 (2003): 469–479; Michele Carlier, Elisabeth Spitz, Marie Cécile Vacher-
Lavenu, Pierre Villéger, Benoît Martin, and François Michel, “Manual Performance
and Laterality in Twins of Known Chorion Type,” Behavior Genetics 26 (1996):
409–417; Charles E. Boklage, “Embryogenesis of Chimeras, Twins and Anterior
Midline Asymmetries,” Human Reproduction 21 (2003): 579–591.
78. Charles E. Boklage, How New Humans Are Made (Singapore: World Scien-
tific, 2010).
79. Sarah E. Medland, Margaret J. Wright, Gina M. Geffen, et al., “Special Twin
Environments, Genetic Influences and Their Effects on the Handedness of
Twins and Their Siblings,” Twin Research 6 (2003): 119–130.
80. R.  C. Oldfield, “The Assessment and Analysis of Handedness: The Edin-
burgh Inventory,” Neuropsychologia 9 (1971): 97–113.
81. Nancy L. Segal and Thomas J. Bouchard Jr., “Hand-Discordance in MZ and
DZ Twins Reared Apart and Together,” paper presented at the III American
Psychological Society, Washington, D.C., June 13–16, 1991
82. Herbert F. Crovitz and Karl Zener, “A Group-Test for Assessing Hand- and
Eye-Dominance,” American Journal of Psychology 75 (1962): 271–276.
83. [χ2 (df = 1) = 1.33, not statistically significant]. MZA female members in two
MZA/DZA triplet sets were included in the individual twin proportions.
84. [χ2 (df = 1) = 1.15, not statistically significant].
 N OT E S TO PAG E S 2 9 5 – 3 0 1 389

85. James Shields, Monozygotic Twins Brought Up Apart and Brought Up

Together (London: Oxford University Press, 1962); Niels Juel-Nielsen, “A
Psychiatric-Psychological Investigation of Monozygous Twins Reared Apart,”
Acta Psychiatrica et Neurologica Scandinavica (1965), Monograph Supple-
ment 183. Handedness data were available for eleven of Juel-Nielsen’s twelve
MZA twin pairs. Twins in the two hand-discordant pairs were described as
“right-handed” and “ambidextrous, probably originally left-handed” (p. 247)
and “both ambidextrous”, but one “predominantly left-handed” (p.  177).
Both twins in a third MZA twin pair were described as “right-handed” (p. 77),
although one twin’s early tendency toward left-handedness was “quickly cor-
rected” by her teacher and she always “felt naturally right-handed” (p. 74).
86. James Shields, Monozygotic Twins (1962). Shields administered Blau’s Ques-
tionnaire for Preferred Laterality to twins who displayed left-handedness,
based on their responses to items in his interview booklet.
87. David J. Merrell, “Dominance of Eye and Hand,” Human Biology 29 (1957):
314–327, p. 325.
88. Chris McManus, “The Inheritance of Left-Handedness,” in Biological Asym-
metry and Handedness, ed. Gregory R. Bock and Joan Marsh, 251–281
(West Sussex, U.K.: CIBA Foundation, 1991).
89. Wendy Johnson and Thomas J. Bouchard Jr., “The MISTRA Data: Forty-
Two Mental Ability Tests in Three Batteries,” Intelligence 39 (2011): 82–88.
90. Ibid., 83.
91. Jason T. Major, Wendy Johnson, and Thomas J. Bouchard Jr., “The Depend-
ability of the General Factor of Intelligence: Why Small, Single-Factor Mod-
els Do Not Adequately Represent g,” Intelligence 39 (2011): 418–433.
92. See, for example, Aja L. Murray, Hayley Dixon, Wendy Johnson, and Thomas
J. Bouchard Jr., “Spearman’s Law of Diminishing Returns: A Statistical Arti-
fact?” (under review); Thomas J. Bouchard Jr., “Heritability,” in Encyclope-
dia of the Mind, ed. Hal Pashler (Thousand Oaks, Calif., in press).

14. Questions, Answers, and Twin Studies of the Future

1. Brian Palmer “Double Inanity: Twin Studies Are Pretty Much Useless,” Slate
Magazine, http://www.slate.com/articles/life/twins/2011/08/double_inanity
.html (8/24/2011); see the response by Nancy L. Segal, Irving I. Gottesman,
Nicholas G. Martin, Eric Turkheimer, and Margaret Gatz, reprinted in
Nancy L. Segal, “The Value of Twin Studies: A Response to Slate Magazine,”
Twin Research and Human Genetics 14 (2011): 593–597.
2. Thomas J. Bouchard Jr., “The Hereditarian Research Program: Triumphs and
Tribulations,” in Arthur Jensen: Consensus and Controversy, ed. Sohan Modgil
and Celia A. L. Modgil, 55–70 (London: Falmer International, 1987).
3. Nancy L. Segal and Jean Russell, “IQ Similarity in Monozygotic and Dizygotic
Twin Children: Effects of the Same versus Separate Examiners: A Research
Note,” Journal of Child Psychology and Psychiatry 32 (1991): 703–708.
4. Thomas J. Bouchard Jr. and Matthew McGue, “Familial Studies of Intelligence:
A Review,” Science 212 (1981): 1055–1059; Thomas J. Bouchard Jr., “Gene-
tic Influence on Human Intelligence (Spearman’s g): How Much?” Annals of
390 N OT E S TO PAG E S 3 0 1 – 3 0 5

Human Biology 36 (2009): 527–544; Nancy L. Segal, “Twins: The Finest Natu-
ral Experiment,” Personality and Individual Differences 49 (2010): 317–323.
5. Matt McGue, Thomas J. Bouchard Jr., William G. Iacono, and David T.
Lykken, “Behavioral Genetics of Cognitive Ability: A Life-Span Perspective,”
in Nature, Nurture and Psychology, ed. Robert Plomin and Gerald E. Mc-
Clearn, 59–76 (Washington, D.C., American Psychological Association, 1993).
6. Shields’s January 24, 1974, letter addressed to Dr. E. B. Hook of the New York
State Birth Defects Institute was provided to me by Irving I. Gottesman.
7. Wendy Johnson and Thomas J. Bouchard Jr., “Sex Differences in Mental
Abilities: g Masks the Dimensions on Which They Lie,” Intelligence 35
(2007): 23–39.
8. Wendy Johnson, Thomas J. Bouchard Jr., Nancy L. Segal, and Jay Samuels,
“General Intelligence and Reading Performance in Adults: Is the Genetic
Factor Structure the Same as for Children?” Personality and Individual Dif-
ferences 38 (2005): 1413–1428.
9. David T. Lykken, Seymour Geisser, and Auke Tellegen, “Heritability Esti-
mates from Twin Studies: The Efficiency of the MZA Design” (unpublished
manuscript, Department of Psychology, University of Minnesota, Minneapo-
lis, 1981).
10. David T. Lykken, “A More Accurate Estimate of Heritability,” Twin Research
and Human Genetics 10 (2007): 168–173.
11. Horatio H. Newman, Frank N. Freeman, and Karl J. Holzinger, Twins: A
Study of Heredity and Environment (Chicago: University of Chicago Press,
1937), 132–133.
12. Steven Rose, Richard C. Lewontin, and Leon J. Kamin, Not in Our Genes:
Biology, Ideology, and Human Nature (New York: Pantheon, 1984); Jay
Joseph, “Separated Twins and the Genetics of Personality Differences: A
Critique,” American Journal of Psychology 114 (2001): 1–30; Jay Joseph,
The Gene Illusion: Genetic Research in Psychiatry ad Psychology under the
Microscope (New York: Algora, 2004).
13. Iva C. Gardner and Horatio H. Newman, “Mental and Physical Traits of
Identical Twins Reared Apart: Case XX. Twins Lois and Louise,” Journal of
Heredity 31 (1940): 119–126, p. 19.
14. Niels Juel-Nielsen, “A Psychiatric-Psychological Investigation of Monozy-
gous Twins Reared Apart,” Acta Psychiatrica et Neurologica Scandinavica
(1965), Monograph Supplement 183.
15. James Shields, Monozygotic Twins Brought Up Apart and Brought Up
Together (London: Oxford University Press, 1962).
16. Our twins were adults, so most were probably conceived before the avail-
ability of assisted reproduction technology. Since the early 1980s, these tech-
niques have substantially raised the DZ twinning rate and slightly elevated
the MZ twinning rate.
17. The similarities of sixteen MZA twin pairs and one DZA twin pair were
detected by others, leading to their reunion. Physical resemblance reunited
twins in another case, in which 64-year-old MZ female twins simply recog-
nized each other; each had always known she had a twin. More recently,
 N OT E S TO PAG E S 3 0 5 – 3 0 8 391

