Attributional Beliefs About the Controllability of Stigmatized

Traits: Antecedents or Justifications of Prejudice?1

Peter Hegarty2 and Anne M. Golden

University of Surrey
Guildford, United Kingdom

Correlational studies show that prejudiced people attribute stigmatized traits to

controllable causes, and blame stigmatized groups for their own fate. Attribution
theory argues that causal attributions cause prejudice, and that changes in attribu-
tional beliefs produce changes in attitudes. In contrast, the justification–suppression
model describes attributions to controllable causes as justifications of pre-existing
prejudices. Study participants reported their attitudes toward 1 of 4 stigmatized
groups, read information that manipulated their attributional beliefs, listed
their thoughts, and reported their attitudes again. Supporting the suppression–
justification model, initially prejudiced participants spontaneously produced more
thoughts about the controllability of stigmatized identities. Refuting attribution
theory, manipulating attributional beliefs had no effect on attitudes. Implications for
applications of attribution theory to reduce prejudice are discussed.

In social psychology and related disciplines, stigmata are defined as deeply

discrediting traits that can reduce a “whole and usual person to a tainted,
discredited one” (Goffman, 1963, p. 3). Social scientists have long pondered
the relationship between prejudice against stigmatized groups and attribu-
tional beliefs about the origins and controllability of stigmatized traits. For
example, Goffman distinguished between tribal stigmas, involving discred-
ited racial and religious identities; abominations of the body, including physi-
cal disfigurements and disabilities; and blemishes of individual character, such
as addictions, alcoholism, mental illness, and homosexuality.
The notion that stigmata with different causes vary in their psychological
impact has remained important. Attribution theory has described causal
beliefs as determinants of emotional, attitudinal, and behavioral responses to
stigmatized individuals and groups (Weiner, Perry, & Magnusson, 1988).
Attributions to controllable causes are held to blame the stigmatized target
for their own fate and to lead to greater prejudice toward that target.

1
This work was supported by an ESRC grant (RES-000-22-0288) to the first author. The
authors thank Lynsey Mahony for research assistance, and Mick Finlay and Thomas Morton
for comments on earlier drafts of the manuscript.
2
Correspondence concerning this article should be addressed to Peter Hegarty, University of
Surrey, Guildford GU2 7XH. E-mail: [email protected]

1023

Journal of Applied Social Psychology, 2008, 38, 4, pp. 1023–1044.

© 2008 Copyright the Authors
Journal compilation © 2008 Blackwell Publishing, Inc.
1024 HEGARTY AND GOLDEN

More recently, the justification–suppression model has hypothesized

further that pre-existing prejudicial attitudes might prompt attributional
beliefs that can justify pre-existing noxious prejudices (Crandall & Eshleman,
2003). In the present paper, we define attitudes as global evaluations of a
target, prejudice as negative attitudes, and attributional beliefs as cognitive
propositions about the origins and controllability of a stigmatized trait. We
ask the question “Which comes first—prejudicial attitudes toward a stigma-
tized group, or cognitive attributional beliefs about the origins and control-
lability of a stigmatized trait?” We describe the predictions of each theory
and report an experiment that tests them both.

Attributional Theory of Stigma

The attributional theory of stigma was inspired by findings that people are
more likely to help a person whose distress originates in an uncontrollable
cause, rather than in a controllable cause (e.g., Piliavin, Rodin, & Piliavin,
1969). This effect is mediated by emotional reactions: Uncontrollable causes
of distress elicit pity and sympathy, while controllable causes of distress elicit
anger and hostility (Reizenstein, 1986). Weiner et al. (1988) argued that
attributions similarly determine reactions to stigmatized individuals and
groups, such that uncontrollable stigmata elicit pity, sympathy, and helping
behavior; while controllable stigmata elicit anger and a refusal to extend aid.
Attributions of controllability are said to affect the degree to which stigma-
tized targets are blamed for their own fate (Weiner, 1993, 1995, 1996). For
example, addicted, obese, gay/lesbian, or mentally ill persons are predicted to
be treated better wherever their stigma is understood to originate in uncon-
trollable biological factors rather than in personal weaknesses or personal
choices.
Several studies of emotional reactions to individual stigmatized targets
support attribution theory. Targets with uncontrollable stigmata elicit more
sympathy and pity, are more likely to receive aid, and elicit less anger (Menec
& Perry, 1998; Rush, 1998; Weiner et al., 1988), regardless of whether their
stigmatized trait is HIV/AIDS (Cobb & deChabert, 2002; Dooley, 1995;
Graham, Weiner, Giuliano, & Williams, 1993; Steins & Weiner, 1999),
mental illness (Corrigan et al., 2001), physical illness (Crandall & Moriarty,
1995), obesity (DeJong, 1980), poverty (Zucker & Weiner, 1993), homosexu-
ality (Armesto & Weisman, 2001), or a failure to seek genetic testing for an
inheritable medical condition (Menec & Weiner, 2000). However, the
hypothesized causal link between anger reactions and refusals to extend aid
has failed to materialize in several studies (see Dooley, 1995; Menec & Perry,
1998; Steins & Weiner, 1999).
 ATTRIBUTIONS AND STIGMA 1025

