Objectives

This report presents national Cohabitation, Marriage,

estimates of the probabilities of marital
and cohabitation outcomes for women
1544 years of age in 1995, by a wide
Divorce, and Remarriage in the
variety of individual- and community-
level characteristics. The life-table United States
analysis in this report takes a life cycle
approach to estimate the probabilities Matthew D. Bramlett, Ph.D. and William D. Mosher, Ph.D.,
that:
Division of Vital Statistics
+ a woman will marry for the first time,
+ an intact first cohabitation will make
the transition to marriage,
+ a first cohabitation will end in
separation,
+ a first marriage will end in separation Highlights probability that a first marriage will
break up, the probability that a woman
or divorce,
whose first marriage has disrupted will
+ a disrupted first marriage will be
enter a new cohabitation, the probability

T
followed by a new cohabitation, his report presents data from
+ a separation from first marriage will Cycle 5 of the National Survey of that a separation from first marriage will
result in divorce, Family Growth (NSFG). The become a legal divorce, the probability
+ a divorce from first marriage will be NSFG is a nationally representative that a divorced woman will remarry, and
followed by remarriage, and survey focused on marriage, divorce, the probability of second marriage
+ a second marriage will end in contraception, infertility, and other disruption. A wide variety of
separation or divorce. characteristics of women and the
factors affecting pregnancy and birth
rates and womens health. Cycle 5 of communities in which they live are used
Methods to examine these cohabitation and
the NSFG was based on face-to-face
The life-table estimates presented marital outcomes.
here are based on a nationally interviews with 10,847 women 1544
years of age in 1995. The analysis of The analyses in this report are
representative sample of women 1544
trends in this report is based on data intended to provide a statistical
years of age in the United States in
1995 from the National Survey of from the 1973, 1976, 1988, and 1995 description, not a definitive or
Family Growth, Cycle 5. cycles of the NSFG. For convenience in exhaustive explanation of these topics.
writing in the text of this report, The data shown here are intended to
Results non-Hispanic white women are often suggest that both characteristics of
The analyses show that various referred to as white and non-Hispanic individuals and the communities in
individual and community-level black women are often referred to as which they live are often important
characteristics are related to the marital black. The full labels are always used factors in understanding cohabitation
and cohabitational outcomes examined and marriage and to encourage
in the tables and graphs.
in this report. The results consistently researchers to consider these factors
This report contains 44 detailed
demonstrate that the cohabitations and when studying these issues. This report
marriages of non-Hispanic black women tables showing analyses of eight
outcomes related to cohabitation and also attempts to shed light on at least
are less stable than those of non-
marriage: the probability that a woman five important issues in the recent
Hispanic white women. An analysis of
trends over time suggests that will marry for the first time, the statistical literature on marriage and
differences by race/ethnicity are probability that an intact first premarital divorce:
becoming more pronounced in recent cohabitation will become a marriage, the + What are the recent trends in marital
years. Racial differences observed are probability that a first premarital breakup, divorce, and remarriage?
associated with individual cohabitation will break up, the + Do the trends in these outcomes
characteristics and with the
characteristics of the communities in differ by race/ethnicity?
which the women live.
The 1995 National Survey of Family Growth was jointly planned and funded primarily by the National
Center for Health Statistics, the National Institute for Child Health and Human Development (NICHD),
Keywords: cohabitation c marriage c the Office of Population Affairs, and the National Center for HIV, STD, and TB Prevention, with
separation c divorce c remarriage c additional support from the Childrens Bureau. The authors gratefully acknowledge the technical
context assistance of Wayne E. Johnson, Ph.D., of the Office of Research and Methodology for assistance in
estimating standard errors of the statistics in this report. The authors gratefully acknowledge the helpful
review and comments of Dr. V. Jeffery Evans of NICHD. This report was edited by Patricia Keaton-
Williams, graphics produced by Jarmila Ogburn, and typeset by Jacqueline M. Davis.

Page 1
Page 2 [ Series 23, No. 22

+ Are characteristics of communities womans race, age, education, the more likely for cohabiting white women
related to couples success in households income, and the economic who report that their religion is either
marriage? opportunities in the community. The somewhat or very important than for
+ Is the statistical portrait of marriage data also show that a great many those who report that their religion is
affected if we measure unmarried marriages end in legal separation but not not important (figure 9).
cohabitation and separation from in divorce, and that looking only at Cohabiting women are more likely
marriage as well as legal marriage divorce greatly understates marital to marry if they live in communities
and divorce? disruption among some groups with lower male unemployment, higher
+ What demographic, economic, and especially non-Hispanic black and median family income, lower poverty,
social factors affect the chances that Hispanic women. and lower receipt of welfare (figure 10).
marriage will succeed or fail? The male unemployment rate seems to
What demographic, economic, and be more important among black women
What are the trends? Our data show an social factors affect the chances that than among white women (figure 11).
increase in the chances that first marriage will succeed or fail? This After the first 3 years of
marriages will end (in separation or report shows that a number of cohabitation, the probability that a
divorce) for marriages that began in the characteristics are closely associated first premarital cohabitation breaks
1950s through the 1970s. From the early with the chances that a marriage will up is higher among black women than
1970s to the late 1980s, the rates of continue or break up. For first among Hispanic or white women
breakup were fairly stable. The marriages, for example, marriages are (figure 12) and is higher among younger
probability of remarriage following less likely to break up, and more likely than older women (figure 13), especially
divorce has decreased slightly, and the to succeed, if the wife grew up in a among white women (figure 14).
probability that the second marriage will two-parent home, is Asian, was 20 years Women who have ever been forced to
break up has risen from the 1950s to the of age or over at marriage, did not have have intercourse before the cohabitation
1980s. any children when she got married, is began are more likely to experience the
college-educated, has more income, or breakup of their first premarital
Do the trends differ by race/ethnicity?
has any religious affiliation. cohabitation than women who have
It appears that these trends were similar
The following highlights illustrate never been forced (figure 15).
for non-Hispanic white and non-
the kinds of findings shown in this Cohabiting women are more likely
Hispanic black women, but black
report: to experience the breakup of their first
women faced higher rates of marital
The probability of first marriage premarital cohabitation if they live in
breakup, lower rates of making the
is lower for non-Hispanic black women communities with higher male
transition from separation to divorce,
than for other women (figures 1 and 2). unemployment, lower median family
and lower rates of remarriage. Among
Getting married by the 18th birthday is income, and higher rates of poverty and
white women, the increasing probability
more likely for Hispanic and receipt of welfare (figures 16 and 17).
of first marriage breakup leveled off in
non-Hispanic white women and less Black women are more likely to
the 1970s but appears to have continued
likely for non-Hispanic black and Asian experience first marital disruption and
rising for black women through the
women (figure 2). First marriage is less Asian women are less likely to
1980s.
likely for women who report that their experience first marital disruption,
Are characteristics of communities religion is not important (figure 3). compared with white or Hispanic
related to success in marriage? This Early marriage is more likely for women (figure 18). First marriages of
report shows clear evidence that women in communities with higher women who are 20 years of age or over
community prosperity is related to male unemployment, lower median at marriage are less likely to break up
successful cohabitations and marriages, family income, higher poverty and than marriages of teenaged brides; but
and that neighborhood poverty increases higher receipt of welfare (figure 4). First there is no significant difference by age
the likelihood that cohabitations and marriage is more likely in at marriage among Hispanic women
marriages will fail. nonmetropolitan areas and less likely in (figure 19). Women whose religion is
central cities (figure 5). somewhat or very important are also
Is the statistical portrait of union
The probability that an intact first less likely to experience a breakup of
formation and dissolution affected if
premarital cohabitation becomes a their first marriage than those whose
we measure unmarried cohabitation
marriage is higher among white women religion is not important (figure 20).
and separation from marriage as well
and lower among black women Women who lived with both parents
as legal divorce? One major advantage
(figure 6); higher among couples with throughout childhood are less likely to
of survey data on marriage is that we
higher incomes than for couples with experience the breakup of their first
are not limited to examining legal
lower incomes (figure 7); and higher for marriage than women who were not
marriage and divorce. The data in this
cohabiting women with any religious raised with two parents throughout
report show that the probability that an
affiliation than for those with no childhood (figure 21). Women who have
intact premarital cohabitation will result
religious affiliation, especially among never been forced to have intercourse
in marriage is 70 percent after 5 years;
white women (figure 8). Marriage is before marriage are less likely to
that probability is associated with the
 Series 23, No. 22 [ Page 3

experience the breakup of their first second marriage disruption is more the stability of marriage than for the
marriage than women who have ever likely among women who were younger stability of cohabitation, and some of
been forced to have intercourse before than age 25 at remarriage than women the effects vary by race/ethnicity.
marriage (figure 22). The chance of who were older at remarriage
marital disruption is lower if the wife (figure 39), more likely among women
had her first birth after marriage who were not raised throughout
Introduction
(figure 23). childhood with two parents (figure 40),

M
Women who have ever suffered more likely among women who have arriage is associated with a
from generalized anxiety disorder ever been forced to have intercourse variety of positive outcomes,
(GAD) are more likely to experience the before marriage than women who have and dissolution of marriage is
breakup of their first marriage than never been forced to have intercourse associated with negative outcomesfor
women who have never suffered from before marriage (figure 41), and more men, women, and their children. A full
GAD (figure 24). Interracial marriages likely among women who have ever analysis of the benefits of marriageto
are more likely to disrupt than marriages suffered from GAD than women who either children or spousesis beyond
in which both spouses are the same have never suffered from GAD the scope of this report; but this brief
race/ethnicity (figure 25). First marriages (figure 42). review should serve to highlight the
are more likely to disrupt in Women with no children at the start importance of the data described in this
communities with higher male of the second marriage are the least report. The purpose of this report is to
unemployment, lower median family likely to experience second marital present estimates of the patterns of
income, higher poverty, and higher disruption. Among those with children cohabitation, marriage, divorce, and
receipt of welfare (figures 26 and 27). at remarriage, those with any unwanted remarriage in the United States as of
Entering a new cohabitation after children are more likely to experience a 1995, by a wide variety of individual-
the first marriage ends is more likely second marital disruption than those and community-level characteristics. We
among white women than black women with no unwanted children (figure 43). do not attempt to provide rigorous
(figure 28); more likely among women Women who live in communities with explanations for the many findings
with no religious affiliation than women higher male unemployment, lower reported here. The intent is to present
with any religious affiliation (figure 29); median family income, higher poverty, the findings in a statistically sound
more likely if she has few or no and higher receipt of welfare are more format, in greater detail than has ever
children (figure 30); and more likely for likely to experience the second marital been done for the United States, and
women who live in communities with breakup (figure 44). thus to encourage more understanding
low male unemployment, poverty, and Although the statistics presented in and further study of these important
receipt of welfare (figure 31). this report are descriptive in nature, it is topics.
Separated white women are more possible to observe how the Compared with unmarried people,
likely to complete the legal divorce characteristics of individuals and married men and women tend to have
process than separated Hispanic or communities may be related to the lower mortality, less risky behavior,
black women (figure 32). The transition stability of cohabitations and marriages. more monitoring of health, more
from separation to divorce is less likely Cohabitations and marriages tend to last compliance with medical regimens,
among women who live in less longer if the woman was older at the higher sexual frequency, more
prosperous communities (figure 33). time the cohabitation or marriage began, satisfaction with their sexual lives, more
The probability of remarriage is if her family income is higher, if she savings, and higher wages (13). The
highest among white divorced women has any religious affiliation or reports differences between married and
and lowest among black divorced that her religion is important to her, if unmarried people may reflect a causal
women (figure 34). Remarriage is more she was raised through childhood in a effect of marriage or a selection effect.
likely among women who were under two-parent intact family, if she had Healthier people may be more likely
age 25 at divorce than among women never been forced to have intercourse, if than others to find mates and marry.
ages 25 and over at divorce (figure 35). she had no children at the start of the Research has suggested that the benefits
Remarriage is more likely for divorced cohabitation or marriage, if her first of marriage may be partially due to a
women who live in communities with birth was at least 8 months after the selection effect and partially due to true
lower male unemployment, poverty, and beginning of the cohabitation or benefits to be gained from being
receipt of welfare (figure 36). marriage, if she has never suffered married as opposed to being unmarried
Remarriage is more likely for women generalized anxiety disorder, if she is (3,4). A lower mortality risk among the
who live in nonmetropolitan areas and is the same race/ethnicity as her husband, married has been shown to persist even
least likely for women who live in the or if she lives in communities with after health in early adulthood was
central cities of metropolitan areas higher median family income, lower controlled, suggesting that at least part
(figure 37). male unemployment, less poverty, less of the benefit of being married is not
Black women are more likely to receipt of welfare, and more adults who the result of selection (4).
experience the breakup of their second are college-educated. Some of these Compared to married individuals,
marriage than other women (figure 38); characteristics show stronger effects for divorced persons exhibit lower levels of
Page 4 [ Series 23, No. 22