separated 35-year-old Argentine MZ female twins met by chance in a cloth-

ing store. The twins were previously unaware of their twinship; “Argentine
Twins Separated at Birth Reunite 35 Years Later,” http://www.20minutos.es
/noticia/460527/0/gemelas/separadas/encuentro/, April 1, 2009. These twins
found each other ten years after the MISTRA ended.
18. Kenneth S. Kendler, Michael C. Neale, Ronald C. Kessler, Andrew C. Heath,
and Lindon J. Eaves, “A Test of the Equal-Environment Assumption in Twin
Studies of Psychiatric Illness,” Behavior Genetics 1 (1993): 21–27; Sandra
Scarr and Louise Carter-Saltzman, “Twin Method: Defense of a Critical As-
sumption,” Behavior Genetics 9 (1979): 527–542.
19. Robert Plomin, John C. DeFries, Gerald E. McClearn, and Peter McGuffin,
Behavioral Genetics, 5th ed. (New York: Worth Publishers, 2008).
20. John C. Loehlin and Robert C. Nichols, Heredity, Environment, and Personal-
ity: A Study of 850 Sets of Twins (Austin: University of Texas Press, 1976);
Robert Plomin, Lee Willerman, and John C. Loehlin, “Resemblance in Appear-
ance and the Equal Environments Assumption in Twin Studies of Personality
Traits,” Behavior Genetics 6 (1976): 43–52; Nikole J. Cronk, Wendy S. Slutske,
Pamela A. Madden, et al., “Emotional and Behavioral Problems among Female
Twins: An Evaluation of the Equal Environments Assumption,” Journal of the
American Academy of Child and Adolescent Psychiatry 41 (2002): 829–837.
21. Robert Bazell, “Sins and Twins,” New Republic, December 21, 1987, 17–18.
22. Lisa Lynch, “Twins’ Personality Study Made Big Splash before Journal Ap-
proval,” Science Writer, Winter 1990, 15–19.
23. Thomas J. Bouchard Jr., “Intensive, Detailed, Exhaustive,” Intelligence 26
(1998): 283–290, p. 288.
24. Bouchard has turned down media invitations since his retirement, but he
forwards all of them to me.
25. Dorothy Nelkin, “An Uneasy Relationship: The Tensions between Medicine
and the Media,” Lancet 347 (1996): 160–1603, p. 1601.
26. Ibid., 1601.
27. Edwin Chen, “Twins Reared Apart: A Living Lab,” New York Times Maga-
zine, December 9, 1979, 112–123.
28. Constance Holden, “Twins Reunited: More Than the Faces Are Familiar,”
Science 80 (1980): 55–59.
29. Lewis Cope, “Twins: Why Are They Alike?” Minneapolis Star Tribune: Pic-
ture Magazine, February 13, 1983, 14–18.
30. Lawrence Wright, “Double Mystery,” New Yorker, August 7, 1995, 45–46,
48–52, 54–62.
31. Jill Neimark, “Nature’s Clones,” Psychology Today, August 1997, 37–45,
64–69.
32. Frank Rich, “Separated at Birth” New York Times, January 12, 1995.
33. Thomas J. Bouchard Jr., “Ideological Obstacles to Genetic Research: Tales
from the Nature-Nurture Wars,” paper presented to “Science and the
Powers” seminar, Swedish Ministry of Education and Science, Stockholm,
Sweden, March 16, 1994.
34. Matt McGue, letter to the editor, Scientific American 269 (1993): 8.
392 N OT E S TO PAG E S 3 0 8 – 3 1 6

35. Thomas J. Bouchard Jr., “Ideological Obstacles,” 1994.

36. Val Dusek, “Bewitching Science,” Science for the People 19 (November/
December 1987), 19–22.
37. John Horgan, “Trends in Behavioral Genetics: Eugenics Revisited,” Scientific
American 268 (1993): 122–131.
38. Dusek, “Bewitching Science.”
39. William H. Tucker, “Burt’s Separated Twins: The Larger Picture,” Journal of
the History of the Behavioral Sciences 43 (2007): 81–86, p. 85.
40. This figure was calculated in 1990. The cost of assessing one twin pair with-
out spouses was $4,280. Participant costs varied due to changing hotel rates,
airline fees, and other factors, explaining why the figures vary across grants.
41. “About the National Science Foundation,” U.S. National Science Founda-
tion, updated August 29, 2011, www.nsf.gov/about/.
42. Leon Festinger, Henry W. Riecken, and Stanley Schachter, When Prophecy
Fails: A Social and Psychological Study of a Modern Group That Predicted
the Destruction of the World (New York: Harper-Torchbooks, 1964; first
published in 1956).
43. Thomas J. Bouchard Jr., letter to Richard Atkinson, April 16, 1979.
44. Proposals are rated as excellent, very good, good, fair, or poor.
45. Richard Lynn, The Science of Human Diversity: A History of the Pioneer
Fund (Lanham, Md.: University Press of America, 2001).
46. Leila Zenderland, Measuring Minds: Henry Herbert Goddard and the Ori-
gins of American Intelligence Testing (Cambridge, Mass.: Cambridge Uni-
versity Press, 1998).
47. Harry F. Weyher, “The Pioneer Fund, the Behavioral Sciences, and the Me-
dia’s False Stories,” Intelligence 26 (1998): 319–336.
48. Paul A. Lombardo, “ ‘The American Breed’: Nazi Eugenics and the Origins of
the Pioneer Fund,” Albany Law Review 65 (2002): 743–829; Paul A. Lom-
bardo, “Pioneer’s Big Lie,” Albany Law Review 65 (2003): 1125–1143;
J. Philippe Rushton, “The Pioneer Fund and the Scientific Study of Human
Differences,” Albany Law Review 66 (2002): 207–262.
49. Ibid. The article was Arthur R. Jensen, “How Much Can We Boost IQ and
Scholastic Achievement?” Harvard Educational Review 39 (1969): 1–123.
50. Lynn, Science of Human Diversity.
51. Harry F. Weyher, “The Pioneer Fund, the Behavioral Sciences, and the Me-
dia’s False Stories,” Intelligence 26 (1999): 319–336.
52. “Pioneer Fund,” The Southern Poverty Law Center, 2011, www.splcenter
.org/get-informed/intelligence-files/groups/pioneer-fund.
53. Irving I. Gottesman, interview, October 15, 2009.
54. Television producers only filmed reenactments of activities twins had already
completed during the study week.
55. Thomas J. Bouchard Jr., Leonard Heston, Elke Eckert, Margaret Keyes, and
Susan Resnick, “The Minnesota Study of Twins Reared Apart: Project De-
scription and Sample Results in the Development Domain,” in Twin Re-
search, vol. 3, Intelligence, Personality, and Development, ed. Luigi Gedda,
Paolo Parisi, and Walter E. Nance, 227–233 (New York: Alan R. Liss, 1981).
 N OT E S TO PAG E S 3 1 7– 32 2 393

56. See the preface to: Morton Hunt, The New Know-Nothings: The Political
Foes of the Scientific Study of Human Nature (New Brunswick, N.J.: Trans-
action, 1999).
57. Koch Industries is the second-largest closely held company in the United
States.
58. David H. Koch is the second wealthiest resident of New York, after Mayor
Michael Bloomberg, per Wikipedia.
59. Koch donated money to New York’s Lincoln Center for renovation of the
opera and ballet theater that now bears his name.
60. Richard B. Call passed away on March 11, 2011. The current President of
the Seaver Institute is Victoria Seaver Dean.
61. Richard Lynn, personal communication, August 4, 2010.
62. The origins of general intelligence are still debated, but many within the sci-
entific community acknowledge significant genetic effects. See Ian J. Deary,
Frank M. Spinath, and Timothy C. Bates, “Genetics of Intelligence,” Euro-
pean Journal of Human Genetics 14 (2006): 690–700, and references therein.
63. Plomin et al., Behavioral Genetics.
64. Avshalom Caspi, Alan Taylor, Terrie E. Moffit, and Robert Plomin, “Neigh-
borhood Deprivation Affects Children’s Mental Health: Environmental Risks
Identified in a Genetic Design,” Psychological Science 11 (2000): 338–342.
65. Gardner and Newman, “Mental and Physical Traits,” p. 126.
66. Ibid.
67. Wendy Johnson, Eric Turkheimer, Irving I. Gottesman, and Thomas J.
Bouchard Jr., “Beyond Heritability: Twin Studies in Behavioral Research,”
Current Directions in Psychological Science 18 (2009): 217–220.
68. Robert Plomin, “Genotype-Environment Correlation and Interaction,” pa-
per presented at the Festschrift Honoring John Loehlin, Newport, Rhode
Island, June 6, 2011.
69. Niels G. Waller and Phillip R. Shaver, “The Importance of Nongenetic Influ-
ences on Romantic Love Styles: A Twin-Family Study,” Psychological Science
5 (1994): 268–274.
70. Justine L. Giddens, Julie Aitken Schermer, and Philip A. Vernon, “Material
Values Are Largely in the Family: A Twin Study of Genetic and Environmen-
tal Contributions to Materialism,” Personality and Individual Differences 46
(2009): 428–431.
71. Superfecundated twins can arise when a woman releases two eggs simultane-
ously and has different sexual partners within a few days of one another.
72. Carol E. Franz, Timothy P. York, Lindon J. Eaves, et al., “Adult Romantic
Attachment, Negative Emotionality, and Depressive Symptoms in Middle
Aged Men: A Multivariate Genetic Analysis,” Behavior Genetics 41 (2011):
488–498; Nicola L. Barclay, Thalia C. Eley, Daniel J. Buysse, et al., “Non-
shared Environmental Influences on Sleep Quality: A Study of Monozygotic
Twin Differences,” Behavior Genetics DOI: 10.1007/s10519-011-9510-1,
October 15, 2011, online; Amélie Petitclerc, Michel Boivin, Ginette Dionne,
Daniel Pérusse, and Richard Tremblay, “Genetic and Environmental Etiology
of Disregard for Rules,” Behavior Genetics 41 (2011): 192–200.
394 N OT E S TO PAG E S 32 2 – 32 5