However, these important studies are not directly relevant to hypoth-

esized effects of causal attributions on attitudes toward stigmatized groups,
and for several reasons. First, information about an individual only affects
beliefs about her group when she is perceived as a typical group member
(Wilder, Simon, & Faith, 1996). Stigmatized individuals—such as an obese
woman with a thyroid condition (DeJong, 1980), or a man who contracted
HIV from a blood transfusion (Dooley, 1995)—may be perceived as indi-
vidually blameless, but they may also be seen as atypical within the larger
stigmatized group.
Second, specific emotional reactions are not the same as global attitudes.
Attribution theory describes pity as a positive emotional reaction, but pity
presumes the legitimacy of the stigmatized target’s lower status (Leach,
Snider, & Iyer, 2003). Pity may engender both positive affect and support for
disempowering interventions toward the stigmatized group (Corrigan,
Markowitz, Watson, Rowan, & Kubiak, 2003), and may be perceived by
stigmatized individuals and groups as patronizing (Fine & Asch, 1988). Thus,
experiments demonstrating that attributions to uncontrollable causes engen-
der ostensibly positive emotional reactions to individuals do not compel the
conclusion that those attributions similarly produce more positive attitudes
toward stigmatized groups.
Existing research on attributional beliefs and attitudes toward stigmatized
groups is also largely correlational. Heterosexuals who believe sexual orien-
tation to be controllable have more heterosexist attitudes (Aguero, Bloch, &
Byrne, 1984; Ernulf, Innala, & Whitam, 1989; Haslam, Rotschild, & Ernst,
2000, 2002; Hegarty, 2002; Hegarty & Pratto, 2001; Sakalli, 2002; Whitley,
1990). Attitudes toward the mentally ill (Corrigan, 2000) and toward fat
people (Crandall, 1994) are also more negative among those who believe the
relevant stigmatized trait to be under personal control. Researchers have
often interpreted these findings along attribution theory lines as demonstrat-
ing effects of attributions on attitudes (e.g., Anderson, Krull, & Weiner,
1996; Crocker, Major, & Steele, 1998), and some have recommended teach-
ing biological determinist theories to reduce stigmatization as a result (e.g.,
Finell, 2002; Sakalli, 2002; Weiner, 1995; Whitley, 1990). However, as
Corrigan and Crandall et al. (2001) have pointed out, correlations are not
evidence of causal effects in this, or any, domain.
In contrast to the vast literature on individual stigmatized targets, only
two experiments have demonstrated the effects of manipulating attributional
beliefs on participants’ subsequent attitudes toward a stigmatized group (see
Crandall, 1994, Study 4 on anti-fat prejudice; Piskur & Degelman, 1992, on
heterosexism). Neither study included measures of premanipulation atti-
tudes, and neither reprinted the causal information used to manipulate
attributional beliefs. Piskur and Degelman failed to show that attributional
1026 HEGARTY AND GOLDEN

beliefs had been manipulated and to control the amount of information

presented. They also observed significant results among women only.
Other experiments have suggested that manipulations of attributional
beliefs do not always affect attitudes as attribution theory would predict.
Participants who are led to believe that schizophrenia is an uncontrollable
condition do not regard perceived schizophrenic targets any more positively
(Corrigan et al., 2001), and participants who imagine having a gay son are
more compassionate when they believe homosexuality to be controllable
(Armesto & Weisman, 2001). In short, direct effects of attributions on group
attitudes have generally been assumed to exist on the basis of correlational
evidence and a small, mixed body of experimental findings. This evidence is
not sufficient as a basis for informed evidence-based stigma-reduction efforts
that involve manipulations of attributional thinking. Recent development of
the justification–suppression model of prejudice casts further doubt on the
assumption that attributional beliefs and attitudes are correlated only
because attributional beliefs have an effect on attitudes.

The Justification–Suppression Model of Prejudice

In a new synthesis of prejudice theories, Crandall and Eshleman (2003)

distinguished a range of factors that lead what they call genuine prejudice to
be either suppressed or justified. Crandall and Eshleman defined genuine
prejudice as “pure, unadulterated, original, unmanaged, and unambivalently
negative feelings toward members of a devalued group” (p. 418). Genuine
prejudice is distinct from the expression of prejudice, and the expression of
prejudice is assumed to be inconsistent with conformity to social norms of
fairness and egalitarianism. To resolve this discrepancy between private,
negative gut-level feelings and social norms, Crandall and Eshleman claimed
that individuals both suppress genuine prejudices, and justify their prejudices
with a range of cognitive rationalizations when they are expressed.
A full review of Crandall and Eshleman’s (2003) model is beyond the
scope of the present paper. However, causal attributions are described in the
model as a possible means of justifying expressed prejudices. Attributions of
stigmatized traits to controllable causes make the expression of prejudice
toward stigmatized people appear to be less noxious (see Rodin, Price,
Sanchez, & McElligot, 1989). Thus, any discrepancy—or dissonance—
between an individual’s genuine prejudice and the social norm to treat a
stigmatized group without prejudice could be resolved by constructing the
relevant stigmatized trait as controllable.
Crandall and Eshleman (2003) allowed that prior attributional beliefs can
affect prejudicial attitudes, as predicted by attribution theory. But they also
 ATTRIBUTIONS AND STIGMA 1027

suggested that individual differences in pre-existing prejudices toward a stig-

matized group may affect whether and how perceivers of that group engage
in attributional thinking about the trait that defines group membership.
Thus, their theory allows an alternative explanation of widespread correla-
tions between prejudicial attitudes toward a stigmatized group and belief in
the controllability of the relevant stigmatized trait. More prejudiced persons
may be motivated to construct that trait as controllable to square their
prejudicial reactions with the norm to be nonprejudiced.

The Present Study

As noted previously, most studies examining attributions about stigma-

tized traits and attitudes toward stigmatized groups have been correlational,
rather than experimental. To address this, we conducted an experiment that
systematically tested both attribution theory and system-justification theory
predictions. We measured participants’ prejudice levels, and exposed them to
information that manipulated their attributional beliefs about one of four
stigmatized traits: homosexual orientation, obesity, depression, and alcohol-
ism. Next, we allowed participants to write about this information to assess
the amount and content of their attributional thinking. System-justification
theory predicts that highly prejudiced participants will spontaneously engage
in more attributional thinking, particularly forms of thinking that assume
the controllability of the stigmatized trait. We also measured participants’
attitudes after they had completed the thought-listing task. Attribution
theory predicts that participants should evince more positive attitudes
after reading that the relevant stigmatized trait was controllable, rather than
uncontrollable.

Method

Participants

Study participants were 166 individuals (97 female, 69 male) who were
recruited from a major university in the United Kingdom through flyers and
classroom announcements (age range = 18 to 56 years; M age = 23.2 years).
All of the participants were students or workers at the university, but none of
them were psychology students. The participants received £5 (approx. $8 US
at the time of the study; approx. $10 US today) in return for their participa-
tion, and a raffle ticket for an additional prize of £50 (approx. $80 US at the
time of the study; approx. $100 US today).
1028 HEGARTY AND GOLDEN

Design

The experiment employed a 2 ¥ 4 design. Participants read about one

of four stigmatized traits: homosexual orientation, obesity, depression, or
alcoholism. Each of the traits was presented as either controllable or
uncontrollable.