psychological well-being, more health of divorce illustrate the importance of Characteristics of individuals related to
problems, greater risk of mortality, more examining trends and differentials in the a higher probability of divorce include
social isolation, less satisfying sex lives, patterns of marriage and divorce over younger age at marriage, lower
more negative life events, greater levels time. education and later birth cohort (23),
of depression and alcohol use, and lower later marriage cohort and presence of a
levels of happiness and self-acceptance Trends and Differences in premarital birth (24), premarital
(5). The economic consequences of cohabitation (25), and premarital sexual
divorce can be severe for women. Most Marriage and Divorce activity (26). Catholic white women are
often, children remain with the mother In the United States during the less likely to divorce than non-Catholic
after divorce; the loss of the ex- second half of the twentieth century, the white women (24). Marital
husbands income often results in a proportion of peoples lives spent in dissatisfaction has been found to be
severe loss of income per capita (6,7). marriage declined due to postponement associated with psychiatric disorders
For a man, the retention of income of marriage to later ages and higher such as GAD, depression, and panic
combined with decreased family size rates of divorce (12). The increase in (27). Other characteristics related to a
may actually result in an increase in his nonmarital cohabiting has also lower probability of remarriage include
new households income per capita (6,8). contributed to the decline in the higher education and older age at
Adverse outcomes accrue to proportion of peoples lives spent in divorce (28) and presence of children
children of divorce and children raised marriage. Increasing rates of from prior marriages (9).
in single-parent families. Although not cohabitation have largely offset Lower economic prospects for
all single-parent families are the result decreasing rates of marriage (13,14). less-educated young men have been
of divorce and not all divorced mothers The proportion of time spent in hypothesized to decrease the probability
remain single, virtually all children of marriage has varied across demographic of marriage. The increasing economic
divorce spend some time in a subgroups. Since 1950, the marital independence of women has also been
single-parent household until the mother patterns of white and black Americans hypothesized to decrease the probability
remarries. Even when the mother does have diverged considerably. About of marriage, although recent evidence
remarry, studies suggest that children in 91 percent of white women born in the suggests that the increasing economic
stepfamilies have similar risks of 1950s are estimated to marry at some independence of women may actually
adverse outcomes as children in time in their lives, compared with only increase the probability of marriage as
single-parent families: both groups of 75 percent of black women born in the earnings and employment may make
children do worse than children living 1950s (13). Black married couples are either partner an attractive potential
with two biological parents in terms of more likely to break up than white spouse (17,21). Marriage market
academic achievement, depression, and married couples, and black divorcees are conditions may also play a role, in that
behavior problems such as drug and less likely to remarry than white the probability of divorce is higher in
alcohol abuse, premarital sexual divorcees (13). areas with large numbers of
intercourse, and being arrested (9). The degree of attachment to economically attractive potential
Single-parent families have lower marriage among black Americans is alternate partners (17,29).
levels of parental involvement in school similar to that of white Americans as A full analysis of all of the
activities and lower student measured by attitudes toward marriage individual- and community-level
achievement, compared to two-parent (15,16). One explanation offered by characteristics associated with
families (10). Children raised in some researchers for the lower cohabitation, marriage, and divorce is
single-parent families are more likely to proportion of time spent in marriage beyond the scope of this report. The
drop out of high school, have lower among black Americans is the idea of a purpose of this report is to present
grades and attendance while in school, marriage squeeze, in which the estimates of the patterns of cohabitation,
and are less likely to attend and marriageable pool of black men is marriage, divorce, and remarriage in the
graduate from college than children low due to high rates of joblessness, United States as of 1995 by a wide
raised in two-parent families (11). They incarceration, and mortality (1719). variety of demographic and community
are more likely to be out of school and Employed men are more likely than characteristics. The individual
unemployed and are also more likely to unemployed men to marry (20). characteristics include some which have
become single parents themselves, than In addition to race and employment been shown to be related to marital
children raised in two-parent families status, other characteristics of outcomes in the literature cited above:
(11). Studies have found that, compared individuals that have been found to be age, race/ethnicity, education, income,
to children in two-parent families, related to a higher probability of getting employment status, religion, family
children of divorce score lower on married include higher education and background, parity, GAD, and whether
measures of self-concept, social earnings (21). Characteristics related to the woman cohabited with her husband
competence, conduct, psychological getting married earlier include growing before marriage (9, 13, 2028). Other
adjustment and long-term health (5). up in a disrupted family and higher individual characteristics have been
The positive health benefits of levels of parents education (22). found in other analyses of the National
marriage and the negative consequences Survey of Family Growth (NSFG) to be
 Series 23, No. 22 [ Page 5

correlated with related variables such as are more or less important than others, although a third wave of data is
marital status, age at marriage, or year but such analysis is beyond the scope of being collected in 200102 (34).
of marriage: forced intercourse, timing this report. The estimates in this report + The U.S. Census Bureaus Survey of
of first birth, and whether births were are based on Cycle 5 of NSFG, Income and Program Participation
unwanted (30). conducted in 1995 by the Centers for (SIPP) is a longitudinal panel survey
Some of these individual Disease Control and Prevention of approximately 37,000 households
characteristics are not available for all (CDC)/National Center for Health that includes a marital history and a
analyses. For example, whether the Statistics (NCHS). Preliminary estimates large number of demographic
marriage was preceded by cohabitation of first marriage disruption, the characteristics. The most recent
is only appropriate for analyses of first- transition from separation to divorce, SIPP data available were from the
and second-marriage duration. Some remarriage, and second marriage 1996 panel (35). There was no
characteristics do not always have disruption by race/ethnicity and age cohabitation history data collected in
enough cases to use in some analyses. based on the 1995 NSFG were SIPP, so analysis of the transition
For example, parity is measured as the published previously (31). from cohabitation to marriage is
number of children born by the start of impossible.
the analysis interval, and the interval for Data Sources + Cycle 5 of the NSFG was collected
the analysis of first marriage begins at in 1995 and contains full
age 15; the number of women who had There have been several sources of cohabitation and marriage histories
given birth before age 15 was data on marriage, divorce, and as well as a large number of
insufficient for analysis of this variable. cohabitation in the United States in potential characteristics to study
Where possible, analyses were run by recent decades, but few are still active: patterns of cohabitation, marriage,
various different measurements of these and divorce. In addition, the NSFG
+ Until 1995, the NCHS Vital
variables. Analyses of all outcomes are Cycle 5 includes data on the
Statistics program included marriage
presented by religious affiliation and the characteristics of the communities in
and divorce registration data. The
importance of religion. For analyses of which the respondents live, allowing
collection of individual record data
first- and second-marriage disruption, for contextual analysis of
ended with data year 1995, and
results are presented by the wifes age cohabitation, marriage, divorce, and
since then only annual total counts
and by the age difference with her remarriage. Cycles 1 through 5 of
of marriages and divorces have been
husband, and by the wifes race/ NSFG can be pooled to perform
available (32). That system
ethnicity and by the race difference with trend analysis. Unlike most of these
previously gave annual rates of legal
her husband (the age difference with other data systems, NSFG is
marriage and divorce by marriage
partner and race difference with partner currently ongoing. Cycle 6 of the
order and age but had no data on
are not available for analysis of the first NSFG is to be collected in 2002,
the lifetime probability of divorce
cohabitation because of the large with public-use data files expected
by other characteristics and included
amount of missing data in the womans to become available in late 2003.
no data on cohabitation or
report of her first cohabiting partners Further analysis of new data on
separation.
characteristics). these topics collected in 2002 will
+ The U.S. Census Bureaus Current
Basic measures of residence such as therefore be possible.
Population Survey (CPS) previously
region of residence and metropolitan
contained a marital history
status are included here. Other measures Life Tables on Marriage
supplement to the June CPS every 5
of the characteristics of the community
years in 1980, 1985, 1990, and
measured at the census-tract or county There have been numerous studies
1995, but was not continued after
level are also included: the male using life-table techniques to study
1995 (33).
unemployment rate, median family marriage and divorce in the United
+ The National Survey of Families
income, percent of households below States. One study presented first and
and Households, conducted by the
poverty, percent of families receiving second marriage dissolution life tables
University of Wisconsin-Madison
public assistance, percent of adults with based on the 1973 NSFG (23). Another
Center for Demography and
college education, the crime rate in the study (1980) constructed similar tables
Ecology, was a comprehensive
county, and the percent of women on first and second marriage based on
survey covering many aspects of
never-married. the Divorce Registration Area annual
cohabitation and marriage and was
The analysis of each outcome is divorce certificate data (36). Life tables
especially useful because of its
presented by each individual and of marriage, widowhood, and divorce
longitudinal design, allowing for the
community characteristic separately. The have been computed based on published
prediction of outcomes based on
results are descriptive and are not meant census and vital statistics data (37,38).
covariates measured before those
to represent a definitive explanation of Other studies have presented statistics
outcomes. However, the sample was
these outcomes. Further analysis using on marriage and divorce that are
originally drawn in 1987 and the
multivariate techniques may reveal that calculated as cumulative percents, which
last data collection was in 199294,
some of the characteristics in this report
Page 6 [ Series 23, No. 22

are similar to estimates obtained in life and Hispanic women were sampled at typically occur in the lives of women
tables. One such study presented higher rates than were other women. and menthat is, in a life-cycle
cumulative probabilities of remarriage Sampling weights account for order. Each outcome was treated
based on the 1976 NSFG (28). Another differential probabilities of sample independently. Although it is possible to
study presented cumulative proportions selection and for nonresponse, and are combine outcomes in multidecrement
of marriages dissolved based on the adjusted to agree with control totals by life tables (such as the formation of the
1982 NSFG (22). Because the focus of age, race, parity, and marital status first union as either cohabitation or
this report is on the occurrence of provided by the U.S. Census Bureau. marriage, or the end of first cohabitation
certain events (marital disruption, The 10,847 women in the 1995 NSFG in either breakup or marriage), that is
remarriage, etc.) within a specified time represent the 60 million women 1544 beyond the scope of this report.
frame (duration of marriage, duration of years of age in the civilian Previous analysis of divorce and
divorce, etc.), life-table techniques are noninstitutionalized population of the remarriage based on Cycle 4 of NSFG
appropriate for this analysis (23). A United States in 1995. On average, each used a measure of the cumulative
detailed description of life-table woman in the 1995 NSFG represents proportion of marriages disrupted as of
techniques appears in the Methods about 5,500 women in the population, interview to describe the phenomena
section, and a sample life table appears although sample weights vary (43). This statistic is a refinement of a
in Appendix II. considerably from this average value rate, approximating the estimates that
The life tables in this report are depending on the respondents race, age, life-table analysis provides. However, it
based on Cycle 5 of the NSFG, the most and Hispanic ethnicity, the response rate is only a single measure of the
recent available data. In addition, a large for similar women, and other factors cumulative proportion at the time of
number of covariates are examined that (30,39). See Appendix I, Technical interview; life tables provide estimates
were not analyzed in the previous Notes for additional information. of cumulative proportions at every time
studies, including the characteristics of The 1995 NSFG collected complete point in the life course of a marriage.
the communities in which women live. retrospective histories of each womans
We also include cohabitation life tables experiences with cohabitation, marriage, Life TablesThe life table is a tool that
that were not available in prior studies, and divorce, including the beginning demographers and statisticians use most
including the probability of cohabitation and ending dates of each cohabitation often to study mortality, but it is also
disruption, the probability of a and marriage and the outcome of each often applied to the study of marital
cohabitation becoming a marriage, and union (marriage, separation, divorce, or stability. In studying mortality, the
the probability of cohabitation after the widowhood) (40). Given these data, the cohort life table is a summary of the
dissolution of first marriage. probabilities shown in this report can be mortality history of a given cohort from
estimated using life-table techniques. birth to death (a cohort is a group of
Previous analyses of marriage and people born in the same year; e.g., the
Methods 1950 cohort includes persons born in
divorce based on vital statistics have
computed and presented annual rates of 1950), and requires data on the
DataThe national estimates of
marriage and divorce (41,42). Rates are longevity of all cohort members, a span
cohabitation, marriage, and divorce
snapshots of data limited to a specific of more than 100 years. As a result, the
patterns in this report are based on data
year. The life-table analysis in this period life table is typically used as a
from the 1995 NSFG. Cycle 5 of NSFG,
report takes a life-cycle approach to model of what would happen to a given
conducted by CDC/NCHS in 1995, was
estimate the probabilities that: cohort if the age-specific death rates
based on a multistage probability sample
from a certain point in time were to
of the civilian, noninstitutionalized + a woman will get married for the remain fixed for the duration of the
population of women in the United first time, cohorts life (44,45).
States, yielding estimates that are + an intact first cohabitation will make As members of the cohort age, they
representative of women 1544 years of the transition to marriage, are subjected to the age-specific death
age in 1995. Between January and + a first cohabitation will end in rates of successive age categories in the
October 1995, in-home computer- breakup, life table. At each interval, the
assisted personal interviews were + a first marriage will end in age-specific death rate for that interval
conducted with 10,847 women, of separation or divorce, is used to calculate how many members
whom 1,553 were Hispanic women, + a disrupted first marriage will be of the cohort die during that interval.
6,483 were non-Hispanic white women, followed by cohabitation, That number of deaths is subtracted
2,446 were non-Hispanic black women + a separation will result in divorce, from the count of cohort members, and
and 365 were women of other races and + a divorce from first marriage will be the result is the number of cohort
ethnic origins. The overall response rate followed by remarriage, and members who survive to go on to the
was 79 percent (30). + a second marriage will end in next interval. Eventually, the last age
The sample list for the 1995 NSFG
separation or divorce. interval is reached and the last cohort
was selected from households that
members die. One overall measure of
responded to the 1993 National Health These outcomes are presented in
longevity is the proportion who survive
Interview Survey. Non-Hispanic black this report in the order in which they
 Series 23, No. 22 [ Page 7