73. Plomin et al., Behavioral Genetics.

74. Sarah E. Harris, Helen Fox, Alan F. Wright, et al., “A Genetic Association
Analysis of Cognitive Ability and Cognitive Ageing Using 325 Markers for
109 Genes Associated with Oxidative Stress or Cognition,” BMC Genetics 8
(2007), doi:10.1186/1471-2156-8-43.
75. “Genome-wide Association Studies,” National Human Genome Research
Institute, last update August 17, 2010, www.genome.gov/20019523; Inter-
national HapMap Project, http://hapmap.ncbi.nlm.nih.gov/.
76. Matt McGue, “The End of Behavioral Genetics?” Behavior Genetics 40 (2010):
284–296.
77. Dorret I. Boomsma, “A Twin Year in Review: Behavior and Psychiatry,”
paper presented at the 13th International Congress of Twin Studies, Seoul,
South Korea, June 4–7, 2010.
78. Peter M. Visscher, William G. Hill, and Naomi R. Wray, “Heritability in the
Genomics Era—Concepts and Misconceptions,” Nature Reviews (Genetics)
9 (2008): 255–266.
79. “Studies of European Volunteer Twins to Identify Genes Underlying Common
Diseases,” GenomEUtwin, last update June 7, 2006, www.genomeutwin.org.
80. U.S. Department of Energy Genome Programs, “SNP Fact Sheet,” Human
Genome Project Information, last updated September 19, 2008, www.ornl
.gov/sci/techresources/Human_Genome/faq/snps.shtml.
81. Ciara Curtin, “The Undiscovered Variants,” Genome Web: Genome Tech-
nology, March 2011, www.genomeweb.com/arrays/undiscovered-variants
?page=4.
82. Gail Davies, Albert Tenesa, Antony Payton, et al., “Genome-wide Associa-
tion Studies Establish That Human Intelligence Is Highly Heritable and
Polygenic,” Molecular Psychiatry 16 (2011): 996–1005.
83. Eric Turkheimer, “Genome Wide Association Studies of Behavior Are Social
Science,” in Philosophy of Behavioral Biology, ed. Katie S. Plaisance and
Thomas A. C. Reydon, chapter 3 (Dordrecht, the Netherlands: Springer, 2011).
84. Mario F. Fraga, Esteban Ballestar, Maria F. Paz, et al., “Epigenetic Differ-
ences Arise during the Lifetime of Monozygotic Twins,” Proceedings of the
National Academy of Sciences of the United States of America 102 (2005):
1413–1414.
85. F. Nipa Haque, Irving I. Gottesman, and Albert H.  C. Wong, “Not Really
Identical: Epigenetic Differences in Monozygotic Twins and Implications for
Twin Studies in Psychiatry,” American Journal of Medical Genetics, Part C:
Seminars in Medical Genetics 151 (2009): 136–141.
86. Andrew P. Feinberg, “Phenotypic Plasticity and the Epigenetics of Human
Disease,” Nature 447 (2007): 443–440.
87. Carl E. G. Bruder, Arkadiusz Piotrowski, Antoinet A. C. J. Gijsbers, et al.,
“Phenotypically Concordant and Discordant Monozygotic Twins Display
Different DNA Copy-Number-Variation Profiles,” American Journal of Hu-
man Genetics 82 (2008): 763–771.
88. Yong-Kyu Kim, Handbook of Behavior Genetics (New York: Springer,
2009).
 N OT E S TO PAG E S 32 5 – 32 7 395

89. Irving I. Gottesman and Todd D. Gould, “The Endophenotype Concept

in  Psychiatry: Etymology and Strategic Intentions,” American Journal of
Psychiatry 160 (2003): 636–645.
90. Raymond C. K. Chan, Irving I. Gottesman, Xiaojia Ge, and Pak C. Sham,
“Strategies for the Study of Neuropsychiatric Disorders Using Endopheno-
types in Developing Countries: A Potential Databank from China,” Frontiers
of Human Neuroscience 4 {2010}, doi:10.3389/fnhum.2010.00207.
91. Nancy L. Segal, “New Sources of Reared Apart Twins,” Twin Research and
Human Genetics 10 (2007): 786–790; Nancy L. Segal, Someone Else’s Twin:
The True Story of Babies Switched at Birth (Amherst, N.Y.: Prometheus
Books, 2011).
92. Zdravka Veleva, Sirpa Vilska, Christel Hydén-Granskog, Aila Tiitinen, Juha
S. Tapanainen, and Hannu Martikainen, “Elective Single Embryo Transfer in
Women Aged 36–39 Years,” Human Reproduction 21 (2006): 2098–2102.
93. The quote from David Lykken was provided by his colleague Robert Krueger,
July 8, 2010; also see Robert F. Krueger, Susan South, Wendy Johnson, et al.,
“The Heritability of Personality Is Not Always 50%: Gene-Environment In-
teractions and Correlations between Personality and Parenting,” Journal of
Personality 76 (2008): 1485–1521.
94. Michel Duyme, Annick-Camille Dumaret, and S. Stanislaw Tomkiewcz.
(1999), “How Can We Boost IQs of ‘Dull Children’? A Late Adoption Study,”
Proceedings of the National Academy of Sciences 96 (1999): 8790–8794.
95. Laura E. Berk, Child Development, 8th ed. (Boston: Allyn & Bacon, 2008).
96. Thomas J. Bouchard Jr., “Whenever the Twain Shall Meet,” The Sciences 37
(1997): 52–57, p. 57.
Acknowledgments

The Minnesota Study of Twins Reared Apart (MISTRA) was a collaborative ef-
fort, and in many ways writing this book was, too. Nearly every former investi-
gator contributed his or her perspectives, insights, and recollections, and many
provided data, documents, and photographs. However, I take responsibility for
putting it all together. A comprehensive work on such a controversial study can-
not please everyone, but I have tried to present a just view.
Writing Born Together—Reared Apart let me reconnect with former colleagues
as well as retrace my steps through the old but familiar twin assessment labora-
tories. Working on this book also put me in touch with many of the reared-apart
twins whom I came to know so well during their weeks in Minneapolis. I never
tire of hearing their life stories that, in my view, offer the best blend of scientific
insight and human interest.
Writing this book was also an opportunity to assess the significance of the
MISTRA in light of current themes and trends in psychology and medicine. The
study withstood the controversies of the past, and, in fact, its findings have been
instrumental in the widespread acceptance of genetic influences on behavior and
health. Future studies of adult reared-apart twins will follow a path different
from ours in view of the molecular genetic advances I discuss in the book. Many
current and future researchers would not be content to disentangle genetic and
environmental influences as we did with inventories, tests, and questionnaires,
but would attempt to tie the MZA twins’ similarities to specific genes and their
differences to epigenetic events. But they would be remiss not to study the twins’
behaviors because, as Bouchard once said, their similarities and differences are “so
up front and personal.” Regardless, the MISTRA continues to inspire research
across many domains of human functioning and promises to do so in the future,
as evidenced by a number of planned and ongoing analyses of our data.
Thanking the many people who provided assistance and encouragement is
such an enjoyable part of writing a book. The core collaborators, Elke D. Eckert,
398 ACKNOWLEDGMENTS