Materials

There were three components to the materials: the pre-manipulation

questionnaire; the texts, which manipulated attributional beliefs; and the
post-manipulation measures. The pre-manipulation questionnaire was
entitled “General Attitudes Survey.” It consisted of 24 thermometer items
assessing attitudes toward 24 social groups, including the four stigmatized
groups of interest: gay men and lesbians, obese persons, depressed people,
and alcoholics. The name of each stigmatized group was presented above a
10-cm unmarked horizontal scale ranging from 0 (unfavorable) to 100 (favor-
able). Participants were instructed to mark the scale at the point that best
represented their attitude, (Campbell, 1971; Haddock & Zanna, 1993; Maio,
Esses, & Bell, 2000; Stangor, Sullivan, & Ford, 1991).
Each of the eight texts used to manipulate attributional beliefs was three
paragraphs long. The first paragraph defined the stigmatized trait, described
its prevalence, and specified negative consequences that can follow from
possessing the trait (including stigmatization). The second paragraph
described the results of bogus behavioral genetics and biological studies. In
the uncontrollable condition, this information supported the idea that the
trait was biologically determined. In the controllable condition, it refuted
biological determinism. The third paragraph described findings about the
controllability of the trait. In the controllable condition, this information
supported the idea that the trait could be changed; while in the uncontrol-
lable condition, it supported the idea that the trait could not be changed.
Texts varied in length from 343 words to 393 words. For each stigmatized
group, the two texts differed at most by 7 words. (See Appendix A for the
obesity texts. All of the other texts are available from the first author upon
request.)
The study included several post-manipulation materials. First, a thought-
listing task was introduced with the following instructions:
Please list the thoughts that occurred to you while you were
reading the essay. These thoughts may have been relevant to the
essay or irrelevant to the essay. Please list at least 5 below, using
one line for each thought, up to a maximum of 12.
 ATTRIBUTIONS AND STIGMA 1029

There were 12 numbered blank lines that followed. This task afforded a
measure of the degree to which participants’ attributional thinking alighted
on controllable and uncontrollable causes of the relevant stigmatized trait.
We used these thoughts both as a manipulation check and as a means of
testing the justification–suppression model.
Second, participants’ comprehension of the texts was assessed using 12
True–False forced-choice items. Some of these items referred to information
that was common across both forms of the texts (4 items in the homosexual
orientation condition; 5 in the alcoholism condition; and 6 in both the obesity
and depression conditions). Other items referred to distinct controllability-
specific information, and the correct answer to these items depended on the
manipulation to which the participant had been exposed (8 in the homo-
sexual orientation conditions; 7 in the alcoholism condition; and 6 in both the
obesity and depression conditions). (See Appendix B for the terms used in the
obesity conditions. All of the others are available from the first author upon
request.)
Third, participants were presented with an inventory assessing their
evaluation of the texts. The inventory contained eight items that were pre-
sented in 7-point Likert-type format. The obesity items are as follows:
1. The article extract was informative about obesity.
2. The article was easy to read.
3. The article extract was difficult to understand. (reverse-scored)
4. The article extract seems to have come from a science textbook or
journal.
5. The argument about the cause of obesity was convincing.
6. The article had insufficient detail about obesity. (reverse-scored)
7. The article has increased my knowledge about people who are
obese.
8. From your reading of the article extract, how favorable would you
consider the author be toward people who are obese as a group?
Equivalent items about homosexuality, alcoholism, and depression were pre-
sented in the relevant conditions.
Fourth, post-manipulation attitudes were assessed using three types of
measures. A thermometer item was presented about the relevant stigmatized
group. Next, a personal-stereotype item was presented (Eagly, Mladinic, &
Otto, 1991; Esses, Haddock & Zanna, 1994). The name of the relevant
stigmatized group was presented. Participants were asked to write down as
many as three terms that characterize the group, to describe the percentage of
the group that is described by each term, and to rate the valence of each term
on a 5-point scale ranging from -2 (extremely negative) to +2 (extremely
positive). Then, attitudes were assessed using standardized scales.
1030 HEGARTY AND GOLDEN

In the sexual orientation conditions, the short form of Herek’s (1984,

1994) Attitudes Toward Lesbians and Gay Men (ATLG) scale was used. All
nine items were presented in 10-point Likert-type format.3 In the obesity
condition, the dislike subscale of Crandall’s (1994) Anti-Fat Attitudes (AFA)
questionnaire was presented. All seven items were presented in 10-point
format. In the alcoholism and depression conditions, five semantic-
differential items selected from Crisp, Gelder, Rix, Meltzer, and Rowlands
(2000) were used. The instructions are as follows: “Think now of someone
who is an alcoholic [depressed]. Please circle the number to show which point
on the scale from 1 to 7 best describes an alcoholic [depressed person] for
each of the following items.” The five items were dangerous–not dangerous,
unpredictable–predictable, hard to talk with–easy to talk with, would improve if
given treatment–would not improve if given treatment, and feel the way we all
do at times–feel different. Thus, the first two attitude measures (i.e., the
thermometer item and the personal stereotype item) allowed comparisons
across stigmatized traits, while the third measure (i.e., the standardized atti-
tude scale) did not.

Procedure

Upon arriving at the laboratory, participants were greeted by the second

author. She briefed them that they would complete some general social
survey questionnaires and evaluate the comprehensibility of an extract from
an A-level psychology textbook by reading the extract and completing some
questionnaires about it.4
Participants sat at individual desks, and the experimenter distributed the
pre-manipulation materials. When these were completed and collected, the
texts were distributed. Participants were randomly assigned to a condition.
They completed the thought-listing task, the manipulation check question-
naire, and the items evaluating the text in sequence. They next completed a
distractor task in which they had to list the 12 nations that use the Euro
currency. Finally, the questionnaires assessing post-manipulation attitudes
with thermometer measures, personal stereotype measures, and standardized
scales were distributed. Experimental sessions included between one and
eight participants and lasted about 30 min.