to specific ages (44). Survivor curves died is unknown, so cases of without first cohabiting or who
can be plotted that show the proportion widowhood are censored (removed from remained unmarried and did not enter a
of the cohort surviving to each the analysis) at the date of the death of new cohabitation by the time of the
successive age category (45,46). the husband. Widowhood is very rare interview are censored. For the
To apply life table analysis to the among women in the age group 1544. transition from separation to divorce,
study of marital (or cohabitation) The mortality of the wives is duration is measured from the date of
stability, the cohort of people is replaced unobservable, as the woman had to have separation from first marriage to the
with a cohort of marriages (or been alive in order to be interviewed. As date the divorce was finalized. Women
cohabitations); age is replaced by union the risk of mortality among women in who never made the transition to
duration, and death is replaced by the age range 1544 is low, this is divorce by the time of the interview are
breakup, separation, or divorce. A unlikely to affect the results censored. For remarriage, duration is
mortality life table is used to analyze substantially. measured from the date of the divorce
death, which is a one-time event that The basic measure used in this to the date of the second marriage.
cannot be reversed, whereas a marital report is the probability that a marriage Women who never remarried by the
life table is used to analyze marriage, or cohabitation will end in separation or time of the interview are censored.
which can occur more than once and divorce. For convenience and brevity in A woman 30 years of age at the
can be reversed. However, there is little this report, this measure is referred to as time of her marriage cannot be included
conceptual difference between the two if the probability of dissolution or the in a measure of the probability of
one considers that the event of a first probability of disruption. In this sense, dissolution after 20 years of marriage,
marriage cannot be reversed (a married dissolution or disruption means to because she would have been 50 years
woman can become unmarried, but break apart or break up. For analysis of age after 20 years of marriage, and
cannot change the fact that she of first- or second-marriage disruption, the maximum age of women in the
experienced the event of a first the duration of the marriage is measured NSFG sample was 44. Because of the
marriage). in months from the start of the marriage age limitation on the sample, the longer
There is an additional issue that until the separation or divorce the period of observation, the younger
must be addressed in order to apply life- (marriages ending in widowhood or still the women must be at marriage to have
table analysis to the study of marital intact at interview are censored). For been 44 years of age or younger when
outcomes. The NSFG sample of women analysis of cohabitation disruption, she was interviewed. Estimates toward
is limited to ages 1544, so the marriage duration is measured from the start of the later durations are therefore biased
histories are incomplete. For respondents the cohabitation until the end of the toward the experiences of younger
whose marriage has not yet ended as of cohabitation, or if the couple married women at marriage. Because younger
interview, the end date of the marriage during the relationship, from the start of age at marriage is associated with a
is unknown, and it is not known how the cohabitation until the separation or higher probability of disruption, this
the marriage will end; therefore the divorce (cohabitations ending in the means that estimates toward the later
duration of the marriage is unknown, death of the partner or still intact at durations may be overestimates of the
and is referred to in statistical literature interview are censored). Cohabitations probability of disruption. To avoid
as censored. Life table procedures that had already made the transition to awkwardness in describing results
allow for the simultaneous analysis of marriage are included in the analysis of affected by this limitation, tables and
complete and incomplete marriage cohabitation disruption because the graphs in this report are truncated as
histories (23). analysis focuses on how long the actual necessary. The events examined in this
Life table analysis can handle relationship endures rather than how report include the first marriage, the
censored cases by keeping such cases in long particular legal definitions endure. transition from first cohabitation to
the analysis as long as they are at risk For the interval to first marriage, marriage, first cohabitation disruption,
of disruption and then dropping them duration is measured from the 15th first marriage dissolution, postmarital
out once the risk is unknown (47). For birthday to the date of first marriage. cohabitation, the transition from
example, when calculating the Women who never married are censored separation to divorce, second marriage,
proportion of marriages that dissolve in at interview. For the transition from and second-marriage dissolution. The
each duration interval, a marriage that cohabitation to marriage, duration is higher the average age at the event, the
has existed for 24 months and still measured from the start of the more truncation is necessary to avoid
exists intact at interview would remain cohabitation to the date of first this potential bias. In the future, the
in the denominator for each duration marriage. Cohabitations ending in death NSFG could address this issue by
interval until 24 months of duration is of the partner or disruption, or still interviewing women up to 54 or 59
reached; after that, the case would no intact and unmarried at interview, are years of age.
longer be used in the calculations. censored. For the interval until The probability of divorce itself is
Widowhood removes a marriage post-marital cohabitation, duration is not always the best measure of marital
from the risk of dissolution. The length measured from the date of the end of instability. While 26.5 percent of women
of time that the marriage would have the first marriage until the start of a new have divorced at the end of 10 years of
endured intact if the husband had not cohabitation. Women who remarried first marriage, 33 percent of all first
Page 8 [ Series 23, No. 22

marriages have disrupted because of Technical Notes.) Non-Hispanic marriage, or after 10 years of marriage.
either separation or divorce at the end of American Indian women are included in Differences that are described in the text
10 years (NSFG Cycle 5, results not analysis of the full sample, but there as statistically significant at certain
shown). Subgroup comparisons of the were not sufficient numbers of durations of marriage may not be
probability of divorce are not non-Hispanic American Indian women statistically significant at other durations
appropriate for subgroups that differ in in the sample to produce reliable of marriage. Differences that are not
the probability that separation will lead estimates separately. discussed in the text are not necessarily
to divorce (48). For example, research Estimates are presented separately statistically insignificant. See the
has shown that the marriages of black for non-Hispanic white women, Technical Notes for details on assessing
women are more likely to end in non-Hispanic black women, and the statistical significance of any
separation than the marriages of white Hispanic women. Analyses by other difference not noted in the text.
women, and that separated black couples characteristics are presented separately Analyses of data by womens
are less likely to make the transition to for non-Hispanic white women, and educational attainment are limited to
divorce than separated white couples non-Hispanic black women, although in women 20 years of age and over at
(23,43). A comparison of the probability some cases the number of non-Hispanic interview because below age 20,
of divorce alone therefore obscures black women in the sample was not education is largely a function of age
some of the difference between these large enough to produce reliable and is often incomplete.
two groups in the probability that a estimates by other covariates. There
marriage will dissolve. For this reason, were enough Hispanic women in the Community Distributions
in this report, marital disruption is sample to present analysis by other by Race/Ethnicity
defined as either separation or divorce, characteristics separately for Hispanic
and a second analysis examines the women for only two outcomes: the As will be shown, the race/ethnicity
probability that separated women will interval until first marriage and the differences in marital and cohabitational
divorce. stability of first marriage. For stability found in this report are
Appendix II presents an example convenience in writing, in the text of substantial, and the trend analysis
life table for the duration of first this report, non-Hispanic white women suggests that the differences are
marriage and describes in detail each are often referred to as white and increasing over time, such that marital
part of the life table and its role in the non-Hispanic black women are often instability has leveled off for
generation of survival statistics. In the referred to as black. The full labels non-Hispanic white women but
following analysis, for the sake of are always used in the tables and continues to increase for non-Hispanic
brevity, only the cumulative proportion graphs. The statistics in this report were black women. In the analyses of marital
dissolved at the beginning of selected computed using the LIFETEST and cohabitational outcomes, the
intervals is presented and compared procedure in Version 8 of PC-SAS (49). consistent finding is that less affluent
across subgroups. (The intervals that The software package SUDAAN, communities as indicated by lower
have been selected are consistent across Version 7.5.6 was used to compute the median family income and percent
outcomes: after 1 year, after 3 years, standard errors of the statistics (50). The college educated and higher
after 5 years, after 10 years, after 15 point estimates derived in SAS and unemployment, poverty, and welfare are
years.) The cumulative proportion SUDAAN are identical, but the standard associated with lower marital and
dissolved after a specified period is a errors computed in SUDAAN correct cohabitational stability. An examination
more stable estimate than the estimates for the complex survey design of the of community distributions by
of individual probabilities of dissolution NSFG Cycle 5. race/ethnicity may suggest avenues for
within each period (23). Although this The statistical significance of further exploration of the race
explanation and the example life table in differences in the probabilities examined differences in marital and cohabitational
the appendix focus on marital duration in this report is assessed by comparing stability.
as the dependent variable of interest, the the boundaries of confidence intervals Table A shows the percentage
methodology is easily adapted to around each estimate (see the Technical distributions of community
examine other cohabitation and marital Notes for further details). Differences characteristics for all women and
outcomes. presented in the text are statistically separately for Hispanic, white, and black
The analyses of the interval until significant at the 5-percent level, women. The community characteristics
first marriage and of first marriage indicating that if the difference were are classified into three categories: the
stability are the only analyses in this merely the result of random chance and top 25 percent, the middle 50 percent,
report in which there were sufficient did not reflect a true difference in the and the bottom 25 percent. The
numbers of non-Hispanic Asian women general population, the difference would percentages in the Total column do
in the NSFG sample to generate reliable only be observed in less than 5 percent not always equal 25, 50, and 25,
estimates. In all other analyses in this of all possible samples. In general, because the value at the quartile does
report, non-Hispanic Asian women are results are described at specific points in not always split the sample up into
included in analysis of the full sample time, for example, the probability of exact quartiles. For example, if the 25th
but are not analyzed separately. (See marital disruption after 5 years of percentile value of median family
 Series 23, No. 22 [ Page 9

Table A. Number of women 1544 years of age (in thousands) and percent distribution, by race/ethnicity and contextual variables:
United States, 1995

Race/ethnicity

Number Non-Hispanic Non-Hispanic

Contextual variable (1,000s)1 Total Hispanic white black

Male unemployment
Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 18,505 30.7 15.8 35.7 15.5
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 28,825 47.9 45.7 50.9 35.5
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 12,870 21.4 38.5 13.4 49.0

Median family income

Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 15,767 26.2 39.9 19.8 51.4
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 30,452 50.6 44.4 54.2 37.8
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 13,981 23.2 15.7 26.0 10.8

Percent below poverty

Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 14,465 24.0 13.0 28.1 9.1
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 30,322 50.4 43.4 54.6 34.3
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 15,414 25.6 43.6 17.3 56.6

Percent receiving welfare

Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 15,695 26.1 15.2 30.4 11.5
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 30,059 49.9 38.7 54.8 33.9
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 14,447 24.0 46.1 14.8 54.6

Percent of adults college-educated

Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 16,781 27.9 38.9 23.0 48.8
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 29,711 49.4 46.7 51.6 39.5
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 13,708 22.8 14.5 25.5 11.7

Crime rate
Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 14,591 25.0 7.1 30.9 12.3
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 29,277 50.2 53.2 50.3 45.6
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 14,469 24.8 39.6 18.8 42.1

Percent of women never-married

Bottom 25 percent . . . . . . . . . . . . . . . . . . . . 16,038 26.6 14.4 32.6 8.3
Middle 50 percent . . . . . . . . . . . . . . . . . . . . 30,221 50.2 50.8 52.8 36.7
Top 25 percent . . . . . . . . . . . . . . . . . . . . . . 13,942 23.2 34.8 14.7 55.1

Metropolitan status
Central city . . . . . . . . . . . . . . . . . . . . . . . . 18,550 30.8 51.0 22.8 55.2
Other SMSA2 . . . . . . . . . . . . . . . . . . . . . . . 29,303 48.7 41.3 52.9 31.7
Not SMSA2 . . . . . . . . . . . . . . . . . . . . . . . . 12,347 20.5 7.7 24.3 13.2

1
The weighted number of women is an estimate of the total population size and does not reflect sample size.
2
SMSA is standard metropolitan statistical area.

income is $10,000, but there are a large unemployment communities and only women are also disproportionately
number of cases with values of $10,000, 13 percent live in high-unemployment present in less affluent areas (table A).
there may not be a clear distinction at communities, compared with 16 percent It will be shown that median family
exactly the 25th percentile. of black women in low-unemployment income in the community is associated
Table A shows that non-Hispanic areas and almost one-half (49 percent) with the probability of first marriage
white women are disproportionately of black women in high-unemployment disruption, such that marriages are more
present in affluent neighborhoods and areas (table A). Only 9 percent of black likely to fail if the woman lives in a
that non-Hispanic black and Hispanic women live in low-poverty communities low-income community. This
women are disproportionately present in and 57 percent live in high-poverty relationship is similar among white and
less-affluent neighborhoods. Roughly areas, compared with 28 percent of black women. Because black women
31 percent of all women live in white women in low-poverty areas and tend to live in communities with low
low-unemployment communities and 17 percent of white women in median family income and communities
21 percent live in high-unemployment high-poverty areas. The distribution of with low income are associated with a
communities, but among white women, Hispanic women falls between that of larger probability of marital disruption,
almost 36 percent live in low- white and black women, but Hispanic black women have a higher chance of
Page 10 [ Series 23, No. 22

Table B. Number of women 1544 years of age (in thousands) and percent distribution, by past cohabitation and marital status and by
age at interview and race/ethnicity: United States, 1995

Past cohabitation and marital status

Never married Ever married

Number Never Ever Never Ever

Age at interview and race/ethnicity (1,000s)1 Total cohabited cohabited cohabited cohabited

Total . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60,201 100.0 27.5 10.2 31.4 30.9

Age at interview
1519 . . . . . . . . . . . . . . . . . . . . . . . . . . . 8,961 100.0 88.6 7.0 2.6 1.9
2024 . . . . . . . . . . . . . . . . . . . . . . . . . . . 9,041 100.0 45.5 20.2 16.2 18.1
2529 . . . . . . . . . . . . . . . . . . . . . . . . . . . 9,693 100.0 20.3 15.4 30.4 33.9
3034 . . . . . . . . . . . . . . . . . . . . . . . . . . . 11,065 100.0 10.8 9.3 37.8 42.1
3539 . . . . . . . . . . . . . . . . . . . . . . . . . . . 11,211 100.0 7.1 6.4 42.9 43.6
4044 . . . . . . . . . . . . . . . . . . . . . . . . . . . 10,230 100.0 5.5 4.1 51.5 38.9

Race/ethnicity
Hispanic . . . . . . . . . . . . . . . . . . . . . . . . . . 6,702 100.0 28.2 10.4 35.1 26.3
White non-Hispanic . . . . . . . . . . . . . . . . . . . 42,522 100.0 24.7 8.9 32.7 33.7
Black non-Hispanic . . . . . . . . . . . . . . . . . . . 8,210 100.0 39.7 17.3 20.2 22.9
Other non-Hispanic2 . . . . . . . . . . . . . . . . . . . 2,767 100.0 33.2 8.3 35.1 23.4

1
The weighted number of women is an estimate of the total population size and does not reflect sample size.
2
Includes Asian and Pacific Islander women and American Indian women, not shown separately.

marital dissolution than white women, classified in table B as never married or 100 percent. Although current cohabitors
who are less likely to live in ever married, with each group further could be never married or formerly
communities with low income. split into two subgroups separating the married, they would not be included in
However, within low-income never cohabited from the ever cohabited. the never married or formerly married
communities, black women still have a These four subgroups are mutually categories because those groups are
greater probability of marital disruption exclusive and exhaustive, summing to restricted to respondents not cohabiting
than white women in low-income 100 percent. at interview in order to focus on the
communities, so some of the race Almost 28 percent of women 1544 proportions of women currently in a
difference remains unexplained. years of age have never married nor marriage or cohabitation.
To fully explore the effects of cohabited (table B). This percentage is Roughly 50 percent of women
individual and community characteristics considerably larger for young women 1544 years of age are currently married
requires multilevel modeling, which is and decreases as age increases. About and 7 percent of women 1544 years of
beyond the scope of this report. 62 percent of women have ever been age are currently cohabiting (table C).
Associations between individual married, one-half of whom have ever One third of women 1544 years of age
outcomes and community characteristics cohabited and one-half of whom have are not cohabiting and have never
could be influenced by unobserved never cohabited. The remaining married. The remaining 10 percent are
factors. The analyses by community 10 percent have cohabited, but never not cohabiting and are formerly married
characteristics are not meant to represent married. Non-Hispanic white women are (separated, divorced, or widowed). The
full explanations of the outcomes more likely to have experienced both percent currently cohabiting is larger for
studied in this report. Researchers are cohabitation and marriage, while young adults in their twenties and then
encouraged to use these results as non-Hispanic black women are more decreases as age increases. The most
starting points to follow up with more likely to have experienced neither striking differences by race/ethnicity are
extensive analysis. cohabitation nor marriage (table B). the higher percent not cohabiting and
Table C shows the distribution of never married and the lower percent
women 1544 years of age in 1995, by currently married among non-Hispanic
Results current cohabitation and marital status at black women. In the remaining text of
interview, age at interview, and this report, non-Hispanic white women
race/ethnicity. Current cohabitation and are often referred to as white and
Cohabitation and Marital marital status is classified as currently non-Hispanic black women are often
Status cohabiting or not currently cohabiting. referred to as black. The full labels
The category not currently cohabiting is are always used in the tables and
Table B shows the distribution of further split into the never married, graphs.
women 1544 years of age in 1995 by formerly married, or currently married.
past cohabitation and marital status, age These four subgroups are mutually
at interview, and race/ethnicity. Past
exclusive and exhaustive, summing to
cohabitation and marital status is
 Series 23, No. 22 [ Page 11