Irving I. Gottesman, Leonard L. Heston, and Auke Tellegen, were as generous

with their time and intelligence as they always were. Each was a compelling pres-
ence in this book because his or her expertise was there from the start. Gottes-
man also read the manuscript in full prior to publication, offering his usual pen-
etrating insights and brilliant suggestions. He generously forwarded important
articles, personal correspondence, and other fascinating material that I incorpo-
rated into the book.
I regret that David Lykken was not around to contribute his remarks and rec-
ollections of the MISTRA. He was a commanding presence, and I sense his ab-
sence from the chapters on psychophysiology, emergenesis, and the roots of hu-
man happiness. I also regret that I never got to meet again with Bill Fox who ran
the psychomotor study or Bill Knobloch who tested the twins’ eyes, colleagues
whose passing came too soon.
I am grateful to all the MISTRA investigators and former students, most of
whom met or spoke with me to reflect on the impact of their findings and the
significance of the MISTRA. At my urgings, many of them read and reread rele-
vant passages from this book so that the text would be exactly right, and they did
so with interest and patience. Some even forwarded unsolicited messages at later
dates with ideas and thoughts for me to consider. Thank you Dorothy Aeppli,
Richard Arvey, Deborah Betsworth, Malcolm Blumenthal, Joseph Bohlen, JoAnn
Boraas, Greg Carey, John Conry, Germaine Cornélissen-Guillaume, Katie Cor-
son, Michael Crawford, David DiLalla, Maurice Dysken, Paul Ekman, Sonya El-
der, John Ficken, Wally Friesen, Paul Fuss, Francie Gabbay, Steven Gangestad,
Jeff Gilger, Will Grove, Franz Halberg, Michael Hammer, Daniel Hankins, Jo-Ida
Hansen, Bruce Hanson, Kevin Haroian, Scott Hershberger, Mark Herzberg, Jim
Hinrichs, Yoon-Mi Hur, Bill Iacono, Jim Jenkins, Wendy Johnson, Lauren Keller,
Margaret Keyes, Laura Koenig, Peter Kohler, Brian Kojetin, Robert Krueger,
Scott Lilienfeld, Kristian Markon, Nick Martin, Matt McGue, Louise Messer,
Bryan Michalowicz, Dan Moloney, Mary Moster, Denise Newman, Bruce Pihl-
strom, Susan Resnick, Stephen Rich, Victor Rivera, Susan Roberts, Beatrice
Roitman-Johnson, Jay Samuels, Margaret Sanderson, Lloyd Sines, Michael Till,
Naip Tuna, Niels Waller, Kimerly Wilcox, Larry Wolff, and Dewey Ziegler. Greg
Carey and Eric Turkheimer generously reviewed selected sections of the manu-
script, offering wise counsel as they always do. Many colleagues contributed to one
or two publications over the course of the study, and scores of graduate and under-
graduate students assisted in collecting and processing data; students are credited
in individual papers.
Several anonymous reviewers provided me with wonderfully comprehensive
sets of suggestions following submission of my book proposal and subsequent
drafts. I discovered that one of these early reviewers was John Loehlin, whose
comments were invaluable. John graciously took a second look at portions of the
manuscript at a later date.
Small acts of kindness are not forgotten. Both Len Heston and Tom Bouchard
pared down their personal libraries just as I was beginning to write, bequeathing
many of the sources and photos that I used in this book. Len’s copies of the three
original reared-apart twin studies and Tom’s collections of some participants’
 ACKNOWLEDGMENTS 399

pictures and tapes were, and always will be, treasured items. The University of
Minnesota Alumni Association helped me locate that wonderful trick photo of
Bouchard sitting simultaneously behind and in front of his desk. Professor Gail
Peterson made sure that I received a CD of Bouchard’s last classroom lecture.
Other individuals provided invaluable assistance from time to time. They in-
clude Don Fowles, David Lubinski, Mary Mount, Antonín Pařícek, Nancy Peder-
sen, Laura Pérgola, Viktória Sas, and Tony Vernon. J. Philippe Rushton verified
several dates for the MISTRA’s receipt of Pioneer Funds, and Richard Lynn pro-
vided information about the Whitfield Institute’s support.
My editor at Harvard University Press, Elizabeth Knoll, was as smart, encour-
aging, and good-humored as she was during her supervision of my previous book,
Indivisible by Two. She is the kind of editor one dreams about. The other staff
members at Harvard University Press, namely, Matthew Hills, Margaux Leonard,
Joy Deng, and Anne Zarrella, were unfailingly helpful and gracious. The copy edi-
tor, Vickie West, and the production editor, Barbara Goodhouse at Westchester
Book Group, were helpful in every way possible. Special thanks go to my friend
and colleague Lauren Gonzalez, whose literary brilliance brought polish and clar-
ity to the text.
California State University–Fullerton granted me the time and facilities I needed
for the successful completion of this book. The psychology department secretaries,
Frances Sanchez and Amanda Hayes, printed many copies of the fifteen chapters
and two appendices for my review. Kelly Donovan (also an identical twin) worked
her usual artistic magic in preparing the photographs and figures as she has done for
me in the past. Several students, namely, Ammar Altowaiji, Shiloh Betterley, Amanda
Killian, and Gayle Dow, helped me obtain research articles and other materials.
Jaimee Munson, Jamie Graham, and Jorge Torres proofed the many tables, charts,
and references with patience and efficiency.
My parents, to whom this book is dedicated, were not here to enjoy my writ-
ing of Born Together—Reared Apart, the one hurtful part of this process. They
were thrilled when I joined the Minnesota Study of Twins Reared Apart and, as
parents of twins, were always eager to learn what we were finding. My dad,
Al Segal, fully supportive of his daughters’ work responsibilities, was understand-
ing when I cut short a home visit to New York just before Father’s Day to test a
new reared-apart pair in Minneapolis. My mom, Esther Segal, always a proud
mother of twins, read nearly every twin-related paper I wrote. My fraternal twin
sister, Anne, is partly—if not solely—responsible for why I do what I do. She has
always been a great supporter of my interests and goals, despite our genetically
based behavioral differences.
My boyfriend, philosophy professor Craig K. Ihara, weathered some lonely
times during my writing and travels. He also reminded me of the other passions
in my life—being with him and being his swing dance partner—and he made sure
I did not forget them. Craig is also credited for suggesting the idea for this book
and its title (the subtitle was mine!), and for reviewing the penultimate version of
the manuscript.
Tom Bouchard’s wife, Pauline, and their two children, Mark and Elizabeth,
were always enthusiastic and supportive of my efforts. They knew better than
400 ACKNOWLEDGMENTS

anyone how much this book needed to be written—I believe his family met more
twins than I did.
And what can I say about Tom Bouchard? That he was brilliant in setting up
the study and in keeping it going? That the twins adored him? That he was the
best mentor a postdoc could have? That he forever changed the way people think
about the roots of human behavior? I could say all these things and more, and it
would not come close to capturing the scientist and person that he is. Bouchard
is a steadfast reductionist, but ironically, just by being himself he has shown that
sometimes you just can’t be.
Index

Abraham (Keller), L. M., 116, 119 Augustine, Saint, 191

Addams, C., 79–80 Authoritarianism, 103, 215, 230–233, 242,
Additive genetic effects, 61–63, 75–76, 99, 253–255, 291, 377n43
102–103, 123, 173, 197, 205–206, 225,
347n40, 366n8 Barron, F., 237–238; Barron-Welsh Art
Adorno, T. W., 230 Scale, 238
Age-sex correction, 44, 62, 350n43, Behavioral Genetics (text), 141–142, 173,
353n45 216–217, 236
Aging, 27, 191–194, 318, 322 Behavior genetics, 9, 42, 58, 70, 100,
Agreeableness. See Big Five personality 133, 135, 148, 189, 214, 216–217, 233,
traits 265, 319, 381n32; and evolutionary
Allele, 56–58, 61, 183 psychology, 111–112, 181–182,
Allen, Garland, 144 184–186, 214–216, 247, 267,
Allen, Gordon, 2, 13, 339n9 382–383n47
Allergies, 16, 26, 87, 162–163, 165, 219, Behavior Genetics ( journal), 208, 322
227, 364n22; asthma, 78, 162–164, Behavior Genetics Association, 7, 70,
364n22; antibodies, 87, 162–163, 72, 95, 182, 204
351nn17–18; seasonal rhinitis, 162–164; Behaviorist perspective, 4, 7, 239
radioallergosorbent test (RAST), 163 Bernard, V., 6, 49
Allport, G., 143, 234 Betsworth, D. G., 208–210
Altemeyer, R., 215, 231, 254 Big Five personality traits, 187–188,
Anthropometric assessment, 39, 73, 93, 214, 223; agreeableness, 187–188;
270–273, 278, 300 conscientiousness, 187–188; extraver-
Arey, L. B., 2 sion, 187–188; neuroticism, 187–188;
Arvey, R., 116, 118–119, 179, 181, 209, openness, 187–188, 224–225; and ego
211–212, 214–215 development, 223–225
Assessment schedule, 8, 13, 15, 22–24, Binet IQ, 60, 107, 261, 356n38
54–55, 63, 65, 74–75, 95–96, 99, 117, Biometrical modeling, 58, 62, 99, 125.
126, 128, 154 See also Model fitting
Association analysis, 322 Birth order, 90, 292–293. See also
Assortative mating, 25, 44, 57, 62, 222, Homosexuality
261, 272, 296, 377n42 Blumenthal, M., 26, 163, 364n22
402 INDEX