3
The short form of Herek’s (1994) ATLG usually contains 10 items. One item (“State laws
prohibiting private, consenting lesbian behavior should be loosened”) was not included, as it is
not relevant to the British legal context.
4
British students take A-level exams at the end of their secondary education in various
chosen subjects, including psychology.
 ATTRIBUTIONS AND STIGMA 1031

After completing all of the materials, participants were carefully debriefed

to check for suspicions about the purposes of the experiment. Process
debriefing (Ross, Lepper, & Hubbard, 1975) was enacted by discussing the
aims and design of the study, presenting participants with the alternate
versions of the relevant text and encouraging them to consider its plausibility.
A fact sheet about the relevant stigmatized group was also distributed. Par-
ticipants were asked not to discuss the study with others; and they were
thanked, paid, and dismissed.

Results

Overview

The present study tested two hypotheses. We tested the suppression–

justification theory’s hypothesis by examining correlations between pre-
existing prejudices and spontaneous attributional thinking, as measured by
the thought-listing task. We tested the attribution theory’s hypothesis by
examining effects of the attributional belief manipulation on each of the three
measures of participants’ post-manipulation attitudes. When testing attribu-
tion theory, we used 2 ¥ 4 ANOVAs using the controllability manipulation
(controllable vs. uncontrollable) and stigma type (alcoholism, depression,
homosexuality, or obesity) as independent variables to test the attribution
theory hypothesis. However, as the standardized measures of attitudes were
not comparable across conditions, we used t tests to compare the post-
manipulation attitudes in the controllable and uncontrollable conditions for
each stigma group.
All of the post hoc tests that we report here are Tukey’s honestly signifi-
cant difference tests with an alpha level of .05. When analyzing post-
manipulation attitude scores, we included pre-manipulation attitude scores
as a covariate. We also conducted several manipulation checks.

Pre-Manipulation Attitudes

Attitudes on the 101-point thermometer measure were scored from 0 to

100 by examining where participants marked each 100-mm line. A 2 ¥ 4
ANOVA using the controllability manipulation (controllable vs. uncontrol-
lable) and stigma type (alcoholism, depression, homosexuality, or obesity)
as independent variables reveals only a main effect of stigma group,
F(3, 158) = 11.46, p < .001. Post hoc tests show that alcoholics were viewed
significantly more negatively than were lesbians and gay men (Ms = 27.00
1032 HEGARTY AND GOLDEN

and 51.95, respectively). Attitudes toward obese people and depressed people
were intermediate and were not significantly different from the means
for either alcoholics or lesbians and gay men (Ms = 46.88 and 50.50,
respectively).

Attributional Thoughts About Texts

The 168 participants produced 1,049 thoughts in total, which were coded
independently by two coders in two stages. First, all thoughts were coded as
attributional if they referred to the cause or to the controllability of the
stigma; or non-attributional if they did not refer to these characteristics of the
stigma. The coders agreed initially on the coding of 93% of cases and resolved
disagreements easily through discussion. In all, 35.7% of the thoughts were
coded as attributional. A 2 ¥ 4 ANOVA using stigma group and controlla-
bility manipulation as independent variables reveals no main effects or inter-
actions on either the total number of thoughts produced (M = 6.33), or the
total number of attributional thoughts produced (M = 2.24; all Fs < 2.10, all
ps > .10).
Next, all attributional thoughts were coded as implying controllability,
uncontrollability, or as ambiguous thoughts. Coders agreed initially on 92%
of cases, and disagreements were easily resolved, as before. Ambiguous
thoughts accounted for 10.5% of all attributional thoughts. For each partici-
pant, we calculated the total number of thoughts that implied controllability
or uncontrollability (Ms = 1.18 and 0.83, respectively).
To assess whether attributional thinking was manipulated effectively by
the texts, a 4 ¥ 2 ¥ 2 ANOVA was conducted, with stigma trait (alcoholism,
depression, homosexuality, or obesity) and attributional belief manipulation
(controllable vs. uncontrollable) as between-subjects factors, and type of
attributional thought (controllability implied vs. uncontrollability implied)
as a within-subjects factor. Participants’ thoughts implied that the stigma-
tized traits were controllable more often than uncontrollable, F(1,
158) = 6.80, p = .01. This effect was moderated by a significant interaction
involving the controllability manipulation, F(1, 158) = 20.49, p < .001. Sig-
nificantly more thoughts implying controllability were produced by those
who read that the stigmatized traits were controllable (Ms = 1.45 and 0.92,
respectively), but significantly more thoughts implying uncontrollability were
produced by participants who read that the stigmatized traits were uncon-
trollable (Ms = 1.19 and 0.46, respectively). In other words, the texts effec-
tively manipulated participants’ causal thinking.
Next, we used the thoughts to test the justification–suppression hypoth-
esis that prejudiced persons spontaneously attribute stigmatized traits to
 ATTRIBUTIONS AND STIGMA 1033

controllable causes. Participants’ pre-manipulation attitudes toward the

stigmatized groups were uncorrelated with the total number of thoughts
produced (r = .013), but were negatively correlated with the total number
of attributional thoughts produced (r = -.182, p < .02). Attitudes were unre-
lated to the production of thoughts implying uncontrollability (r = -.029),
but were negatively correlated with the production of thoughts implying
controllability in the experiment as a whole (r = -.255, p = .001), and in the
controllable and uncontrollable conditions (r = -.259 and -.264, respectively,
both ps < .05). In other words, prejudiced participants spontaneously listed
more thoughts about the controllable origins of the stigmatized traits in
all conditions. This finding is consistent with the justification–suppression
model, which predicts that more prejudiced persons are the most motivated
to engage in such thinking to rationalize their prejudices.