Table C. Number of women 1544 years of age (in thousands) and percent distribution, by current cohabitation and marital status and
by age at interview and race/ethnicity: United States, 1995

Current cohabitation and marital status

Not cohabiting

Age at interview Number Currently Never Formerly Currently

and race/ethnicity (1,000s)1 Total cohabiting married married married

Total . . . . . . . . . . . . . . . . . . . . . 60,201 100.0 7.0 33.4 10.3 49.3

Age at interview
1519 . . . . . . . . . . . . . . . . . . . . 8,961 100.0 4.1 91.5 0.6 3.8
2024 . . . . . . . . . . . . . . . . . . . . 9,041 100.0 11.2 56.1 5.5 27.2
2529 . . . . . . . . . . . . . . . . . . . . 9,693 100.0 9.8 28.9 8.8 52.5
3034 . . . . . . . . . . . . . . . . . . . . 11,065 100.0 7.5 16.2 11.6 64.7
3539 . . . . . . . . . . . . . . . . . . . . 11,211 100.0 5.3 11.9 15.0 67.9
4044 . . . . . . . . . . . . . . . . . . . . 10,230 100.0 4.4 8.8 18.1 68.6

Race/ethnicity
Hispanic . . . . . . . . . . . . . . . . . . . 6,702 100.0 8.2 32.8 11.6 47.4
White non-Hispanic . . . . . . . . . . . . 42,522 100.0 7.0 29.4 9.3 54.3
Black non-Hispanic . . . . . . . . . . . . 8,210 100.0 6.9 52.5 15.5 25.2
Other non-Hispanic2 . . . . . . . . . . . 2,767 100.0 4.6 39.1 7.6 48.8

1
The weighted number of women is an estimate of the total population size and does not reflect sample size.
2
Includes Asian and Pacific Islander women and American Indian women, not shown separately.

The Probability of First actually the month of the 15th birthday, but the differences tend to converge at
Marriage so a 3-year interval ends in the month later durations as most women
just before the 18th birthday and a eventually marry.
Tables 1 and 2 show the probability 5-year interval ends in the month just Table 1 shows that 8 percent of
that a woman marries for the first time before the 20th birthday. A recent census women married for the first time by the
by characteristics of the woman and her report estimated that 90 percent of 18th birthday, 25 percent married by the
community. Tables 3 and 4 show these women will marry at some time in their 20th birthday, and 76 percent married by
estimates for Hispanic women, tables 5 lives (51); because most women the 30th birthday. Figure 1 shows the
and 6 show the estimates for non- eventually marry, the tables presented probability of marriage over time by
Hispanic white women, and tables 7 and here basically show differences in the race/ethnicity: Hispanic and non-
8 show the estimates for non-Hispanic timing of first marriage by Hispanic white women are more likely
black women. These tables show the characteristics of the woman and her to marry by age 25 than non-Hispanic
probabilities of marriage at specific community. A particular variable may black or Asian women, but by age 30,
durations since age 15, the starting point show a significant difference in the non-Hispanic Asian women have caught
for this analysis. The starting point is proportion of women married by age 18, up to Hispanic and white women. Black
women are significantly less likely to
have married by age 30 than any other
Hispanic Non-Hispanic white group (table 1 and figure 1). Figure 2
90 Non-Hispanic black Non-Hispanic Asian shows the probability of marriage by
80 age 18 and by age 30, by race/ethnicity.
Early marriage is more likely for
70
Hispanic women, followed by white
Percent married

60 women, and is less likely for black

50 women and Asian women. Marriage by
age 30 is considerably lower for
40
non-Hispanic black women, with
30 virtually no differences among the other
20
groups (figure 2).
Early marriage (i.e., before the 18th
10 birthday) is more likely among women
0 with less than a high school education at
15 16 18 20 25 30
interview, and among women whose
Age in years mothers had less than a high school
Figure 1. Probability of first marriage by age and race/ethnicity: United States, 1995
education (table 1). The pattern of
differences is similar for the two
Page 12 [ Series 23, No. 22

receipt of public assistance. The

Hispanic Non-Hispanic white differences in figure 4 are substantial:
Non-Hispanic black Non-Hispanic Asian The probability of early marriage is
90
81 200 percent higher in high-poverty
80 77 77 communities compared with low-poverty
70 communities.
Marriage by age 30 is less likely in
Percent married

60
52 metropolitan areas than in
50 nonmetropolitan areas. For all women,
40 and among Hispanic, white, and black
women, the probability of first marriage
30
is lowest in central cities and highest in
20 nonmetropolitan areas (figure 5).
13
10 8 Figure 5 also shows that within each
5 3
0
category of metropolitan status,
Age 18 Age 30 non-Hispanic black women are
significantly less likely to marry by age
Figure 2. Probability of first marriage by ages 18 and 30 by race/ethnicity: United States, 30 than Hispanic or non-Hispanic white
1995 women.

Very important Somewhat important Not important

The Probability That an Intact
80 First Cohabitation Makes the
70 Transition to Marriage
60 Tables 9 and 10 show the
probability that the first premarital
Percent married

50 cohabitation becomes a marriage by

40
characteristics of the woman and her
community. Tables 11 and 12 show
30 these estimates for non-Hispanic white
women, and tables 13 and 14 show
20
these estimates for non-Hispanic black
10 women. Table 9 shows that for all
women, the probability of a first
0
15 16 18 20 25 30 premarital cohabitation becoming a
Age in years marriage is 58 percent after 3 years of
cohabitation and 70 percent after 5 years
Figure 3. Probability of first marriage by age and importance of religion: United States,
of cohabitation. This means that
1995
58 percent of cohabitations that have
lasted at least 3 years have made the
measurements of education, but the Differences by these variables are
transition to marriage by that time and
differences are larger by the womans smaller and nonsignificant among
that 70 percent of cohabitations that
own education than by her mothers non-Hispanic black women, although
have lasted for 5 years have made the
education. Because of data quality issues the differences by education and
transition to marriage by that time.
(the higher level of imputation of mothers education remain statistically
Table 9 and figure 6 show that the
mothers education), the respondents significant (table 7).
probability that a first premarital
own education is retained for the Figure 3 shows the probability of
cohabitation becomes a marriage is
remaining analysis in this report and first marriage by the importance of
higher for white women, lower for black
mothers education is dropped. religion. Women who reported that their
women, and intermediate for Hispanic
Other characteristics that are religion is not important are less likely
women (figure 6). The probability that
associated with early marriage include to marry than other women at all ages
the first cohabitation becomes a
lower family income at interview, being through age 30 (figure 3).
marriage within 5 years is 75 percent for
affiliated with a fundamentalist Early marriage is more common in
white women, 61 percent for Hispanic
Protestant faith, and living in the South less-affluent communities. Figure 4
women, and only 48 percent for black
(table 1). Among Hispanic women, shows that marriage by age 18 is more
women (table 9). Table 9 also shows that
affiliation with either fundamentalist likely in communities with higher male
the probability of the transition to
Protestant or other Protestant faiths is unemployment, lower median family
marriage is higher for women with
associated with early marriage (table 3). income, higher poverty, and higher
higher family income (the measurement
 Series 23, No. 22 [ Page 13

of family income in the NSFG Cycle 5

15 includes any income from the
Low Middle High
cohabitating partner). Figure 7 reveals
12
that this difference by family income is
11 11
10 much larger among black women than
Percent married

10
9 white women: Among white women,
8 8 8 there is only a nonsignificant 4
6
percentage point difference in the
probability of the transition to marriage
5
4 4 between the low-income and high-
3 income groups, whereas the difference is
32 percentage points among black
women (figure 7).
0
Unemployment Income Poverty Welfare Although differences between
specific denominations are small, the
Figure 4. Probability of first marriage by age 18 by community male unemployment rate, probability of the transition to marriage
median family income, percent below poverty, and percent receiving public assistance:
differs significantly between women
United States, 1995
with any religious affiliation and women
with no religious affiliation. Figure 8
Central city Other SMSA1 Not SMSA1 shows that the probability of the
transition to marriage within 5 years is
90 88 87
84 82 65 percent for women with no religious
79 79
80 affiliation and 72 percent for women
73 72
70 66
with any religious affiliation, and the
difference is larger among white
Percent married

60 56 58
women. Figure 9 shows that among
50 48 white women, women to whom religion
40 is not important are less likely to make
the transition to marriage than women to
30
whom religion is somewhat or very
20 important, although the difference
10 converges to nonsignificance at later
0 durations of cohabitation.
All races Hispanic Non-Hispanic Non-Hispanic A greater probability of making the
white black
1SMSA is standard metropolitan statistical area. transition from cohabitation to first
marriage within 5 years is also
Figure 5. Probability of first marriage by age 30 by race/ethnicity and metropolitan status: associated with higher education, having
United States, 1995 a two-parent intact family throughout
childhood, having no children at
cohabitation, and having children after
100 Hispanic Non-Hispanic white Non-Hispanic black the start of the cohabitation as compared
90 with before the cohabitation or never
80
(table 9).
The probability that a first
70 premarital cohabitation makes the
Percent married

60 transition to marriage within 5 years is

higher in communities with a lower
50
male unemployment rate, a higher
40 median family income, a lower percent
30 of families below poverty, and a lower
percent of households receiving public
20
assistance (table 10 and figure 10). The
10 probability of the transition to marriage
0 is higher in communities with a higher
0 1 3 5 10 percent college-educated and a higher
Duration in years percent of women never-married, is
lower in central cities than in other
Figure 6. Probability that an intact cohabitation makes the transition to first marriage by
duration of cohabitation and race/ethnicity: United States, 1995 areas, and does not differ significantly
Page 14 [ Series 23, No. 22

by the crime rate in the county

Less than $25,000 $25,000-$49,999 $50,000 or more
(table 10). In a typical comparison in
80 75 77 table 10, the difference between an
71 73 73 71 extreme category and the middle
70
62 category is not significant, whereas the
60 difference between the bottom quarter
Percent married

50
50 and the top quarter is significant.
40
39 Among white women, the effects of
the community male unemployment
30 rate, the percent below poverty, and the
20 percent college-educated fade to
10 nonsignificance (table 12). Among black
women, only the median family income
0
All races Non-Hispanic Non-Hispanic and percent college-educated fade to
white black nonsignificance (table 14). Figure 11
shows that the effect of community
Figure 7. Probability that an intact cohabitation makes the transition to first marriage
within 5 years by race/ethnicity and family income: United States, 1995
male unemployment is stronger among
black women than among white women:
The difference in the probability of the
90 No religious affiliation Any religious affiliation
transition to marriage within 5 years
between low-unemployment and
80 77
72
high-unemployment communities is
70 65 65 twice as high among black women
(27 percentage points) than among white
Percent married

60
women (13 percentage points, figure 6,
50 based on tables 12 and 11).
40
The Probability of First
30
Cohabitation Disruption
20
Tables 15 and 16 show the
10 probability of first premarital
0 cohabitation disruption by characteristics
All races Non-Hispanic white of the woman and her community.
Tables 17 and 18 show these estimates
Figure 8. Probability that an intact cohabitation makes the transition to first marriage
for non-Hispanic white women, and
within 5 years by religious affiliation: United States, 1995
tables 19 and 20 show these estimates
for non-Hispanic black women.
Cohabitation disruption includes
100 Very important Somewhat important Not important cohabitations that made the transition to
90 marriage and then disrupted, because we
are more interested in seeing how long
80
the relationship endures than in seeing
70 how long certain legal definitions of the
Percent married

60 relationship endure. The probability of

first cohabitation disruption for all
50
women is 39 percent within 3 years and
40 49 percent within 5 years (table 15).
30 Figure 12 shows that black women are
more likely to experience a cohabitation
20
disruption than either white or Hispanic
10 women. The difference in the
0 probability of cohabitation disruption
0 1 3 5 10 between Hispanic and white women is
Duration in years
only statistically significant at 3 years
Figure 9. Probability that an intact cohabitation makes the transition to first marriage by duration of cohabitation (table 15).
duration of cohabitation and importance of religion: Non-Hispanic white women, The cohabitations of women who
United States, 1995 were at least 25 years of age at the start
 Series 23, No. 22 [ Page 15

100
of the cohabitation are less likely to
Low Middle High disrupt than those of women younger
90
80 80
than 25 years of age at cohabitation.
80 78 78
72 73 72 73 Figure 13 shows that there are very
70 66 small differences among women under
Percent married

60 58 58
55 18, 1819, and 2024 years of age, and
50 a much larger difference between these
40
groups and women 25 years of age and
over. Figure 14 reveals that the
30
difference by age at cohabitation is
20
twice as large among white women as it
10 is among black women. The difference
0 between the two extreme age categories
Unemployment Income Poverty Welfare
is 31 percentage points for white women
Figure 10. Probability that an intact cohabitation makes the transition to first marriage and only 15 percentage points among
within 5 years by community male unemployment rate, median family income, percent black women, based on tables 17 and
below poverty, and percent receiving public assistance: United States, 1995 19.
Women who have ever been forced
to have intercourse at some time before
the cohabitation are more likely to
90 Low unemployment Medium unemployment High unemployment experience cohabitation disruption than
79 those who have not been forced to have
80 78 76
72 intercourse (table 15). Figure 15 reveals
70 66 66 66 that this difference is larger for white
women than for black women. After 5
Percent married