Body composition, nutrition, and energy Candidate genes, 322

expenditure (Tufts University), 191, Cardiac studies, 22–23, 41, 82, 122;
218–219, 271, 274–282, 296; BOD electrocardiogram (ECG), 22, 26,
POD, 277 123–125, 299; circadian system, 85–86,
Bohlen, J., 91 222; Holter monitor, 85–86; exercise
Boomsma, D., 323 stress test, 123; vectorcardiogram,
Boraas, J., 96, 153–156, 162 123–124; echocardiogram, 123, 219;
Bouchard, C., 229 cardiovascular disease, 280, 282.
Bouchard, T. J. Jr., 9, 19; twin method, See also Body composition, nutrition,
2, 340n17; photographs, 5, 46, 48, and energy expenditure (Tufts University)
192–194; beginnings of the study, 7–13, Carey, G., 205–206, 241, 317
15, 18–25, 27–30, 343n69, 344nn5,6,21; Carroll three-strata model, 262
participants, 33–35, 49, 96, 151, 170, Cattell-Horn fluid-crystallized model,
172, 218, 227, 278, 325, 345n2; study 262–263
procedures, 40, 44, 54–55, 62, 73, 88, 95, Chomsky, N., 8
322; early findings, 65–66, 68, 80, 93, Chromosomes, 7, 56–57, 137, 172, 322;
344n12; teaching, 68, 70, 301; personal X chromosome, 2, 122
harassment, 70, 348n10; Juel-Nielsen Claridge, G., 77
interview, 79, 297; Minnesota Center for Cloning, 216, 373n67
Twin and Adoption Research, 79, 81–82, Cognitive ability tests. See Special mental
95; personality studies, 97, 103–104, 148, abilities
150, 182, 187, 213–214, 242, 251, Cognitive psychology, 7–8
372n53; Science IQ paper, 105–112, 216, Common environment. See Shared
357n46; job satisfaction, 116–117; environment
medical studies, 123, 136; information Comprehensive Ability Battery. See Special
processing, 126–132, 179; public interest, mental abilities
133–135, 218; religiosity, 142, 144, Concordance, 27, 38–39, 66, 74, 88–91,
233–235, 256, 387n67; vocational 93, 140–141, 155, 163–165, 243, 294;
interests, 165, 167, 169; childhood difficulties determining, 66; probandwise
environment, 195, 197; MMPI, 204–205, concordance, 140, 164, 361n15
241; sensation seeking, 208; end of the Conry, J., 154, 156–157
study, 220, 244–245, 250, 257, 326, Conscientiousness, 281. See also Big Five
391n24; free speech movement, 231. See personality traits
also Authoritarianism; Body composition, Conservatism, 32, 77, 99, 142, 215, 231,
nutrition, and energy expenditure (Tufts 253–256, 291–292
University); Conservatism; Controversies Constructive replication, 70, 75, 287, 302
(MISTRA); Creativity; Diet; Ego Continuous (biometrical, polygenic) trait,
development; Evolutionary psychology; 57
Fluctuating asymmetry (FA); Funding, Controversies (MISTRA), 13–14, 29,
General mental ability; Headache study; 101–102, 104, 190, 212, 216, 244,
Heritability; IQ; Morningness; Reading; 255, 266, 296, 298–309, 315, 368n55;
Science (IQ paper); Shared environment; sample size, 13, 32–34, 239, 302–303,
Traditional moral values triad; Twin 309, 313–314; separation and contact,
Relationship Survey; Zygosity diagnosis 13, 42, 44, 108, 299, 302, 304–305,
Briggs Life History Form, 185. See also 307–308; environmental similarities,
Minnesota-Briggs History Record 13, 51–52, 71, 104; funding sources, 13,
Burks, B., 7, 19 309–317; data collection, 26, 28, 40–42,
Burt, Sir Cyril, 4–5, 49, 80, 101, 113, 219, 55–56, 67, 83, 91, 139, 154, 158, 220,
285, 302, 313, 317, 320, 341n24 283, 298–302, 307; restricted environ-
ments, 149; participant recruitment,
California Psychological Inventory, 27, 240, 302–306; data quality, 255; media
150, 188, 234–235, 239–240, 251. coverage, 306–308; chance findings,
See also Personality assessment 307–308; publication record, 308.
 INDEX 403

See also Juel-Nielsen, N.; Kamin, L.; Ekman, P., 282–283

Newman, H. H.; Sample characteristics; Elder, S., 280–281, 296
Science (IQ paper); Shields, J. Electrocardiogram (ECG). See Cardiac
Cope, L., 35, 307 studies
Copy number variations, 324, 326 Electroencephalogram (EEG), 74–75,
Correlation, 27, 29, 58–59, 114; intraclass 109–110, 114–115, 172, 350n41
correlation, 41, 58–59, 61–62, 114; and Emergenesis, 58, 61, 73, 75–78, 171–174,
emergenesis, 58–59; genetic correlation, 176, 186; and epistasis, 58, 61, 172–173
140; phenotypic correlation, 140; Emotionality, 100, 102–103, 237,
cross-twin correlation, 199 251–252, 267, 355n26. See also
Corson (McCourt), K., 218, 231–233, 242 Neuroticism
Creativity, 126, 173–174, 180, 201, 212, Endophenotype, 325
221, 236–239, 296, 366n18 Epigenetics, 324, 326
Criticisms. See Controversies (MISTRA) Epistasis, 57–58, 61, 172–173, 187.
Crystallized intelligence, 190, 262, 382n42. See also Emergenesis
See also Cattell-Horn fluid-crystallized Equal environments assumption, 3, 73
model Erlenmeyer-Kimling, N., 7
Eveningness, 221–222
Daniels, D., 147 Evolutionary psychology, 15, 111–112,
Darley, J., 22, 25, 310–311, 344n9 181–182, 186, 247, 267; life history
Darwin, C., 239 theory, 182, 184. See also Behavior
Dental assessment, 22–23, 96, 153–157, genetics; Sociobiology
243–244 Extraversion, 32, 77, 98–99, 149–151, 172,
Diener, E., 198, 200 185, 187–188, 209–210, 239, 351n15.
Diet, 227–229; and dental assessment, See also Big Five personality traits
153–154; and allergies, 227 Extrinsic religiousness, 233–234, 236
DiLalla, D., 204–206, 240–241
DiLalla, L., 253 Facial expressivity, 282–283; Facial Action
DiLeonardi, J., 33, 50, 170 Coding System, 283
Discordance, 38–39, 41, 66, 74–75, 78, 89, Family environment, 12, 62, 71, 100, 102,
91, 93, 139–141, 164, 206, 294–295, 119, 147–150, 165–169, 187, 195–197,
323–325; difficulties determining, 66 233, 243–244; Moos Family Environ-
Discrete trait, 57 ment Scale, 108–109, 148–150, 166–167,
DNA, 18, 56, 213, 282, 322–324, 345n5, 196–197, 233, 251; Block Environmen-
373n67; and cardiac measures, 125; and tal Questionnaire, 196–197, 233, 251;
evolution, 173. See also Zygosity and personality, 251–254
diagnosis Farber, S. L., 71, 153, 226, 230,
Dominant gene, 57 348–349n20, 364n22
Dusek, V., 308 Fine, W. 35, 317
Dysken, M. W., 25, 137 Fingerprints, 38–39; ridge count, 38,
72–73, 82, 94. See also Fluctuating
Eating behavior, 68, 279–281, 296; Food asymmetry (FA)
Frequency Questionnaire, 228; and Finnish reared-apart twin study, 12, 140;
disinhibition, 280–281; and hunger, reared-together twin study, 101, 123,
280–281, 385n38; and restraint, 125, 164
280–281. See also Diet “Fireman twins,” 15, 49, 64, 111, 132,
Eaves, L. J., 233, 240 146, 151, 181, 177–178, 181, 225, 228,
Eckert, E. D., 20–24, 67–69, 85, 93, 138, 255, 272, 289
228, 308 Fletcher, R., 5
Ego development, 218, 221, 223–225; Fluctuating asymmetry (FA), 270–271,
and cognition, 223–224; Washington 322; dermatoglyphic asymmetry, 270;
University Sentence Completion Test, composite measure vs. single traits,
223–224, 375n12 272–273; and general intelligence, 273
404 INDEX