Comprehension

On average, participants answered 88.1% of the common comprehen-

sion items correctly. They comprehended more of the alcoholism texts
(M = 94.5%) than the texts about obesity, homosexual orientation, or depres-
sion (Ms = 85.5%, 86.4%, and 86.1%, respectively), F(3, 157) = 3.94, p = .01.
However, there was no effect of controllability manipulation or interaction
(both Fs < 1.50, both ps > .20).
We calculated the percentage of each participant’s responses to the dis-
tinct controllability items implying that the stigma was controllable. Unsur-
prisingly, participants’ responses suggested controllability much more after
reading the controllability texts, compared to the uncontrollability texts,
F(3, 157) = 412.38, p < .001 (Ms = 82.1% and 21.8%, respectively). Partici-
pants also endorsed fewer items about the controllability of sexual orienta-
tion than about the controllability of obesity, F(3, 157) = 3.24, p < .05
(Ms = 48.6% and 58.5%, respectively). There was no significant interaction
between controllability manipulation and stigma group (F < 1.40, p > .25). In
short, the participants understood the vignettes, including those portions that
were used to manipulate their attributional beliefs.

Evaluation of Texts

The eight items used to assess evaluation of the texts were not sufficiently
correlated to form a single measure (Cronbach’s a = .49) and were analyzed
separately. Stigma group affected responses to three items. Responses to Item
1 show that the sexual orientation texts were viewed as significantly less
1034 HEGARTY AND GOLDEN

informative (M = 5.11) than were the obesity and depression texts (Ms = 5.85
and 5.78, respectively), F(3, 158) = 3.61, p < .05. Responses to Item 2 show
that the sexual orientation texts were viewed as significantly harder to read
than were the obesity texts (M = 5.64 and 6.25, respectively), F(3, 158) = 2.77,
p < .05. Finally, a significant effect was observed with regard to Item 7, F(3,
158) = 2.70, p < .05, but post hoc tests reveal no clear significant differences
between conditions.
The controllability manipulation affected only responses to Item 7. Par-
ticipants in the uncontrollable conditions perceived that they learned more
than did participants in the controllable conditions, F(1, 158) = 7.69, p < .01
(Ms = 5.14 and 4.46, respectively). This finding, along with the tendency to
spontaneously produce thoughts implying controllability, suggests that attri-
butions of stigma to controllable factors were the default among these par-
ticipants. No other main effects of controllability approached significance (all
Fs < 2.70, all ps > .10). None of the interaction effects were significant (all
Fs < 1.47, all ps < .22).
Thus, the participants both understood information that the texts com-
municated about controllability, and spontaneously produced attributional
thoughts consistent with our predictions. In other words, the texts were
effective. Under such conditions, attribution theory predicts that attitudes
toward stigmatized groups should be affected. We next examined if this was
the case by examining participants’ attitudes as assessed by thermometer
scales, personal stereotypes, and standardized measures.

Post-Manipulation Thermometers

We first tested for effects of the attributional belief manipulations on

attitudes by examining the post-manipulation thermometer scores. A 2
(Stigma Controllability) ¥ 4 (Stigmatized Trait) ANOVA was conducted
using pre-manipulation thermometer scores as a covariate. The effect of
the covariate was significant, F(1, 156) = 151.56, p < .001; but neither of the
main effects nor the interaction approached significance (all Fs < 1.10, all
ps > .35).5 Thermometer scores were equivalent after reading that the stigma-
tized traits were controllable or uncontrollable (Ms = 50.3 and 49.7, respec-
tively). This result failed to confirm attribution theory.

5
We also examined change scores on the thermometer measure. Attitudes toward the stig-
matized groups were more positive after reading the texts than before, t(164) = 4.32, p < .001. A
2 ¥ 4 ANOVA using change scores as a dependent variable reveals no significant effects of
stigma group, controllability manipulation, or interaction between them (all Fs < 1.70, all
ps > .17).
 ATTRIBUTIONS AND STIGMA 1035

Personal Stereotypes

Personal stereotypes for each participant for each stigma were computed
according to the formula

∑ ( p × v) n
where p is the percentage of the group judged to share the trait (i.e., 0 to 100),
v is the valence of the trait (i.e., -2 to 2), and n is the total number of traits
described. Thus, personal stereotype scores ranged from -200 to +200. A
2 ¥ 4 ANOVA including pre-manipulation thermometer scores as a covariate
was conducted. Once again, the effect of the covariate was significant, F(3,
157) = 9.91, p < .01. In addition, there was a significant main effect of stig-
matized group, F(3, 157) = 10.08, p < .001. Post hoc tests show that partici-
pants had significantly more positive stereotypes of gay men and lesbians
(M = 21.40) than of obese people and depressed people (Ms = -36.07 and
-48.01, respectively). Stereotypes of alcoholics were significantly more nega-
tive than were stereotypes of all other groups (M = -91.38). However, there
was no effect of the attributional belief manipulation, nor any interaction
between stigma group and attributional belief manipulation (both Fs < 1).
Personal stereotypes were equally negative after reading that the stigmatized
traits were controllable or uncontrollable (Ms = -36.54 and -37.25, respec-
tively). This result also failed to confirm attribution theory.

Standardized Measures of Attitudes

Next, we examined the standardized measures of attitudes. As these dif-

fered by stigma group, we examined each stigma group separately. The
appropriate ATLG items were reverse-coded, and all items were averaged to
form a reliable measure (Cronbach’s a = .89). High scores reflect higher levels
of heterosexism. Pre-manipulation thermometer scores were highly corre-
lated with ATLG (r = -.81, p < .001) and were included as a covariate in a
one-way ANOVA examining the effect of the controllability manipulation on
ATLG. A significant effect of the covariate, F(1, 41) = 73.49, p < .001, but no
effect of the controllability manipulation was observed (F < 1). Contrary to
attribution theory predictions, ATLG scores were lower, but not significantly
so, after reading that homosexual orientations were controllable, rather than
uncontrollable (Ms = 2.47 and 3.11, respectively).
The appropriate AFA items were reverse-coded and averaged to form a
reliable measure (a = .73). High scores reflect higher anti-fat prejudice. Pre-
manipulation thermometer scores were highly correlated with AFA (r = -.81,
1036 HEGARTY AND GOLDEN

p < .001) and were included as a covariate in a one-way ANOVA, as

mentioned previously. Again, the effect of the covariate was significant,
F(1, 39) = 6.00, p < .01, but not the main effect of controllability informa-
tion (F < 1). AFA scores were marginally higher after reading that body
weight was controllable, rather than uncontrollable (Ms = 3.41 and 2.61,
respectively).6
The semantic-differential items applied to depression and alcoholism were
not consistently correlated with each other, and so were analyzed as indi-
vidual items. Pre-manipulation thermometer scores were largely uncorrelated
with these items, so they were not included as covariates in the analysis. In the
alcoholism and depression conditions, responses on all of the semantic-
differential items were equivalent across both conditions (all |t| < 1.60, all
p > .10), with one exception. Participants judged that depressed people “feel
as we do” somewhat more in the uncontrollable condition than in the con-
trollable condition, F(1, 40) = 3.08, p < .09. Thus, across all conditions, none
of the data from any of the standardized attitude measures supported attri-
bution theory.