60
53 years of cohabitation, the probability of
50
disruption is 60 percent for white
40 39 women who have ever been forced to
30
have intercourse and 46 percent for
white women who have never been
20 forced to have intercourse, a difference
10 of 14 percentage points; the analogous
difference among black women is
0
All races Non-Hispanic white Non-Hispanic black 5 percentage points (figure 15, based on
tables 17 and 19). These data do not
Figure 11. Probability that an intact cohabitation makes the transition to first marriage identify or classify whether the forced
within 5 years by race/ethnicity and male unemployment rate in the community: intercourse was with the cohabiting
United States, 1995
partner or someone else. It is therefore
not clear what role, if any, the forced
intercourse had in the disruption of the
80 cohabitation. This finding, like many
Hispanic Non-Hispanic white Non-Hispanic black
other findings in this report, deserves
70 further study.
60
Women who have ever had GAD
are more likely to experience a
Percent disrupted

50 cohabitation disruption than those who

have never had GAD (table 15). The
40
American Psychiatric Association
30 defines generalized anxiety disorder as
unrealistic or excessive anxiety or worry
20
about two or more life circumstances for
10 6 months or longer (52). Women were
classified as having ever suffered from
0 GAD if they reported that they had
0 1 3 5 10
Duration in years suffered for at least 6 months from
worry or anxiety, including feelings of
Figure 12. Probability that the first cohabitation breaks up by duration of cohabitation and restlessness, feeling keyed up or on
race/ethnicity: United States, 1995 edge, irritability, a pounding or racing
Page 16 [ Series 23, No. 22

Women with no religious affiliation are

Under age 18 Age 18-19 Age 20-24 Age 25 and over more likely to experience a cohabitation
70
disruption than Catholic or
60 nonfundamentalist Protestant women.
Although parity and the wantedness
Percent disrupted

50
status of children present at the start of
40 the cohabitation are not significantly
30 related to the probability of disruption,
the timing of the first birth matters:
20
Women whose first birth was more than
10 7 months after the cohabitation began
0
are less likely to experience a
0 1 3 5 10 cohabitation disruption than women
Duration in years whose first birth was before the union,
Figure 13. Probability that the first cohabitation breaks up by duration of cohabitation and
and women who have never had a birth
age at the beginning of cohabitation: United States, 1995 are more likely to experience disruption.
Cohabitations in the South are slightly
more likely to disrupt than those in the
Northeast and Midwest, although the
difference is not statistically significant
90 Under age 18 Age 18-19 Age 20-24 Age 25 and over until after 10 years of cohabitation
80 77 77 (table 15).
70 72 The effect of work status at
70 68 68 68
62 cohabitation differs between white and
60
Percent disrupted

60 58 black women: White women not

50 working at the time the cohabitation
43 began are more likely to experience
40 39
cohabitation disruption after 10 years
30 than those working at cohabitation (and
20 full-time/part-time status makes no
difference), but black women working
10
part-time at cohabitation are more likely
0 to experience cohabitation disruption
All races Non-Hispanic white Non-Hispanic black
after 10 years, followed by those not
Figure 14. Probability that the first cohabitation breaks up within 10 years by race/ working, and full-time workers
ethnicity and age at the beginning of cohabitation: United States, 1995 (tables 17,19).
The probability of cohabitation
disruption is higher in communities with
higher unemployment, lower median
Ever had forced intercourse Never had forced intercourse family income, and a higher percent of
70

60 58 60 59 families either below poverty level or

54 receiving public assistance (table 16 and
Percent disrupted

50 47 46 figure 16). Cohabitation disruption is

40 also more likely in communities with a
higher percent of women never-married
30
and in central cities (table 16). Among
20 white women, the probability of
10 cohabitation disruption is higher in
counties with higher crime (table 18).
0
All races Non-Hispanic white Non-Hispanic black Figure 17 shows that the effect of the
community male unemployment rate is
Figure 15. Probability that the first cohabitation breaks up within 5 years by race/ethnicity
similar for white and black women: The
and forced intercourse before cohabitation: United States, 1995
difference in the probability of
disruption between low-unemployment
heart, getting tired easily, trouble falling of cohabitation disruption. Higher
and high-unemployment areas is
or staying asleep, and feeling faint. family income and being raised in an
12 percentage points among white
Other characteristics of individuals intact two-parent family are associated
women and 10 percentage points among
are also associated with the probability with a lower probability of disruption.
 Series 23, No. 22 [ Page 17

The Probability of First

80 Low Middle High
Marriage Disruption
69 71 69 69
70
61 61 Tables 21 and 22 show the
58 60 59
60 56 probability of first marriage disruption
Percent disrupted

55 54
50 due to separation or divorce by
40
characteristics of the woman and her
community. Tables 23 and 24 show
30 these estimates for Hispanic women,
20 tables 25 and 26 show these estimates
10 for non-Hispanic white women, and
tables 27 and 28 show these estimates
0
Unemployment Income Poverty Welfare for non-Hispanic black women. After 5
years, 20 percent of all first marriages
Figure 16. Probability that the first cohabitation breaks up within 10 years by community have disrupted, due to either separation
male unemployment rate, median family income, percent below poverty, and percent
or divorce. After 10 years, one-third of
receiving public assistance: United States, 1995
first marriages have disrupted (table 21).
After 10 years, 32 percent of white
womens first marriages have dissolved,
and 34 percent of Hispanic womens
Low unemployment Medium unemployment High unemployment first marriages have dissolved (table 21).
80 76 In contrast, 47 percent of black womens
69 69 70 first marriages have dissolved after 10
70 66
60 60
years. Asian womens first marriages
60 58 57 dissolve at a considerably slower rate:
Percent disrupted

50 After 10 years, only 20 percent have

40
disrupted. Figure 18 shows the similarity
of white and Hispanic womens
30 probabilities of disruption, the higher
20 likelihood of disruption among black
10 women and the lower likelihood of
disruption among Asian women.
0
All races Non-Hispanic white Non-Hispanic black Age at marriage is associated with
the risk of marital disruption. Table 21
Figure 17. Probability that the first cohabitation breaks up within 10 years by race/ shows that after 10 years of marriage,
ethnicity and male unemployment rate in the community: United States, 1995 48 percent of first marriages of brides
under age 18 have disrupted, compared
with only 24 percent of those to brides
at least age 25 at marriage (table 21).
Hispanic Non-Hispanic white Women in the youngest age category are
60 Non-Hispanic black Non-Hispanic Asian twice as likely to experience marital
50 disruption within 10 years as women in
Percent disrupted

the oldest age category. Figure 19 shows

40
that this difference is even larger among
30 white women, a difference of
28 percentage points. Among black
20
women, the difference is 17 percentage
10 points. Figure 19 also shows that there
0 is virtually no difference by age at
0 1 3 5 10 15 marriage among Hispanic women.
Duration in years Figure 20 shows the relationship of
first marital disruption with the
Figure 18. Probability that the first marriage breaks up by duration of marriage and
race/ethnicity: United States, 1995
importance of religion. The graph shows
that the higher the importance attached
black women. Within each of black women are more likely to to religion, the lower the likelihood of
unemployment group, the cohabitations disrupt than those of white women. marital disruption (although the
Page 18 [ Series 23, No. 22

chance of marital dissolution after 10

Under age 18 Age 18-19 Age 20-24 Age 25 and over years of marriage; those who did not
60 55
52 have an intact family have a 41-percent
48 50
50
45 chance of dissolution, a difference of
Percent disrupted

40
40 38 39 38 12 percentage points (figure 21). The
36
33 31 analogous difference among black
30 29 27
24 women is 13 percentage points, and
22
20 among Hispanic women, 17 percentage
points. This finding is consistent with
10 the notion of the intergenerational
0 transmission of divorce (53).
All races Hispanic Non-Hispanic white Non-Hispanic black Marital dissolution is more likely
for a woman who was ever forced to
Figure 19. Probability that the first marriage breaks up within 10 years by race/ethnicity
have intercourse by a man at some time
and age at the beginning of marriage: United States, 1995
in her life before she was married
(table 21). Figure 22 shows that this
effect is found for Hispanic, white, and
black women. The difference in
60 Very important Somewhat important Not important probabilities between women who have
50 and have not ever been forced to have
Percent disrupted

intercourse is large: 17 percentage points

40 for Hispanic women, 22 percentage
30 points for non-Hispanic white women,
and 20 percentage points for non-
20
Hispanic black women (figure 22). It
10 was noted earlier that the effect of
0
forced intercourse on the probability of
0 1 3 5 10 15 cohabitation disruption was larger for
Duration in years white women than for black women, but
for first marriage dissolution, the effect
Figure 20. Probability that the first marriage breaks up by duration of marriage and
appears to be very similar for the two
importance of religion: United States, 1995
racial groups.
Figure 23 shows that among
Hispanic, white, and black women,
Intact two-parents Otherwise those whose first birth was more than 7
60
53 months after first marriage have the
50 45 lowest chance of marital disruption.
Percent disrupted

43 41
40
40 Differences among the other three
categories of timing of first birth are not
30 29 28 29
statistically significant, but the data
20 suggest that for Hispanic and black
women, those whose first birth is within
10 7 months of marriage are the most
0 likely to experience marital disruption
All races Hispanic Non-Hispanic white Non-Hispanic black (tables 23,25,27).
Women who have ever suffered
Figure 21. Probability that the first marriage breaks up within 10 years by race/ethnicity
and intact status of family of origin: United States, 1995 from GAD are more likely to experience
first marital disruption than women who
difference between the very important The probability of first marriage have not had GAD. Figure 24 shows a
and the somewhat important groups is dissolution is substantially higher for difference of 16 percentage points after
not statistically significant) (table 21 and women who did not grow up in a 15 years of marriage. This difference is
figure 20). Women who reported that two-parent intact family (table 21). smaller for black women than for white
religion is not important to them are Figure 21 shows that this effect is women. The difference in the
more likely to have experienced first similar for Hispanic women, white probability of marital disruption between
marital dissolution than women who women, and black women. Among women who have ever suffered from
reported that religion is somewhat white women, those who were raised in GAD and those who have not is
important or very important. an intact family have a 29-percent 17 percentage points for non-Hispanic
white women, 10 percentage points for
 Series 23, No. 22 [ Page 19

be made is limited because of small sample

Ever had forced intercourse Never had forced intercourse sizes. While specific pairings such as
70
63 white/black or black/Asian are not
60 shown, comparisons that can be shown
53 51 reliably are presented in table 21.
Percent disrupted

50 49
43 White/any other couples have similar
40
32
chances of marital disruption as all
30 29
30 different race couples, which is not
20
surprising as the majority of different
race couples are white/any other
10 pairings. Black/any other couples appear
0 to have chances of marital disruption
All races Hispanic Non-Hispanic white Non-Hispanic black similar to those for all black couples.
Figure 22. Probability that the first marriage breaks up within 10 years by race/ethnicity
Other individual characteristics of
and forced intercourse: United States, 1995 women that are associated with a greater
probability of marital dissolution include
lower education, lower family income,
not working at the beginning of
marriage, working full time as opposed
Before first marriage 0-7 months after to working part time at marriage, having
More than 7 months after Never had first birth
40 no religious affiliation, already having
36
35 one child or more at the start of the
31
Percent disrupted

30 29 30
28 marriage, and living in the South
26
25 23 22 23 (table 21). First marriages that were
20 18 preceded by cohabitation are more likely
15 to disrupt than those that were not
15 12
10 preceded by cohabitation. The only
5 variable in table 21 that does not show a
0 significant effect on the probability of
Hispanic Non-Hispanic white Non-Hispanic black first marital dissolution is the age
Figure 23. Probability that the first marriage breaks up within 5 years by race/ethnicity
difference between husband and wife.
and timing of first birth: United States, 1995 First marriages are more likely to
disrupt in communities with higher
unemployment, lower median family
income, and a higher percent of families
Ever had GAD 1 Never had GAD1 below poverty level or receiving public
70
64 assistance (table 22 and figure 26). First
58
60 55 54 54
marriages are also more likely to disrupt
in central cities, and in communities
Percent disrupted

50 38
39
with a lower percent college-educated, a
40 37
higher crime rate, and a higher percent
30 of women never-married (table 22).
20
Figure 27 shows that the effect of
community median family income is
10 similar for Hispanic, white, and black
0 women. The difference in the
All races Hispanic Non-Hispanic white Non-Hispanic black
1GAD is generalized anxiety disorder. probability of marital disruption between
low-income and high-income
Figure 24. Probability that the first marriage breaks up within 15 years by race/ethnicity communities is 12 percentage points for
and generalized anxiety disorder: United States, 1995 Hispanic women, 20 percentage points
for white women, and 23 percentage
non-Hispanic black women, and spouses are of different race/ethnicity points for black women.
20 percentage points for Hispanic (table 21 and figure 25). After 10 years
women (figure 24). of marriage, interracial marriages have a The Probability of Cohabitation
First marriages in which the 41-percent chance of disruption and After the End of First Marriage
husband and wife are both members of same-race marriages have a 31-percent
the same race/ethnicity are more likely chance of disruption (table 21). The Tables 29 and 30 show the
to succeed than those in which the number of specific comparisons that can probability of postmarital cohabitation
Page 20 [ Series 23, No. 22

contextual variables. The probability of

Different race Same race postmarital cohabitation indicates the
50
probability that a woman will enter a
40 new cohabiting relationship after the end
Percent disrupted

of her first marriage. The probability of

30 cohabitation after the end of the first
marriage is 53 percent after 5 years and
20 70 percent after 10 years (table 29).
Black women are significantly less
10 likely to cohabit after marriage than
Hispanic or white women. Five years
0 after the end of the first marriage, the
0 1 3 5 10 15
Duration in years
probability of postmarital cohabitation is
50 percent for Hispanic women,
Figure 25. Probability that the first marriage breaks up by duration of marriage and race 58 percent for white women, and only
difference with husband: United States, 1995 31 percent for black women (table 29
and figure 28). The difference between
Hispanic and white women is not
statistically significant.
Low Middle High Women with no religious affiliation
50
44 43
are the most likely to have cohabited
41 41 after marriage (86 percent within 10
40
years), and women affiliated with
Percent disrupted

33 33 34 34
30 29 fundamentalist Protestant faiths are the
26 least likely (56 percent within 10 years,
23 24
20 table 29). Figure 29 shows that women
with any religious affiliation are less
10 likely to cohabit after marriage than
women with no religious affiliation. For
0 women whose religion is very
Unemployment Income Poverty Welfare important, the probability of
cohabitation after marriage is 62 percent
Figure 26. Probability that the first marriage breaks up within 10 years by community
male unemployment rate, median family income, percent below poverty, and percent
within 10 years, compared with
receiving public assistance: United States, 1995 77 percent of those for whom religion is
not important (table 29).
Having no children at the end of
first marriage is associated with a higher
Low income Medium income High income probability of entering a new
60
56 cohabitation: Figure 30 shows that the
50
probability of postmarital cohabitation
44 43 within 10 years of separation is
Percent disrupted