Fluid intelligence, 190, 382n42. See also Grants, 13, 25, 29, 157–158, 218–220,
Cattell-Horn fluid-crystallized model 275; grant reviews, 29, 219–220,
Fox, P. W. (Bill), 202–203 309–314. See also Funding
Fraga, M. F., 324 Grove, W., 23, 138–140
Freedman, D. G., 183, 250
Freeman, F. N., 9, 12, 60 Halberg, F., 85–86
Friesen, W., 283 Hamilton, W. D., 183, 249
Funding, 13, 19, 29, 36, 104, 133–134, Hammer, M., 230
218–220, 244; sources of, 309–319, Handedness, 3, 193, 272, 293–295,
329–331. Also see Appendix A 389nn85–86
Fuss, P., 218, 274, 276, 278, 282, 300 Hanson, B., 26, 85–86, 124, 164, 296,
Future of twin studies, 319–325 345n12
Happiness, 198–201, 268, 283, 322,
Gabbay, F., 26 369n8, 370n14; happiness set point,
Galton, F., 1–4, 191, 202, 238–239 195, 199–201, 252
Gametes, 56 Harcourt Brace Jovanovich, 25, 311, 329
Gangestad, S., 182 Harkness, A., 242
Garcia Effect, 6 Harmon, L. R., 240
Gardner, H., 382–383n47 Harris, R. E., 240; Harris-Lingoes (MMPI)
Genain quadruplets, 67, 138 subscales, 240–241
Gene, 56–58, 61, 177; periodontal risk Hawaii Ability Battery. See Special mental
(inflammation), 161; mutations, 182, abilities
270; molecular genetics, 322–324 Headache study, 225–227; headaches,
General mental ability, 68, 70, 76, 84, 10, 27, 325–326
105, 129, 190–191, 259–262, 284–289, Height, 11, 38, 61, 72–73, 82, 92–94, 228,
340n22, 393n62; Raven/Mill-Hill, 271–273, 323–324, 326; and homosexu-
68, 75–76, 105, 107, 350n44; Raven ality, 93
Progressive Matrices, 68, 75–76, Hereditary Genius, 238. See also Galton, F.
105–107, 350n44; challenges to MZA Heritability, 1, 41, 60–62, 71, 101,
findings, 70–72, 308–309, 368n55; first 110–112, 177, 190, 203–204, 241, 261,
principal component, 105, 107, 217. 264, 292, 295–296, 302–303, 320–324,
See also Creativity; g factor; Information 346–347n37, 347n40, 350n51, 356n40;
processing; Jensen, A. R.; Verbal- broad heritability, 61–62; narrow
Perceptual-Image Rotation Model; heritability, 61–62, 123, 176; Falconer’s
Wilson-Patterson Conservatism Formula, 62. See also specific areas of
Scale study
GenomeEUTwin Project, 323 Hershberger, S. L., 204
Genotype, 10, 57, 111, 150, 322–323 Herzberg, M., 24, 96
Genotype-environment correlation (G-E), Heston, L. L., 7, 20–24, 27, 66–67, 90, 93,
59–60; passive G-E, 60; active G-E, 60, 300, 316
104, 111, 118, 253; reactive (evocative) Holden, C., 307, 348n12
G-E, 60, 197 Holzinger, K. J., 9, 12, 60
Genotype × environment interaction Homosexuality, 67, 90–93, 141, 308;
(G × E), 59–60, 140 previous twin studies, 90
g factor, 217, 261–262, 285–287. See also Horgan, J., 306, 308
General mental ability Human Genome Project, 213, 322
Goleman, D., 97–98, 100 Hur, Y.-M., 195–197, 207–208, 221–222,
Gorsuch, R. L., 234 228–229, 251, 267, 280
Gottesman, I. I., 7, 9–10, 20–22,
204–205, 253, 295, 297, 311, 315, Iacono, W., 75, 120
324–325, 341n25, 338n20, 390n6, Immunology, 82, 87–88, 165
392n53 Impulsivity, 259; and sensation seeking,
Gough, H., 9, 148–149, 174 195, 207–208; and eating behavior, 281
 INDEX 405

Inclusive fitness, 183–184. See also Jarvik, L., 7

Hamilton, W. D. Jensen, A. R., 4, 306, 315
Information processing, 7, 22, 72, 82–85, Jim twins, 9–10, 17–23, 25–32, 49, 65,
110, 116, 126, 128–132, 167, 179, 221, 87, 111, 151, 205, 218, 226, 297, 303,
351n1; Posner Letter Identification, 83, 305, 310–312, 320, 326, 329, 342nn58–
130; Sternberg Memory Search, 83–84, 59, 344n6, 345n5; headaches, 10, 27,
130–131; Shepard-Metzler Cube 226; school subjects, 19; weight
Rotation, 83–84, 130, 287; and gain, 27
cognitive measures, 83–84, 130–132; Job satisfaction, 11, 115–119, 133,
Experience Producing Drive Theory– 208–209, 211–212, 214, 222, 236,
Revised, 129; Innate Neurological 292, 322, 326; extrinsic job satisfaction,
Structure Theory, 129, 132; Experience 117–118, 211, 236; intrinsic job
Producing Drive Theory, 129, 132, 216 satisfaction, 117–119, 133, 211–212
Intercourse (age at), 91, 185, 269–270, 292 Johnson, W., 217, 257–263, 272, 284–290,
Interests, 11, 82, 111, 250, 297, 301, 322, 295–296, 302–303, 321
333, 362n26, 367n25; recreational Joseph, J., 42, 44, 345n12
(leisure), 77–78, 110, 172, 174–175; Joynson, R. B., 5
occupational (career), 110, 117–118, Juel-Nielsen, N., 4–5, 9, 12, 20, 28, 35, 70,
174, 236; and experience producing 73–74, 79, 91–92, 105, 107, 189, 217,
drive theory-revised, 129; interests and 247, 295, 297, 304–305, 340n19,
talents, 174–178, 215; feminine interests 341n30, 342n56, 348n20, 368n56,
(MMPI), 206; activity-ability groupings, 389n85
287–290. See also Personality assess-
ment; Religiosity; Vocational interests Kagan, J., 101–102
International HapMap Project, 322 Kamin, L., 4, 101, 302, 308, 320,
Intrauterine environment, 71, 216. See also 341nn24–25, 348–349n20
Prenatal environment Kaplan, J., 71
Intrinsic religiousness, 233–236, 256 Kety, S., 241
Introversion, 151, 209–210, 239, 242 Keyes, M., 19, 26, 218
IQ, 4–5, 7, 10, 12–14, 27, 29, 32, 37, Kit of Factor-Referenced Cognitive Tests,
40–41, 44, 59, 60, 62, 68, 70–72, 75–76, 84, 126. See also Special mental abilities
80, 95, 104–114, 130, 133, 144, 186, Knobloch, W. H., 88–89
189–190, 215–218, 261, 285, 287, Koch, D. H., 317; David H. Koch
289, 295, 326, 356n38, 382n47; data Charitable Foundation, 218–219,
collection, 105, 107, 299–300, 368n55; 317–318, 329–331, 393nn57–59
heritability, 107–108, 112, 286–287, Koenig, L., 256, 291, 361n22, 381–382n32
356n40; and religiosity, 144; and Kohler, P., 87
ego development, 224, 375n12; and Kojetin, B., 142, 144, 236, 361n22
conservatism, 254; and Stroop Color- Korean twins reared apart, 207, 229, 250,
Word Test, 259; and fluctuating 267
asymmetry, 272–273; within-pair Koshland, D. E., 13, 105
differences, 286–287, 386n51; and brain Krueger R. F., 241–242, 251–253,
volume, 289; mean of MISTRA sample, 395n93
295; MMPI, 362n31. See also Burt, Sir
Cyril; Controversies (MISTRA); General Leahy, A., 7
mental ability; g factor; Science (IQ Lewis, J. See Jim twins
paper); Wechsler Adult Intelligence Life history interview, 32, 44, 47, 51, 108,
Scale (WAIS) 117, 138, 185; clinical interview, 24; life
stress interview, 24
Jackson Vocational Interest Survey, Lingoes, J. C., 240. See also Harris, R. E.
166–167, 209, 290. See also Vocational Linkage analysis, 322
interests Locus, 56–57, 61; DNA markers, 38,
Japanese reared-apart twin study, 12 345n5
406 INDEX