Discussion

This experiment tested two hypotheses about the relationship between

prejudicial attitudes toward stigmatized groups and attributional beliefs
about stigmatized traits. Contrary to attribution theory, manipulations of
attributional beliefs had no effect on attitudes toward stigmatized groups.
However, consistent with the justification–suppression model, prejudiced
people spontaneously produced a greater number of causal thoughts imply-
ing that the stigmatized traits were controllable.
This is one of very few experiments to have directly tested the hypothesis
that causal attributions affect attitudes toward stigmatized groups. Attribu-
tions may affect attitudes in some situations, but both the present data and
other published studies (e.g., Armesto & Weisman, 2001; Corrigan et al.,
2001) show that attribution theory’s predicted effects do not always occur. It
is entirely possible that the attributional-belief manipulation used here was
insufficient to produce attitude change, and that other interventions might be
more effective. However, like others who have observed effects consistent
with attribution theory (e.g., Crandall, 1994; Piskur & Degelman, 1992), we
used bogus information to manipulate attributional beliefs, and observed

6
None of the statistical conclusions about the lack of an effect of the manipulation on
attitudes were affected by running the analysis again without including pre-manipulation atti-
tudes as a covariate, with one exception. A marginally significant effect of the manipulation on
AFA scores was observed, t(38) = 1.76, p < .09.
 ATTRIBUTIONS AND STIGMA 1037

attitudes immediately afterward. Unlike other researchers, we examined

reactions to several groups. We also used several attitude measures, and
showed clear effects of our manipulation on spontaneous causal thinking.
Our results do not imply that manipulations of attributional beliefs never
affect prejudicial attitudes in the manner that attribution theory predicts.
However, researchers who have theorized along the lines of attribution
theory have relied too heavily on correlational findings as the basis of their
recommendations to promote biological information about stigmatized traits
in the service of stigma reduction (e.g., Anderson et al., 1996; Crocker et al.,
1998; Finell, 2002; Sakalli, 2002; Weiner, 1995; Whitley, 1990). However,
biological theories implying that stigmatized identities are uncontrollable
also imply that stigmatized groups are separate and distinct. Their use in
stigma-reduction interventions may engender essentialist thinking, and may
even backfire (Haslam et al., 2000, 2002; Hegarty & Pratto, 2001; Keller,
2005). Applied researchers and the stigmatized groups that they are aiming to
help deserve a much higher standard of contextual, nuanced thinking backed
up by clear experimental results in this domain.
In the present experiment, people with differing levels of prejudice spon-
taneously produced different kinds of attributional thinking about stigma-
tized identities when prompted to do so. These findings are consistent with
the justification–suppression model, which specifies that prejudiced persons
alight on controllable causes of stigmatized traits to resolve discrepancies
between pre-existing prejudices and norms to appear nonprejudiced. Other
studies have also suggested that attributional thinking is related to prejudice
because of its capacity to express tolerance or prejudice, rather than because
of its inherent propositional content.
In individualistic cultures, judgments of personal responsibility are more
central to impression formation than in collectivistic cultures; and stronger
relationships between causal attributions and attitudes toward stigmatized
groups are observed in individualistic cultures (Crandall et al., 2001;
Crandall & Martinez, 1996). Within individualistic cultures, people vary in
the degree to which they believe that attributions to uncontrollable beliefs
express tolerant or prejudicial attitudes. Correlations between causal attri-
butions and attitudes toward stigmatized groups are limited to those who
believe most strongly that attributions to uncontrollable beliefs express tol-
erant attitudes (Hegarty, 2002). Weiner (1996) argued that attribution theory
is so well supported by research evidence that little can be gained by further
exploring the effects of attributions on emotional reactions to individual
targets. We broadly agree and urge researchers to consider the opposite
process by which attributional thinking is prompted by discrepancies
between an individual’s prejudices and his or her adherence to social norms
to be nonprejudiced. Researchers might further consider the ideological work
1038 HEGARTY AND GOLDEN

that the resolution of those discrepancies through the medium of attribu-

tional beliefs might be doing.
While our claims that prejudice affects causal attribution more than the
reverse may seem problematic for attribution theory, this is not necessarily
the case. Rather, the justification–suppression model may simply be remind-
ing attribution researchers of an aspect of the attributional process that has
received comparatively little research attention. Weiner et al. (1988) specified
that “negative events or effects in particular initiate attributional search”
(p. 739). Subsequent research has paid much attention to the kinds of attri-
butions that people alight on once a search has been initiated, often at an
experimenter’s specific request. But very little attention has been paid to the
question of when people initiate attributional thinking to begin with, and to
what variability in the depth of their interest in nature/nurture questions
might signify. Consistent with Weiner et al.’s original thinking, our results
show that attributional thinking is more common among initially prejudiced
persons who perceive stigmatized groups in the most negative terms. Attri-
butional thoughts may affect attitudes, but initially prejudiced attitudes
appear to initiate the attributional thinking process.