40 38 39 77 percent for childless women,

33
31
33 33 70 percent for women with one child,
30 26 and 63 percent for women with more
23 23 than one child. Figure 30 also shows
20 that this effect is smaller among white
women.
10
Other characteristics of individuals
0 that are associated with a higher
All races Hispanic Non-Hispanic white Non-Hispanic black probability of cohabitation after
Figure 27. Probability that the first marriage breaks up within 10 years by race/ethnicity marriage include not growing up in a
and median family income in the community: United States, 1995 two-parent intact family (table 29).
Women at least 25 years of age at the
by characteristics of the woman and her black women and Hispanic women in end of the first marriage are less likely
community. Tables 31 and 32 show the sample whose first marriages have than younger women to cohabit after
these estimates for non-Hispanic white ended is not large enough for reliable marriage, although differences between
women; the numbers of non-Hispanic estimation by independent and teenagers and women in their early
 Series 23, No. 22 [ Page 21

twenties are not statistically significant

90 Hispanic Non-Hispanic white Non-Hispanic black (table 29).
80
The probability of postmarital
cohabitation is lower in central cities
70 than in other areas (table 30).
Percent cohabiting

60 Cohabitation after marriage is less likely

in communities with high
50
unemployment, low median family
40 income, a high percent of families
30 below poverty or receiving public
assistance, a low percent college-
20 educated, a high percent of women
10 never-married, and high crime rates
(table 30 and figure 31).
0
0 1 3 5 10 15 For several of these community-
Duration in years level factors, the difference in the
probability of postmarital cohabitation
Figure 28. Probability of postmarital cohabitation by duration of separation and race/ that is statistically significant is between
ethnicity: United States, 1995
one extreme category and the two other
categories. For example, the likelihood
100
No religious affiliation Any religious affiliation
of cohabitation after marriage is
90
significantly lower in communities with
very high unemployment (the top
80 quartile) than in communities with low
70 or moderate rates of unemployment.
Percent cohabiting

60
This is the pattern that is repeated for
all variables in table 30 except
50 metropolitan status and median family
40 income. Cohabitation after marriage is
30
significantly less likely in communities
with very high unemployment, very
20 high poverty and receipt of welfare,
10 very low education, and a very high
percent of women never-married.
0
0 1 3 5 10 15 Cohabitation after marriage is more
Duration in years likely in communities with very low
crime rates (although the difference by
Figure 29. Probability of postmarital cohabitation by duration of separation and religious crime rate is only significant at 5 years
affiliation: United States, 1995
duration of separation, table 30).

The Probability of Separation

No children One child Two or more
90 Making the Transition to
81
80 77 Divorce
75 73
70
70 Tables 33 and 34 show the
63
probability that a separation will lead to
Percent cohabiting

60
divorce by characteristics of the woman
50 and her community. Tables 35 and 36
40 show these estimates for non-Hispanic
white women; the numbers of
30
non-Hispanic black women and
20 Hispanic women in the sample whose
10 first marriages have ended in separation
are not sufficient for reliable estimation
0
All races Non-Hispanic white by independent and contextual variables.
Most separated women make the
Figure 30. Probability of postmarital cohabitation within 10 years of separation by transition to divorce very quickly:
race/ethnicity and parity at separation: United States, 1995
84 percent make the transition to divorce
Page 22 [ Series 23, No. 22

Midwest. Because the vast majority of

Low Middle High women make the transition to divorce
80 76
within 5 years, the differences tend to
76 76 75
73 72 72 disappear at later durations, although
70 67
63 some of the differences can still be
60 58 59 59 detected at 5 years duration.
Percent cohabiting

50
The contextual variables in table 34
show a consistent pattern in that,
40 typically, one extreme category differs
30 significantly from the other two
20
categories in the probability of the
transition to divorce. The transition to
10 divorce is less likely in communities
0 with very high unemployment, poverty,
Unemployment Income Poverty Welfare receipt of public assistance and percent
of women never-married, and very low
Figure 31. Probability of postmarital cohabitation within 10 years of separation by median family income and education
community male unemployment rate, median family income, percent below poverty, and
percent receiving public assistance: United States, 1995
(table 34 and figure 33). Differences
between the median category and the
opposite extreme are smaller. The
transition from separation to divorce is
Hispanic Non-Hispanic white Non-Hispanic black less likely in central cities and more
100 likely in communities with very low
90 crime rates.
80
The transition from separation to
divorce is virtually universal among
70 non-Hispanic white women, as noted
Percent divorced

60 previously. Hence, differences among

non-Hispanic white women tend to
50
disappear by 5 years of separation.
40 Differences found for the full sample
30 are also smaller when the statistics are
limited to non-Hispanic white women
20
(tables 35 and 36).
10

0 The Probability of Remarriage

0 1 3 5 10
Duration in years
Following Divorce
Tables 37 and 38 show the
Figure 32. Probability of transition from separation to divorce by duration of separation
and race/ethnicity: United States, 1995
probability of remarriage by
characteristics of the woman and her
community. Tables 39 and 40 show
within 3 years, and 91 percent do so Figure 32 shows the large gap that exists
these estimates for non-Hispanic white
within 5 years (table 33). between white women and Hispanic or
women; the numbers of non-Hispanic
The separations of white women are black women.
black and Hispanic divorced women in
much more likely to result in divorce The transition from separation to
the sample were not large enough for
than the separations of black or divorce is less likely for women without
reliable estimation by independent and
Hispanic women. Virtually all a high school degree, for women with
contextual variables. Table 37 shows that
separations of white women result in low income, for women not working at
54 percent of divorced women remarry
divorce quickly. Table 33 shows that the time of separation, for women with
within 5 years and 75 percent of
97 percent of white women make the children at separation, and (among those
divorced women remarry within 10
transition to divorce within 5 years of with children) for women with any
years. Black women are the least likely
the separation, compared with unwanted children at separation
to remarry, and white women are the
77 percent of Hispanic women and only (table 33). The transition to divorce is
most likely to remarry (figure 34). After
67 percent of black women. A also less likely if the first birth occurred
5 years of divorce, the probability of
substantial proportion (15 percent or before the marriage or during the first 7
remarriage is 58 percent for white
more) of Hispanic and black separations months of marriage and is less likely in
women, 44 percent for Hispanic women,
remain as separations for the long term. the Northeast and more likely in the
and only 32 percent for black women
 Series 23, No. 22 [ Page 23

(table 37). A study based on Cycle 2 of

100 Low Middle High the NSFG, conducted in 1976, showed
91 90 92 that the probability of remarriage within
90 87 88 89 88
85
5 years of divorce was 73 percent for
80 78
72
white women and 49 percent for black
72 70
70 women (28). The results in table 37
Percent divorced

60
suggest that the probability of
remarriage within 5 years of divorce has
50 decreased over the 19-year interval
40 between Cycles 2 and 5 (a decrease of
one-fifth for white women and one-third
30
for black women).
20 The probability of remarriage is
10 higher for women who were under age
25 at divorce. Women under age 25 at
0
Unemployment Income Poverty Welfare divorce have an 81-percent chance of
remarriage within 10 years, while
Figure 33. Probability of transition from separation to divorce within 3 years of separation women 25 years of age or over at
by community male unemployment rate, median family income, percent below poverty, divorce have only a 68-percent chance
and percent receiving public assistance: United States, 1995
of remarriage within 10 years, a
difference of 13 percentage points
(table 37). Figure 35 shows that the
Hispanic Non-Hispanic white Non-Hispanic black difference among white women is
80
11 percentage points.
70 Table 37 shows that the probability
60 of remarriage is greater for women with
higher family income, although this
Percent remarried

50
result should be interpreted with caution
40 because family income was measured at
30 interview and could have increased as a
result of remarriage. Women living in
20
the South are more likely to remarry
10 than other women (table 37).
Differences by parity at divorce are
0
0 1 3 5 10 small and not significant, although the
Duration in years results do suggest that remarriage is
Figure 34. Probability of remarriage by duration of divorce and race/ethnicity: more likely if there are no children
United States, 1995 present at the time of the divorce
(table 37). Women whose first birth was
at least 7 months after first marriage are
more likely to remarry after divorce
Under 25 years 25 years and over than women whose first birth occurred
90
81 83 before or just after marriage. Women
80 who have never had a birth by the time
72
70 68 of the NSFG interview are less likely to
remarry than women who had had a
Percent remarried

60
birth by the time of the interview
50 (table 37). Many of these differences in
40 the probability of remarriage diminish
after 10 years of divorce.
30 The probability of remarriage is
20 lower in communities with very high
10
unemployment, poverty and receipt of
public assistance, and very low median
0 family income (although the difference
All races Non-Hispanic white
by family income is not significant,
Figure 35. Probability of remarriage within 10 years of divorce by race/ethnicity and age at table 38 and figure 36). Women living in
divorce: United States, 1995 communities with a higher percent of
Page 24 [ Series 23, No. 22

quite large: The probability is about

Low Middle High 45 percent higher for nonmetropolitan
80 79 77 77 78 79 areas than for central cities, for all
75 74 75
71 women and white women (figure 37).
70 67 66 65 Among divorced white women, the
60 chance of remarriage within 5 years of
Percent remarried

50 divorce is 47 percent in central cities,

58 percent in the suburbs of
40
metropolitan areas, and 68 percent in
30 nonmetropolitan areas (table 40).
20 Many of these community effects
found for the full sample fade to
10 nonsignificance for white women. The
0 probability of remarriage for white
Unemployment Income Poverty Welfare
women is lower in central cities and
Figure 36. Probability of remarriage within 10 years of divorce by community male higher in rural areas, and lower in
unemployment rate, median family income, percent below poverty, and percent receiving communities with a higher percent of
public assistance: United States, 1995 women never-married; no other
contextual variable in table 40 shows a
significant effect for white women.
Central city Other SMSA 1 Not SMSA1
The Probability of Second
70 68
64 Marriage Disruption
60 58
54 Tables 41 and 42 show the
Percent remarried

50 47
44 probability of second marriage
40 disruption due to separation or divorce
30
by characteristics of the woman and her
community. Tables 43 and 44 show
20 these estimates for non-Hispanic white
10 women; the numbers of remarriages
among non-Hispanic black and Hispanic
0
All races Non-Hispanic white women in the sample are too small for
1SMSA is standard metropolitan statistical area.
separate analysis by individual and
Figure 37. Probability of remarriage within 5 years of divorce by race/ethnicity and contextual variables. Fifteen percent of
metropolitan status: United States, 1995 remarriages have dissolved after 3 years
and almost a quarter after 5 years
(table 41).
The remarriages of black women
Hispanic Non-Hispanic white Non-Hispanic black are more likely to disrupt, and those of
50
Hispanic women are less likely to
40
disrupt, although differences by
Percent disrupted

race/ethnicity are not statistically

30 significant, due to the small sample
sizes of Hispanic and black women
20 (table 41 and figure 38). Prior research
has suggested that the remarriages of
10 black women are less stable than those
of white women (23). That study did not
0 consider Hispanic women separately
0 1 3 5 10
Duration in years
because the number of Hispanic women
in the sample was too small to be
Figure 38. Probability that the second marriage breaks up by duration of marriage and analyzed separately.
race/ethnicity: United States, 1995 The data also show that second
marriage disruption is more likely for
women never-married are less likely to in central cities (table 38 and figure 37). women under age 25 at remarriage than
remarry after divorce, and remarriage is The difference in the probability of for women at least age 25 at remarriage.
more likely in rural areas and less likely remarriage by metropolitan status is After 10 years of remarriage, the
 Series 23, No. 22 [ Page 25

GAD are nearly 50 percent more likely

Under 25 years 25 years and over
to experience a second marriage
50 49 disruption than women who have never
47
45 suffered from GAD (figure 42).
40 Women who have children at the
Percent disrupted

34
35 33 time of remarriage are more likely to
30 experience second marriage disruption
25 than women who do not have any
20 children, and if the children were
15 unwanted, the probability of disruption
10 is even higher (table 41 and figure 43).
5 Figure 43 shows that after 10 years of
0 remarriage, the probability of disruption
All races Non-Hispanic white
is 32 percent for women with no
Figure 39. Probability that the second marriage breaks up within 10 years by race/ children at remarriage. For women with
ethnicity and age at remarriage: United States, 1995 children, but none of whom were
reported as unwanted, the probability is
40 percent, and for women with
Intact two-parents Otherwise children, and any of whom were
60
51 reported as unwanted, the probability is
49
50 44 percent (slightly higher, at 47 percent,
Percent disrupted

40 among white women, figure 43). It is

33 33
not surprising that the presence of
30
children from a prior relationship can
20 affect the stability of a second marriage,
10 nor is it surprising that the presence of
unwanted children may have a greater
0
All races Non-Hispanic white effect.
The probability of second marriage
Figure 40. Probability that the second marriage breaks up within 10 years by race/
disruption is significantly higher for
ethnicity and intact family of origin: United States, 1995
women with lower family income and is
lower in the Northeast and higher in the
Midwest (table 41). Other individual
Ever had forced intercourse Never had forced intercourse characteristics did not show significant
50 48 effects, although the data suggest that
45
second marriage disruption may be more
Percent disrupted

40 36 36 likely for women without a high school

30 education, for women with no religious
20 affiliation, for women whose first birth
was before or during the first 7 months
10
of first marriage as opposed to after 7
0 months of marriage, and for women
All races Non-Hispanic white
who are older than their husbands.
Figure 41. Probability that the second marriage breaks up within 10 years by race/ Interestingly, although the probability of
ethnicity and forced intercourse: United States, 1995 first marriage disruption is higher if the
first marriage was preceded by
probability of disruption is 47 percent family (49 percent) than for women who cohabitation, this is not the case for
for women who were under age 25 at did (33 percent, figure 40). Women who second marriage; if anything,
remarriage and 34 percent for women at have ever been forced to have cohabitation before remarriage may be
least age 25 at remarriage (table 41 and intercourse are more likely to experience associated with a lower probability of
figure 39). This difference is slightly second marriage disruption (figure 41). disruption, although the difference is
larger among white women (figure 39). The probability of second marriage small and not statistically significant.
Stressful events in the past may disruption is about 25 percent higher for Second marriage disruption is
impact the stability of remarriages. The all women who have ever been forced significantly more likely in communities
probability of second marriage to have intercourse, and about one-third with a high percent of households below
disruption is higher for women who did greater among white women (figure 41). poverty, low median family income, and
not grow up in a two-parent intact Women who have ever suffered from low percent college-educated (table 42
and figure 44). Other contextual
Page 26 [ Series 23, No. 22