Loehlin, J. C., 3, 142, 182, 187, 217, 231, Minnesota Twin Registry. See Twin
240, 361n20, 393n68 registries
Loevinger, J., 223–225, 375n15 Mischel, W., 1
Lykken, D. T., 9, 18–22, 26, 30, 46, 65, Model fitting, 123, 126, 131, 144, 181,
72–78, 96–97, 101, 104, 107, 119–122, 187, 197, 203, 213, 227, 229, 232.
125, 162, 172, 174–178, 226, 231, 233, See also Biometrical modeling
237, 272, 299, 331, 314, 344n6, 355n21. Molecular genetics, 144, 213, 215, 265,
See also Constructive replication; 320, 322–327
Electroencephalogram (EEG); Emergen- Moloney, D., 83, 165–169, 174, 208–209
esis; Funding; Happiness; Twin registries Morningness, 221–222
Lynch, L., 97–98, 306–307 Moster, M., 83
Muller, H. J., 4
MacKinnon, D., 9 Multidimensional Personality Questionnaire
Markon, K., 241–242, 251 (MPQ), 22, 76, 99, 102–104, 172, 176,
Martin, N. G., 32, 111, 151, 231, 253–254 237, 242, 251, 253–255; higher order
McClearn, G. E., 189 factors, 99–103, 355n26; well-being
McClintock, M., 268 scale, 199. See also Personality assessment
McGue, M., 24, 29, 44, 62, 96, 128–132, Myers, D., 198
148, 150, 167, 179, 189–190, 217, 251,
257, 308, 316, 361n15 National Institutes of Mental Health,
McGuire, S. A., 292 13, 67, 112, 219
Mealey, L., 184–186, 274 National Science Foundation, 219, 311, 329
Media coverage. See Controversies Nelkin, D., 307
(MISTRA) Neubauer, P., 5–6, 49, 341n30
Memory, 190, 206. See also Information Neuroticism, 32, 77, 98–99, 101. See also
processing; Special mental abilities; Big Five personality traits
Verbal-Perceptual-Image Rotation Neuroticism-Extraversion-Openness
Model (NEO) Inventory, 32
Menarche (age at), 93, 267–270, 292, 322; Newman, D., 223–225
Belsky-Draper hypothesis, 268–269 Newman, H. H., 5, 9, 12, 60, 70, 73,
Merrell, D., 295 79, 98, 105, 107–108, 189, 217, 247,
Merriman, C., 2 271, 295, 302–304, 321, 356n38,
Messer, L. B., 153–154 368n56
Metabolic syndrome, 280. See also Body New York Review of Books, 71
composition, nutrition, and energy New York Times, 13, 19–20, 27, 97, 134,
expenditure (Tufts University) 200, 305, 307, 348n20
Michalowicz, B. S., 24, 68, 157–162, Nichols, R., 3, 142, 231
243–244, 296, 310 Nightline, 104
Minneapolis Star Tribune, 19, 35 Nisbett, R., 71
Minnesota-Briggs History Record, 251. Nonadditivity, 61–63, 99, 102, 125, 176,
See also Briggs Life History Form 182, 187–188, 203, 205, 212, 347n40
Minnesota Center for Twin and Adoption Nonshared environment, 63, 99, 102–103,
Research, 8, 79, 81–82, 95 144, 169, 177, 190, 197, 209, 214, 224,
Minnesota Daily, 23, 348n11 232, 251, 293, 324, 326
Minnesota Multiphasic Personality Nontwin participants. See Sample
Inventory (MMPI), 22, 47, 72, 95, 99, characteristics
195, 204–207, 302; religious beliefs, Novitski, E., 263
142; content scales, 205–206; clinical
scales, 205, 240; profile analysis, Occupational similarity. See Job
206–207; creativity, 237–238; psycho- satisfaction
motor acceleration scale, 240; Harris- Oldfield Handedness Inventory, 293
Lingoes Scales, 240–241; Subtle-Obvious Openness, 32, 174. See also Big Five
Scales, 240–241 personality traits
 INDEX 407

Ophthalmology, 82, 88–89, 221 Placement (in rearing homes), 50, 71, 94,
Otis IQ, 107, 356n38 108–109, 149, 235, 254, 303; selective
placement, 224, 261, 308
Parental characteristics. See Sample Platt, J. R., 292
characteristics Plomin, R., 100, 112, 147, 173, 189, 217,
Participant identification. See Sample 236
characteristics Polesky, H. F., 24, 344n10
Pedersen, N. L., 189, 217, 240 Popenoe, P., 3–4
Periodontal examination, 22, 24, 68, 96, Prenatal environment, 10, 13, 41, 286,
152, 154, 157–162, 165, 243–244, 296; 293, 304; and homosexuality, 90.
alveolar bone height, 159, 161; oral See also Cloning; Twin method
bacteria, 159, 162, 243–244; funding, Propping, P., 2, 186, 190–191
219, 310, 330–331; temporomandibular Psychomotor assessment, 23, 202–204,
joint symptoms, 243 299, 370n26; practice effects, 201–204;
Personality assessment, 11, 32, 116, 222, Rotary pursuit test, 202–203. See also
354nn2,6,8, 355n21, 360n13, 368n52, Minnesota Multiphasic Personality
386n43; Differential Personality Inventory (MMPI)
Questionnaire, 22, 76–77; Multidimen- Psychopathology, 116, 323–324; Tourette
sional Personality Questionnaire, 22, syndrome, 24, 136–138, 141; drug
76–77, 97–104, 172, 176, 199–200, abuse/dependence, 69, 139–140; alcohol
254; California Psychological Inventory, abuse/dependence, 69, 139–140, 144,
27, 148–150, 234–235, 239–240; 320; Diagnostic Interview Schedule,
heritability, 41, 102–103, 144, 165, 138–141; Diagnostic Statistical Manual
181–182, 195, 214–215, 319, 322, 326; (DSM)-III, 138–139, 360n6; antisocial
temperament, 66; and IQ, 71, 108, personality disorder, 139–141; MMPI
110–112; and weight, 73, 281; and twin personality correlates, 195, 204–207,
relationship, 76–77, 188, 240, 250; and 242; and religiosity, 234; and talent,
environment, 147–151, 188, 214, 238. See also Minnesota Multiphasic
251–253; and creativity, 173–174, Personality Inventory (MMPI)
237–238; psychopathology, 205, 242; Pulmonary study, 22, 78
personality “set point,” 252; and birth
order, 292–293; prenatal effects, Raven/Mill-Hill, 68, 76, 105, 107
349n28. See also Authoritarianism; Big Raven Progressive Matrices. See General
Five personality traits; Conservatism; mental ability
Controversies (MISTRA); Ego develop- Reading, 8, 32, 77, 257–261; Oskar and
ment; Interests; Minnesota Multiphasic Jack, 132; genetic factor structure,
Personality Inventory (MMPI); 259–260
Multidimensional Personality Question- Recessive gene, 57
naire (MPQ); Psychopathology; Recreational interests. See Interests
Religiosity; Reproduction-related Recruitment (of participants), 9, 35–37,
behaviors; Sensation seeking; Stroop 53–54, 90, 218–219, 302; participant
Color-Word Test; Twin method; sources, 33; pairs studied per year, 37.
Vocational interests See also Controversies (MISTRA);
Peterson, G., 19 Sample characteristics
Phenotype, 57, 59, 112, 322–323. See also Religiosity, 10–11, 13, 47, 109, 116,
Endophenotype 141–149, 173, 177, 181, 185, 196,
Phenylketonuria (PKU), 7, 57, 342n42 206, 213, 218, 221, 231, 233, 255–256,
Physical facilities, 51, 108, 109, 130, 361n22; and traditionalism, 144,
166, 251 189, 199; Age Universal Religious
Physical similarity. See Twin method Orientation Scale, 234. See also
Pihlstrom, B., 24, 96, 157–159, 161–162, Extrinsic religiousness; Intrinsic
243–244, 310, 330 religiousness; Traditional moral
Pioneer Fund. See Funding values triad
408 INDEX