Limitations and Future Directions

The present study has many limitations, most of which are typical of
research in this area. First, we relied on a college sample, but we conscien-
tiously avoided recruiting psychology students. Second, our studies were
based in the United Kingdom, while most studies in this domain have been
conducted with samples in the United States. Previous research, at least in the
domain of sexual orientation, has shown correlations between attitudes and
attributional beliefs to be stronger in the U.S. than in the UK, and we
recommend that researchers located in the U.S. explore the effects of preju-
dice on attributional thinking (Hegarty, 2002). Third, like other researchers
(e.g., Crandall, 1994), we only examined immediate effects of our manipula-
tions, and long-term effects ought to be examined in future studies, too.
Finally, as we observed no effects of attributional belief manipulations on
attitudes, we were unable to answer questions about possible moderators or
mediators of those effects.
We encourage other researchers in this area to employ diverse samples
from more than one country to test theories across multiple stigmatized
groups. Research should explore the when, how, and why of the reciprocal
relationship between thinking about the cause of a stigmatized trait and
evaluations of the group that is characterized by that trait.
 ATTRIBUTIONS AND STIGMA 1039

References

Aguero, J. E., Bloch, L., & Byrne, D. (1984). The relationships among sexual
beliefs, attitudes, experience, and homophobia. Journal of Homosexuality,
10, 95–107.
Anderson, C. A., Krull, D. S., & Weiner, B. (1996). Explanations: Processes
and consequences. In E. T. Higgins & A. W. Kruglanski (Eds.), Social
psychology: Handbook of basic principles (pp. 271–296). New York:
Guilford.
Armesto, J. C., & Weisman, A. G. (2001). Attributions and emotional reac-
tions to the identity disclosure (“coming out”) of a homosexual child.
Family Process, 40, 145–161.
Campbell, A. (1971). White attitudes toward Black people. Ann Arbor, MI:
Institute for Social Research.
Cobb, M., & deChabert, J. T. (2002). HIV/AIDS and care provider attribu-
tions: Who’s to blame? AIDS Care, 14, 545–548.
Corrigan, P. W. (2000). Mental health stigma as social attribution: Implica-
tions for research methods and attitude change. Clinical Psychology:
Research and Practice, 7, 48–67.
Corrigan, P. W., Markowitz, F. E., Watson, A., Rowan, D., & Kubiak,
M. A. (2003). An attribution model of public discrimination towards
persons with mental illness. Journal of Health and Social Behavior, 44,
162–179.
Corrigan, P. W., River, L. P., Lundin, R. K., Penn, D. L., Uphoff-Wasowski,
K., Campion, J., et al. (2001). Three strategies for changing attributions
about severe mental illness, Schizophrenia Bulletin, 27, 187–195.
Crandall, C. S. (1994). Prejudice against fat people: Ideology and self-
interest. Journal of Personality and Social Psychology, 66, 882–894.
Crandall, C. S., D’Anello, S., Sakalli, N., Lazarus, E., Nejtardt, G. W., &
Feather, N. T. (2001). An attribution-value model of prejudice: Anti-fat
attitudes in six nations. Personality and Social Psychology Bulletin, 27,
30–37.
Crandall, C. S., & Eshleman, A. (2003). A justification–suppression model of
the expression and experience of prejudice. Psychological Bulletin, 129,
414–446.
Crandall, C. S., & Martinez, R. (1996). Culture, ideology, and anti-fat atti-
tudes. Personality and Social Psychology Bulletin, 22, 1165–1176.
Crandall, C. S., & Moriarty, D. (1995). Physical illness stigma and social
rejection. British Journal of Social Psychology, 34, 67–83.
Crisp, A. H., Gelder, M. G., Rix, S., Meltzer, H. I., & Rowlands, O. J. (2000).
Stigmatisation of people with mental illnesses. British Journal of Psychia-
try, 177, 4–7.
1040 HEGARTY AND GOLDEN

Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. T. Gilbert,
S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (Vol.
2, pp. 504–553). New York: Oxford University Press.
DeJong, W. (1980). The stigma of obesity: The consequences of naïve
assumptions concerning the causes of physical deviance. Journal of Health
and Social Behavior, 21, 75–87.
Dooley, P. A. (1995). Perceptions of the onset controllability of AIDS and
helping judgments: An attributional analysis. Journal of Applied Social
Psychology, 25, 858–869.
Eagly, A. H., Mladinic, A., & Otto, S. (1991). Are women evaluated more
favorably than men? An analysis of attitudes, beliefs, and emotions.
Psychology of Women Quarterly, 15, 203–216.
Ernulf, K. E., Innala, S. M., & Whitam, F. (1989). Biological explanation,
psychological explanation, and tolerance of homosexuals: A cross-
national analysis of beliefs and attitudes. Psychological Reports, 65, 1003–
1010.
Esses, V. M., Haddock, G., & Zanna, M. P. (1994). The role of mood in
the expression of intergroup stereotypes. The psychology of prejudice:
The Ontario Symposium (Vol. 7, pp. 77–101). Hillsdale, NJ: Lawrence
Erlbaum.
Fine, M., & Asch, A. (1988). Disability beyond stigma: Social interaction,
discrimination, and activism. Journal of Social Issues, 44, 3–21.
Finell, D. S. (2002). The case for teaching patients about the neurobiological
basis of addictions. Journal of Addictions Nursing, 12, 149–158.
Goffman, E. (1963). Stigma: Notes on the management of spoiled identity.
Englewood Cliffs, NJ: Prentice Hall.
Graham, S., Weiner, B., Giuliano, T., & Williams, E. (1993). An attributional
analysis of reactions to Magic Johnson. Journal of Applied Social Psy-
chology, 23, 96–1010.
Haddock, G., & Zanna, M. P. (1998). On the use of open-ended measures to
assess attitude components. British Journal of Social Psychology, 37, 129–
149.
Haslam, N., Rotschild, L., & Ernst, D. (2000). Essentialist beliefs about
social categories. British Journal of Social Psychology, 39, 113–127.
Haslam, N., Rotschild, L., & Ernst, D. (2002). Are essentialist beliefs asso-
ciated with prejudice? British Journal of Social Psychology, 41, 87–100.
Hegarty, P. (2002). “It’s not a choice, it’s the way we’re built”: Symbolic
beliefs about sexual orientation in the U.S. and Britain. Journal of Com-
munity and Applied Social Psychology, 12, 153–166.
Hegarty, P., & Pratto, F. (2001). Sexual orientation beliefs: Their relationship
to anti-gay attitudes and biological determinist attitudes. Journal of
Homosexuality, 41, 121–135.
 ATTRIBUTIONS AND STIGMA 1041