1 of the NSFG was conducted in 1973,

60 Ever had GAD 1 Never had GAD1 Cycle 2 was conducted in 1976, Cycle 3
50 50 in 1982, and Cycle 4 in 1988. Pooling
50
Percent disrupted

Cycles 1, 2, 4, and 5 provides enough

40 cases to generate reliable estimates of
34 34
30 the probabilities by marriage cohort. A
marriage cohort refers to all marriages
20
that occurred within a particular period.
10 Figure 45 presents a trend analysis
0 of the probability of first marriage
All races Non-Hispanic white disruption due to either separation or
1GAD is generalized anxiety disorder.
divorce. The graph plots the probability
Figure 42. Probability that the second marriage breaks up within 10 years by race/ of marital disruption within 10 years of
ethnicity and generalized anxiety disorder: United States, 1995 marriage for 5-year marriage cohorts
between 1950 and 1984. The 5-year
cohorts of 198589 and 199094 are
No children Any children, none unwanted Any children, any unwanted
excluded because most of the marriages
50
begun in those years had not had the
47
45 44 chance to exist for 10 years as of the
40
40 40 1995 Cycle 5 interview (only those that
Percent disrupted

35 began in the very beginning of the

32 32
30 198589 cohort would have had the
25 chance to last for 10 years by the 1995
20 interview). The plotted statistics are
15 presented in tabular form in table D.
10 Figure 45 shows that the probability
5 of first marriage disruption within 10
0 years of marriage increased over time
All races Non-Hispanic white for marriages begun from the 1950s
Figure 43. Probability that the second marriage breaks up within 10 years by race/ through the 1970s and then leveled off
ethnicity and presence of unwanted children at remarriage: United States, 1995 in 197584. Among non-Hispanic black
women, however, the leveling off did
not occur. The probability of first
Low Middle High
60 marital disruption among black women
54 decreased for marriages begun in the
50 49
45 46 1950s, increased slightly in the 1960s,
Percent disrupted

41 39 and then increased at a greater rate

40 38 36
31 33 through the 1970s and early 1980s. This
30 26 28
sharp increase for black women resulted
20 in a larger gap between white and black
women such that for first marriages
10
begun in the 1980s, the probability of
0 marital disruption within 10 years was
Unemployment Income Poverty Welfare
about two-thirds greater for black than
Figure 44. Probability that the second marriage breaks up within 10 years by community for white women. The gap between
male unemployment rate, median family income, percent below poverty, and percent white and black women was smallest in
receiving public assistance: United States, 1995 the late 1960s and largest in the 1980s.
Figure 46 shows the trend over time
variables do not show significant effects, Trends Over Time in the probability that a separation from
although the data suggest that second first marriage will make the transition to
marriage disruption is higher in The statistics presented thus far in divorce. The graph shows that for
communities with higher unemployment, this report are based on data from Cycle marriages begun from the 1950s through
higher percent receiving public 5 of the NSFG, collected in 1995. the 1980s, the probability of completing
assistance, and higher percent of women Previous cycles of the NSFG also the divorce process within 5 years of
never-married (table 42 and figure 44). collected marital histories, allowing for separation was mostly unchanged,
an analysis of trends over time in the ranging from 85 percent to 90 percent
probabilities of some of the marital during the entire period (figure 46). The
outcomes examined in this report. Cycle probability of a separation making the
 Series 23, No. 22 [ Page 27

marriages begun in the late 1980s, and

Total Non-Hispanic white Non-Hispanic black
50 the probability of a separated black
woman finalizing her divorce decreases
for first marriages begun in the late
Percent disrupted

40
1980s (figure 46). This means that
analyses that only examine divorce as
30
the indicator of marital disruption and
ignore separation will understate the
20 racial difference in marital disruption to
an even greater extent over time. It will
10 be important to continue to monitor
1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 these trends through the 1990s once
Marriage cohort Cycle 6 of the NSFG is completed in
Figure 45. Probability of first marriage disruption within 10 years by marriage cohort and
2002.
race/ethnicity: United States, marriages begun in 195084 Figure 47 shows the trend in the
probability of remarriage within 5 years
of divorce. The cohorts in this case refer
Table D. Probabilities of first marriage disruption, of transition from separation to divorce, to divorce cohorts, that is, all the
of remarriage, and of second marriage disruption by race/ethnicity and cohort:
United States
divorces that occurred in a particular
10-year period. Ten-year cohorts are
Probability used instead of 5-year cohorts because
Race/ethnicity of sample size considerations. The trend
shows that the overall pattern is one of
NonHispanic Non-Hispanic
Marital outcome and marriage cohort Total1 white black decreasing chances of remarriage over
time, for all women (figure 47). The
First marriage disruption within 10 years chances of remarriage were lower for
195054 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.14 0.13 0.24 black women than for white women
195559 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.14 0.13 0.19
throughout the period.
196064 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.18 0.17 0.21
196569 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.19 0.19 0.22 Figure 48 shows the trend in second
197074 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.28 0.27 0.32 marriage disruption, again in 10-year
197579 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.31 0.30 0.43 cohorts. The overall pattern is one of
198084 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.31 0.28 0.47
increasing chances of second marriage
Separation to divorce within 5 years disruption over time. For non-Hispanic
195054 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.85 0.90 0.50 white women, the chance of second
195559 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.86 0.93 0.54 marriage disruption decreased slightly
196064 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.88 0.93 0.57 from the 1950s to the 1960s and then
196569 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.86 0.93 0.52
197074 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.90 0.96 0.65
increased in the 1970s and 1980s. For
197579 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.90 0.96 0.64 non-Hispanic black women, the
198084 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.86 0.93 0.64 probability of second marriage
198489 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.88 0.96 0.56 disruption was initially lower than that
Divorce to remarriage within 5 years of white women in the 1960s. The
1950592 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.65 0.70 0.43
probability of disruption among black
196069 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.60 0.63 0.40 women in the 1950s cannot be shown
197079 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.58 0.60 0.38 because there were not enough black
198089 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.50 0.55 0.27 women whose second marriage had
Second marriage disruption within 5 years begun in the 1950s in the combined
195059 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.16 0.16 *
NSFG samples to generate reliable
196069 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.14 0.15 0.12 estimates. The probability of disruption
197079 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.18 0.18 0.18 among black women then increased
198089 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.22 0.22 0.27
through the 1970s and 1980s, and that
* Figure does not meet standard of reliability or precision.
increase was greater than that among
1
Includes Asian and Pacific Islander women and American Indian women not shown separately.
white women, such that by the late
2
For the probability of remarriage, the cohorts refer to divorce cohorts rather than marriage cohorts.

1980s, the probability of second

marriage disruption was considerably
transition to divorce was much lower for racial divergence that occurred for higher among black women than among
black women than for white women marriages begun in the late 1980s. The white women.
throughout the period. The most probability of a separated white woman
interesting feature of figure 46 is the finalizing her divorce increases for first
Page 28 [ Series 23, No. 22

100 Total Non-Hispanic white Non-Hispanic black Summary of Findings

for Each Independent
90 Variable
Percent divorced

T
80 his section summarizes the
findings by the individual and
community variables. Table E
70
indicates whether each individual
characteristic is statistically significantly
60
related to each outcome, and table F
indicates whether each community
characteristic is statistically significantly
50 related to each outcome. The remainder
1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 1985-89 of this section briefly summarizes how
Marriage cohort the characteristics are related to each
outcome.
Figure 46. Probability of transition from separation to divorce within 5 years by marriage
cohort and race/ethnicity: United States, marriages begun in 195084 Individual CharacteristicsAge has
significant effects for some of the
marital and cohabitation outcomes
considered in this report. Women 25
years of age or over at the start of the
Total Non-Hispanic white Non-Hispanic black cohabitation are less likely to experience
70 cohabitation disruption than women
under 25 years of age. Likewise, higher
60 age at marriage is associated with a
Percent remarried

lower probability of marital disruption.

50 These effects are found for the full
sample, for non-Hispanic white women,
40
and for non-Hispanic black women,
30
although the differences are not always
significant among black women,
20 probably due to small sample size.
1950-59 1960-69 1970-79 1980-89 Cohabitation after marriage and
Divorce cohort remarriage after divorce are less likely
for women 25 years of age or over at
Figure 47. Probability of remarriage within 5 years by divorce cohort and race/ethnicity:
United States, divorces occurring in 195089
marital dissolution.
Education and family income are
correlated and typically relate to these
marital outcomes in similar ways.
Higher education, higher mothers
education, and higher family income are
30
Total Non-Hispanic white Non-Hispanic black associated with a lower likelihood of
first marriage by age 18 (although
Percent disrupted

income is not related to early marriage

among black women). Higher education
20 and higher income are associated with a
greater probability of the first
cohabitation making the transition to
marriage, although these effects are
stronger for black women than for white
10 women. Higher family income is
1950-59 1960-69 1970-79 1980-89
Marriage cohort
associated with a lower likelihood of
cohabitation disruption. Higher
Figure 48. Probability of second marriage disruption within 5 years by marriage cohort education and income are associated
and race/ethnicity: United States, marriages begun in 195084
with a lower probability of marital
disruption. Higher family income is
 Series 23, No. 22 [ Page 29

Table E. Statistical significance of each individual characteristic by outcome affiliation are less likely to marry by
Transition age 30 than women with any religious
Transition to from First First affiliation; women of fundamentalist
first marriage cohabitation cohabitation marriage
Characteristic (all women) to marriage disruption disruption Protestant faiths are more likely than
other women to marry by any age (up to
Race/ethnicity . . . . . . . . . . . . . . . significant significant significant significant age 30); and women who report that
Age at start of interval . . . . . . . . . . ... significant significant significant
Education . . . . . . . . . . . . . . . . . significant significant significant significant
their religion is not important are less
Family income . . . . . . . . . . . . . . . significant significant significant significant likely to marry by age 30. Women with
Work status at start of interval . . . . . ... not significant significant significant no religious affiliation are less likely to
Religious affiliation . . . . . . . . . . . . significant significant significant significant make the transition from cohabitation to
Importance of religion . . . . . . . . . . significant significant not significant significant
Intact family of origin status . . . . . . significant significant significant significant
marriage than women with any religious
Forced intercourse . . . . . . . . . . . . ... not significant significant significant affiliation. Women with no religious
Parity at start of interval . . . . . . . . . ... significant significant significant affiliation are also more likely to
Wantedness of children . . . . . . . . . ... not significant not significant not significant
experience cohabitation disruption, are
Timing of first birth . . . . . . . . . . . . ... significant significant significant
Generalized anxiety disorder . . . . . . significant not significant significant significant more likely to experience first marriage
Region of residence . . . . . . . . . . . significant not significant significant significant disruption, and are more likely to
Age difference with spouse . . . . . . . ... ... ... not significant cohabit after the dissolution of first
Race difference with spouse . . . . . . ... ... ... significant
marriage. Women affiliated with
Cohabited before marriage . . . . . . . ... ... ... significant
fundamentalist Protestant denominations
Transition are less likely to cohabit after the first
from Second
Postmarital separation marriage marriage ends. White women who report
Characteristic cohabitation to divorce Remarriage disruption that religion is not important to them are
Race/ethnicity . . . . . . . . . . . . . . . significant significant significant not significant
more likely to experience first marriage
Age at start of interval . . . . . . . . . . significant not significant significant significant disruption and are less likely to cohabit
Education . . . . . . . . . . . . . . . . . significant significant not significant not significant after the first marriage ends; these
Family income . . . . . . . . . . . . . . . significant significant significant significant differences are not significant among
Work status at start of interval . . . . . significant significant not significant not significant
Religious affiliation . . . . . . . . . . . . significant not significant not significant not significant
black women.
Importance of religion . . . . . . . . . . significant significant not significant not significant Whether the woman was raised in
Intact family of origin . . . . . . . . . . . significant significant not significant significant an intact, two-parent family throughout
Forced intercourse . . . . . . . . . . . . significant not significant not significant significant
childhood is related to some of the
Parity at start of interval . . . . . . . . . significant significant not significant not significant
Wantedness of children . . . . . . . . . not significant significant not significant not significant marital and cohabitation outcomes in
Timing of first birth . . . . . . . . . . . . not significant significant significant significant this report. The effect of family
Generalized anxiety disorder . . . . . . not significant significant not significant significant background on first marriage varies by
Region of residence . . . . . . . . . . . not significant significant significant significant
race/ethnicity. Among Hispanic women,
Age difference with spouse . . . . . . . ... ... ... not significant
Race difference with spouse . . . . . . ... ... ... not significant growing up in an intact family is
Cohabited before marriage . . . . . . . ... ... ... significant associated with a higher chance of first
marriage at all ages. Among non-
. . . Category not applicable.
Hispanic white women, an intact family
NOTES: Statistical significance is indicated if at least one category differed significantly from at least one other category in at
least one duration interval; significance refers to findings for all women and may differ by race/ethnicity. Individual characteristics is associated with a lower chance of
were analyzed independently; multivariate analysis may reveal that some of these effects are influenced by unobserved factors. early marriage, and the difference by
family background converges at later
associated with a higher likelihood of more likely to make the transition to ages. There is no effect of family
postmarital cohabitation. Higher marriage than full-time workers or background among non-Hispanic black
education is associated with a lower nonworkers. White women who were women. Growing up with two parents
probability of postmarital cohabitation not working at cohabitation and black has only a small effect on the
among white women but not among women who were working part time at probability of making the transition
black women. Low education and cohabitation are more likely to from cohabitation to marriage. But
income suggest a lower probability of a experience cohabitation disruption. Part growing up with two parents is
separation making the transition to time workers are less likely to associated with a lower likelihood of
divorce. Higher income, but not experience first marriage disruption, and cohabitation disruption, a much lower
education, is associated with a greater nonworkers are less likely to make the likelihood of first marriage disruption, a
likelihood of remarriage and a lower transition from separation to divorce. lower likelihood of postmarital
likelihood of second marriage Religious affiliation and the cohabitation, and a much lower
disruption. importance of religion are related to likelihood of second marriage
Work status at cohabitation is many of the marital and cohabitation disruption.
related to the probability of the outcomes in this report. Among Women who have ever been forced
transition to marriage among black Hispanic and non-Hispanic white to have intercourse have a greater
women only: Part-time workers are women, women with no religious probability of cohabitation disruption
Page 30 [ Series 23, No. 22