Reproduction-related behaviors, 184–186. Shared environment, 26, 72, 100–103,

See also Intercourse (age at); Menarche 135, 147–148, 312, 319; defined, 63;
(age at) intrauterine, 71; MZA and MZT twins,
Resnick, S., 26, 316, 342n59 73, 79, 99, 122, 200, 267; parenting, 98;
Rich, S., 96, 354n48 and intelligence, 109–110, 356n40;
Right-Wing Authoritarianism Scale, 215, cardiac study, 125, 133; special abilities,
231–233, 254–255 128; virtual twins, 133, 190; attitudes
Roberts, S., 218, 274–276, 279–282, and beliefs, 142, 144, 362n30; aller-
309–310, 323, 330 gies,165; vocational interests, 169, 209;
Roe, A., 19 personality, 187–188, 214, 242, 244;
Rose, R. J., 101, 355nn21–22 morningness-eveningness, 222; ego
Rosenthal, D., 7 development, 225; headaches, 227;
Ross, J. M., 234 authoritarianism, 232; oral bacteria,
Rowe, D. C., 112, 187, 196, 251, 269 243; temporomandibular joint symp-
toms, 244; teenage emotions, 253; age
Sample characteristics, 21, 25, 29, 37, 141, at menarche, 267; weight, 271; love
206, 217; rearing circumstances, 13, styles, 322
50–51; sample size, 21, 29, 41–42, 44, Shields, J., 5, 8, 12, 30, 72–73, 99, 343n68,
68, 239, 266, 273; nontwin participants, 354n10; schizophrenia research, 7, 9, 22;
25, 44–45; separation and contact, 32, twin method, 28; zygosity diagnosis, 39;
37–38, 41–43, 48, 70–73, 106, 139, 168, separation and contact, 51, 53; EEG, 74;
235, 249, 299, 302, 356n39; representa- personality, 77, 98–99, 107, 150, 188,
tiveness, 36–37, 140–141, 188, 196, 206, 354n8; homosexuality, 90–91; weight,
240; age at assessment, 37–38, 41–44, 94; IQ, 107, 189, 217, 386n51; voice,
73, 106; zygosity, 38–39, 41–42, 51, 230; social relationship, 247; handed-
154, 224, 247, 302; individual sample, ness, 295, 389n86; and criticisms,
41; paired sample, 41; sex, 41–42; 302, 314, 341n25, 390n6; participant
country of residence, 44–45; rearing recruitment, 304, 380n9; endophentypes,
circumstances, 50–51; number of 325
residences, 51, 130; parental character- Siemens, H., 2
istics, 53; reunion circumstances, 53; Sines, L., 24
rule of two-thirds, 175. See also Skin conductance, 75, 115–116, 119–122,
Controversies (MISTRA); Homo- 350n42, 358n9
sexuality; Recruitment (of participants); Slossen Oral Reading Test, 258–259
Zygosity diagnosis Social attitudes, 11, 103, 109, 212, 231,
Samuels, J., 8, 258–259 253, 256, 264, 386n43; Wilson-
Sanderson, M., 91 Patterson Conservatism Scale, 32, 231,
Scarr, S., 292, 348–349n20 253–255; Public Opinion inventory, 32,
Science (IQ paper), 13, 36, 104–114, 116, 377n54; and fitness, 292
133, 186, 216, 355n29, 357n46; Social relationships. See Twin Relationship
environmental measures, 108–109; Survey
special issue, 213. See also General Sociobiology, 6, 181, 183. See also
mental ability; IQ Evolutionary psychology
Scientific American, 308, 391n34 Special mental abilities, 5, 23, 40, 68, 82,
Seaver Institute, 318–319, 329–331, 393n60 84, 93–94, 107, 116, 125–128, 257,
Sensation seeking, 195, 207–208 259, 299, 308; Comprehensive Ability
Sex differences, 284, 287–290 Battery, 68, 105, 125–127, 130, 238,
Sexual attraction, 15, 91–93; histocompat- 261, 287; ability factors, 68, 126–128,
ibility, 92; genetic sexual attraction, 92, 130–131; and information processing,
343n71; Westermarck Effect, 92, 343n71 83–85, 128–131, 351n1; Hawaii Ability
Sexual life history, 23, 54, 89–93, 141. Battery, 84, 94, 105, 125–127, 130–131,
See also Homosexuality; Intercourse 224, 261, 287; Kit of Factor-Referenced
(age at); Menarche (age at) Cognitive Tests, 84, 126; cousin
 INDEX 409

correlations, 94, 126, 301; Stroop Twin method, 1–3, 191, 241; criticisms,
Color-Word Test, 259; sex differences, 13, 71, 114, 216, 305; statistical vs.
287. See also General mental ability anecdotal, 27–28. See also Equal
Spelling, 8, 128, 204, 258–260, 288, 313. environments assumption
See also Jim twins Twin registries, 12, 35–36, 90, 195, 302;
Spencer Foundation, 25, 311, 329, 330 Danish Twin Registry, 9, 79, 304–305;
Springer, J. See Jim twins Minnesota Twin Registry, 21–22, 35, 99,
Stahl, L., 264 133, 143, 158, 163, 174, 211, 220, 226,
Stassen, H. H., 114 231, 267, 331; National Academy of
Strategy of Strong Inference, 292 Sciences-National Research Council
Strong-Campbell Interest Inventory, 143, Twin Registry, 212; Kansas Twin
166, 237, 290; Hansen combined scales, Registry, 226
209. See also Vocational interests Twin Relationship Survey, 15, 41, 44, 47,
Strong Vocational Interest Blank-Strong- 247–250, 268, 270; twin relationships,
Campbell Interest Inventory, 166 11, 28, 47, 49, 63, 98, 147, 169, 180,
Stroop Color-Word Test, 257–259 183–184, 246–247, 319; twin vs.
Stroud, J., 47, 95–96 adoptive sibling, 248, 250
Sulloway, F., 71, 292, 293 Twins Magazine, 96
Swedish Adoption and Twin Study of Twin type, 2, 13, 38–40, 89, 158, 305;
Aging (SATSA), 5, 12, 114, 140, 172, conceptualization of, 249
187, 189, 196, 240, 287, 293, 300, 302
Szasz, T., 241 Unique environment. See Nonshared
environment
Taylor, H., 70–71, 113, 348n17. See also University of Pennsylvania Smell Identifica-
General mental ability; IQ tion Test, 191, 242
Tellegen, A., 9, 20–22, 76, 96, 99, 100, Unrelated siblings, 26, 94, 248, 250, 322.
103, 169, 198–201, 223, 242, 254, 259, See also Virtual twins
297, 316
Tesser, A., 212–213 Vectorcardiogram. See Cardiac studies
Till, M. J., 23, 153–155 Venable, G. D., 234
Traditionalism, 100, 102–103, 142, 144, Venezky, R., 258
199, 237, 253–255, 291–292, 322 Verbal-Perceptual-Image Rotation Model,
Traditional moral values triad, 215, 262–263, 284–285, 287, 290; compared
291–292. See also Authoritarianism; to other models, 262–263; content
Conservatism; Religiosity memory, 262, 285, 287–288. See also
Triplets, 104, purposeful separation, 6, General mental ability; g factor; IQ
49; and Tourette syndrome, 24–25, Vernon, P. A. (Tony), 129, 263, 285,
136–138; statistical analysis, 41, 316–317
386n47, 388n83; MISTRA participants, Vernon, P. E., 262–263
42, 45, 48, 50, 52, 72, 82, 85; 89, 95, Virtual twins, 71, 133, 190, 217,
105, 126; reunion, 246. See also Bernard, 301, 322
V.; Cardiac studies; Creativity; Dental Vocational interests, 93–94, 113, 165–169,
assessment; Handedness; Interests; 174, 208–210
Neubauer, P.; Psychopathology; Voice quality, 230
Pulmonary study
Tucker, W. H., 49, 309 Wachs, T., 148
Tufts University study, 191, 218–219, 271, Waller, N., 142–144, 173–174, 236,
274–280, 296, 300; body size, 191, 271; 362nn26–28
energy expenditure, 274, 276–278, 281; Watson, J., 4, 239
food intake, 276–281; funding, 309, Web site (DrNancySegalTwins.Org), 14,
330. See also Diet 23, 56, 99, 124–125, 213–214, 277
Tuna, N., 23, 85, 123–125 Wechsler Adult Intelligence Scale (WAIS),
Turkheimer, E. 169, 356n40 26, 32, 68, 84, 105–107, 110, 224, 254,
410 INDEX

WAIS (continued) Wilson, R. S. 114, 215

259, 261, 302; placement coefficients, Wilson-Patterson Conservatism Scale.
108–109. See also General mental ability See Social attitudes
Weight, 11, 27, 38, 57–58, 66, 73, 82, Woodcock-Johnson Reading Mastery Test,
93–94, 122, 141, 207, 228–229, 271, 32, 258
274–275, 277, 279–281; birth weight, Work values, 13, 178–181, 209, 212–214,
10, 216, 257. See also Body composition, 221. See also Job satisfaction
nutrition, and energy expenditure (Tufts
University) X chromosome, 2, 122
Well-being, 76, 100, 103, 172, 199–200,
214, 242. See also Happiness Ziegler, D., 226–227
Whitfield Institute, 244, 319, 329, 331 Zuckerman, M., 207
Wiener, D. N., 240 Zygosity (MISTRA twins). See Sample
Wilcox, K. J., 80, 96, 133 characteristics
Wilde, G. J. S., 76 Zygosity diagnosis, 38–39, 345n5, 369n10;
Willerman, L., 7 behavioral similarity, 305. See also Twin
Wilson, E. O., 6, 112, 183, 234 type