Table F. Statistical significance of each contextual characteristic by outcome with a greater chance of first marriage,
Transition although the difference disappears by
Transition to from First First age 30. GAD is also associated with a
first marriage cohabitation cohabitation marriage
Characteristic (all women) to marriage disruption disruption greater likelihood of cohabitation
disruption, a greater likelihood of first
Male unemployment rate . . . . . . . . significant significant significant significant marriage disruption, and a greater
Median family income . . . . . . . . . . significant significant significant significant
Percent below poverty . . . . . . . . . . significant significant significant significant
likelihood of second marriage
Percent on public assistance . . . . . . significant significant significant significant disruption.
Percent college-educated . . . . . . . . significant significant not significant significant Women in the South tend to marry
Crime rate . . . . . . . . . . . . . . . . . significant not significant significant significant earlier than women in other regions,
Percent of women never married . . . significant significant significant significant
Metropolitan status . . . . . . . . . . . . significant significant significant significant
although regional differences in first
marriage converge at age 30. Region of
Transition residence is not related to the likelihood
from Second
Postmarital separation marriage of the transition from cohabitation to
Characteristic cohabitation to divorce Remarriage disruption
marriage for the full sample, but among
Male unemployment rate . . . . . . . . significant significant significant not significant black women, the transition is more
Median family income . . . . . . . . . . significant significant not significant significant likely in the South and less likely in the
Percent below poverty . . . . . . . . . . significant significant significant significant Northeast. Cohabitation disruption and
Percent on public assistance . . . . . . significant significant significant not significant
Percent college-educated . . . . . . . . significant significant not significant significant
first marriage disruption are more likely
Crime rate . . . . . . . . . . . . . . . . . significant significant significant not significant in the South and, among white women,
Percent of women never married . . . significant significant significant not significant first marriage disruption is also more
Metropolitan status . . . . . . . . . . . . significant significant significant not significant likely in the West. The transition from
NOTES: Statistical significance is indicated if at least one category differed significantly from at least one other category in at
separation to divorce is more likely in
least one duration interval; significance refers to findings for all women, and may differ by race/ethnicity. Contextual the Midwest and less likely in the
characteristics were analyzed independently; multivariate analysis may reveal that some of these effects are influenced by
unobserved factors.
Northeast. Remarriage is more likely in
the South, and second marriage
and a greater probability of first before or during the first 7 months of disruption is more likely in the Midwest
marriage disruption. Forced intercourse marriage, and if the woman had any and less likely in the Northeast.
is associated with a higher likelihood of children at separation. Among women Community InfluencesNot
postmarital cohabitation among black with children, the transition from surprisingly, affluence is good for the
women. Forced intercourse is not related separation to divorce is less likely if any stability of marriages and cohabitations;
to the probability of remarriage but is of the children were unwanted. poverty is not. In this report,
associated with a higher likelihood of Remarriage is more likely for community-level socioeconomic status
second marriage disruption. women with no children at divorce than (SES) has been measured by the male
Childbearing has been measured in for women with any children at divorce, unemployment rate, median family
a variety of ways in this report, although this difference is small and not income, the percent of families below
including parity, whether any children significant. Remarriage is more likely poverty, the percent of households
were unwanted, and timing of first birth. for women whose first birth was more receiving public assistance, and the
The transition to marriage is more likely than 7 months after first marriage, and percent of adults who are college-
for women with no children at the start is less likely for women whose first educated. Affluence is indicated by
of the cohabitation, and for women who birth was before first marriage or who higher median family income and
had their first birth after the start of the had never had a birth by the time of the percent college-educated, and lower
union compared with before the union. interview. Second marriage disruption is unemployment, percent below poverty,
These differences are not significant more likely for women who had any and percent receiving public assistance.
among black women, however. children at remarriage or whose first The five indicators of community SES
Cohabitation disruption is less likely for birth occurred before or during the first are almost always consistent in their
women whose first birth was more than 7 months of first marriage (although the effects on cohabitation and marital
7 months after cohabitation and is more difference by timing of first birth is not outcomes. Any lack of agreement among
likely for women who have never had a significant). the five indicators is a result of one or
first birth. First marriage disruption is Women who had ever had GAD more not showing a significant effect. In
more likely for women whose first birth reported that they had suffered for at no analyses in this report is there a
was before marriage compared with least 6 months from worry or anxiety, direct contradiction between two
afterwards. Women with no children at including feelings of restlessness, feeling indicators of community SES that
the end of first marriage are more likely keyed up or on edge, irritability, a suggest two statistically significant
to cohabit after the marriage ends. The pounding or racing heart, getting tired effects that operate in opposite
transition from separation to divorce is easily, trouble falling or staying asleep, directions of the community SES
less likely if the first birth occurred and feeling faint. GAD is associated continuum.
 Series 23, No. 22 [ Page 31

Higher community-level SES is

associated with a lower probability of
because of the level of context at which
it is measured. While all the other
Discussion
early first marriage (indicated by all five continuous contextual variables are
community SES variables), a greater measured at the census tract level, the

I
n each comparison of racial/ethnic
probability of an intact cohabitation crime rate was only available at the subgroups, the results consistently
making the transition to marriage county level, and the county may be too suggest that the unions of
(indicated by all five community SES large an area to measure the context in non-Hispanic black women are less
variables), a lower probability of which these outcomes occur. Among stable than those of non-Hispanic white
cohabitation disruption (indicated by all white and black women, low-crime or Hispanic women. Black women are
five community SES variables except communities are associated with higher less likely to marry by age 30 and less
percent college-educated), a lower chances of first marriage (although the likely to make the transition from
probability of first marriage disruption differences are not significant for black cohabitation to marriage, and their
(indicated by all five), a higher women). Higher crime rates in the cohabitations are more likely to disrupt
probability of cohabitation after county are associated with a higher than those of other women. The first
marriage (indicated by all five), a likelihood of first marriage disruption marriages of black women disrupt faster
greater probability of making the and a lower likelihood of postmarital than the first marriages of other women.
transition from separation to divorce cohabitation. The crime rate is not Black women are less likely to enter a
(indicated by all five), a higher related to the other outcomes examined cohabitation after the dissolution of the
probability of remarriage (indicated by in this report. first marriage. The separations of black
all five except percent college-educated), Not surprisingly, the marriage women are less likely to make the
and a lower probability of second market context seems to matter for transition to divorce, and the interval
marriage disruption (indicated by all five, cohabitation and marital outcomes. A between divorce and remarriage is
although the effects of unemployment and higher percent of women never-married longer for black women. The data
percent receiving public assistance are not is associated with a lower probability of suggest that the remarriages of black
statistically significant). first marriage, a lower probability that women disrupt faster than the
This remarkable consistency is also the cohabitation makes the transition to remarriages of other women. The trend
apparent in analyses run separately for marriage, a higher probability of analysis suggests that, at least for some
Hispanic women, non-Hispanic white cohabitation disruption, a higher of these marital outcomes, the
women, and non-Hispanic black women. probability of first marriage disruption, a differences by race are increasing over
Although some or all of the effects of lower probability of cohabitation after recent decades. The differences between
these community SES indicators may marriage, a lower probability of the white and Hispanic women are smaller.
fade to nonsignificance among either transition from separation to divorce, Some researchers have suggested
Hispanic, white, or black women, there and a lower probability of remarriage. that these differences may be related to
are still no direct contradictions to Metropolitan status indicates higher rates of unemployment,
indicate anything other than the whether the woman lives in a central incarceration, and mortality among the
conclusion that affluence is good for city, the suburbs, or in a rural area. The black population, their lower levels of
cohabitation and marriage. However, the typical finding in this report is that educational attainment and earnings,
results by race/ethnicity do suggest that central cities are different from other their previous experiences as children of
community affluence may matter more areas in terms of the cohabitation and unmarried or less-educated parents, and
for black women than for white women marital outcomes considered here, and higher rates of poverty and lack of job
for at least some of these cohabitation differences between suburbs and opportunities in the communities in
and marital outcomes. However, it is nonmetropolitan areas are much smaller. which they live (13,19,20). The findings
difficult to simultaneously evaluate all First marriage is more likely in in this report suggest that individual
five indicators of community affluence nonmetropolitan areas and less likely in characteristics such as race/ethnicity
in this regard, because some are central cities. The transition from may not be the sole determinants of
significant for white women while cohabitation to marriage is less likely in marital and cohabitation success. The
others are significant for black women, central cities. Cohabitation disruption neighborhoods in which people live may
depending on the outcome in question. and first marriage disruption are more be important, and differences in marital
Other contextual characteristics likely in central cities. Postmarital and cohabitation outcomes between
considered in this report include the cohabitation is less likely in central white and black women may depend to
crime rate, marriage market (measured cities. Remarriage is much less likely in some extent on the community
by the percent of women never- central cities and more likely in environments in which the women live.
married), and metropolitan status nonmetropolitan areas. The overall Both non-Hispanic white women
(central city, other SMSA, or pattern suggests that central cities have and non-Hispanic black women who
nonmetropolitan). The crime rate was lower rates of union formation and live in neighborhoods with high levels
the least successful indicator of context higher rates of cohabitation and of poverty, receipt of welfare and
in that it showed the lowest number of marriage disruption than suburbs or unemployment, and low levels of
significant effects. This is most likely nonmetropolitan areas. income and education are more likely to
Page 32 [ Series 23, No. 22

experience separation and divorce. Black analyses, the measurement of children at if she had never been forced to have
women live disproportionately in the beginning of the interval is less than intercourse, if she had no children at the
low-SES neighborhoods. Whether the satisfactory. For example, in the analysis start of the cohabitation or marriage, if
lower marital success of black women is of first marriage disruption, it would be her first birth was more than 7 months
due to their disproportionate prevalence better to know how many children were after the beginning of the cohabitation
in low-SES neighborhoods, to individual present in the family at the time of the or marriage, if she has never suffered
characteristics, or to other factors, is a disruption, rather than at the time the GAD, if she is the same race/ethnicity
question for further study. marriage began, but because many of as her husband, or if she lives in
This report presents analyses of the first marriages had not disrupted by communities with higher median family
eight cohabitation and marital outcomes interview (were censored), it did not income, lower male unemployment, less
by a wide variety of individual and make sense to code the presence of poverty, less receipt of welfare, and
community characteristics. This children at the time of disruption. more adults who are college-educated.
presentation is not meant to represent a Although it makes sense to measure the Some of these characteristics show
definitive explanation of any of the number of children at the beginning of stronger effects for the stability of
outcomes presented here. Rather, the the interval when studying the marriage than for the stability of
intention is to provide benchmark transitions from separation to divorce, cohabitation and some of the effects
statistics by a wide variety of separation to postmarital cohabitation, vary by race/ethnicity. With the
characteristics, and to encourage and divorce to remarriage, it is less than exception of controlling for race/
researchers to consider these factors satisfactory for the earlier outcomes. In ethnicity, none of these characteristics
when studying marital outcomes. Of addition, characteristics of the children were tested in a multivariate context,
particular note is the presentation of themselves could be studied, such as and it may be that some of the effects
statistics on the probability of their age, gender, and whether they are are spurious, reflecting the effects of
postmarital cohabitation. The authors biological, adopted, or step-children. To other variables. Researchers are
know of no other study that analyzes do so, however, would require an encouraged to consider analysis of these
this facet of the marital life course with analysis specifically designed to outcomes with a multivariate approach,
nationally representative data. measure the effects of the characteristics and to include controls for race/ethnicity
The outcomes analyzed in this of children on these marital outcomes. and community context in the analysis
report deserve further study using Finally, another avenue of further of these outcomes.
multivariate statistical techniques. These research is non-coresidential families
techniques allow the statistician to (i.e., families that do not live together).
determine whether characteristics used The term fragile families has been References
in this report are less important or more used to describe families that do not
important when other characteristics are share a single residence and are at 1. Waite LJ. Does marriage matter?
controlled. Hazards models are one higher risk of poverty and family Demography 32 (4): 483507. 1995.
method of controlling for multiple instability than married-couple families 2. Rogers RG. Marriage, sex, and
characteristics simultaneously (47). For (55). This report focuses on cohabitation mortality. Journal of Marriage and the
example, a hazards model could show and marriage, both of which are Family 57 (2):51526. 1995.
the effect of income on the chances of shared-residence relationships, and does 3. Waite LJ, Gallagher M. The case for
marital disruption, controlling for not analyze alternative forms of family marriage: Why married people are
happier, healthier and better off
education, race/ethnicity, age, and other life. The stability of such families and
financially. New York: Doubleday.
characteristics. Another possibility for the effects of individual and community
2000.
further study is multilevel modeling, to characteristics on that stability, is an 4. Murray JE. Marital protection and
control for the effects of the community important topic, especially for marital selection: Evidence from a
(or neighborhood) environment (54). understanding low-income families. historical-prospective sample of
This report suggests that community Although the statistics presented in American men. Demography 37
characteristics are important for a full this report are descriptive in nature, it is (4):51121. 2000.
understanding of the outcomes analyzed possible to draw some conclusions about 5. Amato PR. The consequences of
in this report. the characteristics of individuals and divorce for adults and children. Journal
This report has focused on changes communities that may contribute to the of Marriage and the Family 62
in the relationships between spouses (or stability of cohabitations and marriages. (4):126987. 2000.
6. Duncan GJ, Hoffman SD. A
between cohabiting partners) and has Cohabitations and marriages tend to
reconsideration of the economic
not dealt, in detail, with the effects of demonstrate more stability if the woman
consequences of marital dissolution.
children on these outcomes. Although was older at the time the cohabitation or Demography 22 (4):48597. 1985.
we have tried to examine the presence marriage began, if her family income is 7. Smock PJ. The economic costs of
of children as a characteristic that may higher, if she has a religious affiliation marital disruption for young women
be associated with a marital or or reports that her religion is important over the past two decades. Demography
cohabitation outcome, this is a topic that to her, if she was raised through 30 (3):35371. 1993.
deserves closer scrutiny. In many of our childhood in a two-parent intact family,