Herek, G. M. (1984). Attitudes towards lesbians and gay men: A factor

analytic study. Journal of Homosexuality, 10, 1–21.
Herek, G. M. (1994). Assessing heterosexuals’ attitudes toward lesbians and
gay men: A review of empirical research with the ATLG scale. In
B. Greene & G. M. Herek (Eds.), Lesbian and gay psychology: Theory,
research, and clinical applications. Psychological perspectives on lesbian
and gay issues (Vol. 1, pp. 206–228). Thousand Oaks, CA: Sage.
Keller, J. (2005). In genes we trust: The biological component of psycho-
logical essentialism and its relationship to mechanisms of motivated
social cognition. Journal of Personality and Social Psychology, 88, 686–
702.
Leach, C. W., Snider, N., & Iyer, A. (2003). “Poisoning the consciences of the
fortunate”: The experience of relative advantage and support for social
equality. In I. Walker & H. J. Smith (Eds.), Relative deprivation: Specifi-
cation, development, and integration (pp. 136–163). Cambridge, UK: Cam-
bridge University Press.
Maio, G. R., Esses, V. M., & Bell, D. W. (2000). Examining conflict between
components of attitudes: Ambivalence and inconsistency are distinct con-
structs. Canadian Journal of Behavioural Science, 32, 71–83.
Menec, V. H., & Perry, R. P. (1998). Reactions to stigmas among Canadian
students: Testing an attribution–affect–help judgment model. Journal of
Social Psychology, 138, 443–454.
Menec, V. H., & Weiner, B. (2000). Observers’ reactions to genetic testing:
The role of hindsight bias and judgments of responsibility. Journal of
Applied Social Psychology, 30, 1670–1690.
Piliavin, I. M., Rodin, J., & Piliavin, J. A. (1969). Good Samaritanism: An
underground phenomenon? Journal of Personality and Social Psychology,
13, 289–299.
Piskur, J., & Degelman, D. (1992). Effect of reading a summary of research
about biological bases of homosexual orientation on attitudes toward
homosexuals. Psychological Reports, 71, 1219–1225.
Reizenstein, R. (1986). A structural equation analysis of Weiner’s
attribution–affect model of helping behavior. Journal of Personality and
Social Psychology, 50, 1123–1133.
Rodin, M., Price, J., Sanchez, F., & McElligot, S. (1989). Derogation, exclu-
sion, and unfair treatment of persons with social flaws: Controllability of
stigma and the attribution of prejudice. Personality and Social Psychology
Bulletin, 15, 439–451.
Ross, L., Lepper, M. R., & Hubbard, M. (1975). Perseverance in self-
perception and social perception: Biased attributional processes in the
debriefing paradigm. Journal of Personality and Social Psychology, 32,
880–892.
1042 HEGARTY AND GOLDEN

Rush, L. J. (1998). Affective reactions to multiple social stigmas. Journal of

Social Psychology, 138, 421–430.
Sakalli, N. (2002). Application of the attribution-value model of prejudice to
homosexuality. Journal of Social Psychology, 142, 264–271.
Stangor, C., Sullivan, L. A., & Ford, T. E. (1991). Affective and cognitive
determinants of prejudice. Social Cognition, 9, 359–380.
Steins, G., & Weiner, B. (1999). The influence of the perceived responsibility
and personality characteristics on the emotional and behavioral reactions
to people with AIDS. Journal of Social Psychology, 139, 487–495.
Weiner, B. (1993). On sin versus sickness: A theory of perceived responsibil-
ity and social motivation. American Psychologist, 48, 957–965.
Weiner, B. (1995). Inferences of responsibility and social motivation.
Advances in Experimental Social Psychology, 27, 1–47.
Weiner, B. (1996). Searching for order in social motivation. Psychological
Inquiry, 7, 199–216.
Weiner, B., Perry, R. P., & Magnusson, J. (1988). An attributional analysis of
reactions to stigmas. Journal of Personality and Social Psychology, 55,
738–748.
Whitley, B. E., Jr. (1990). The relationship of heterosexuals’ attributions for
the causes of homosexuality to attitudes toward lesbians and gay men.
Personality and Social Psychology Bulletin, 16, 369–377.
Wilder, D. A., Simon, A. F., & Faith, M. (1996). Enhancing the impact of
counterstereotypic information: Dispositional attributions for deviance.
Journal of Personality and Social Psychology, 71, 276–287.
Zucker, G. S., & Weiner, B. (1993). Conservatism and perceptions of
poverty: An attributional analysis. Journal of Applied Social Psychology,
23, 925–943.

Appendix A

Obesity Text: Uncontrollable Condition

Obesity is a condition of excess fat. In the short term, everyone experi-

ences small fluctuations in body weight due to fluid retention or loss.
However, over half of the adult population in the UK is medically defined as
overweight, and 19% are clinically obese. Obesity is associated with more
than 30 medical conditions, including diabetes, coronary heart disease,
cancer, hypertension, and stroke. In addition to being the butt of “fat jokes”
and other kinds of prejudice, obese people often face discrimination from the
medical profession. Very few doctors specialize in the treatment of this
condition.
 ATTRIBUTIONS AND STIGMA 1043

Weight gain is highly visible when it occurs. But why do some people
become fat, while others do not? Recent medical research in the UK has
shown that 8 out of 10 children with two obese parents will become obese
themselves, as compared with 2 out of 10 children with two lean parents. But
is this due to nature or to nurture? Studies of adopted children have revealed
that their weight patterns are similar to those of their biological [adoptive],
rather than their adoptive [biological] parents. In addition, scientists have
found differences [have consistently found no differences] in the brain activi-
ties of obese and non-obese people in response to food stimuli. Therefore,
most scientists now agree that genes and other biological factors play a large
part in the development of obesity [are largely irrelevant in obesity]. Life
experience seems to play a minor role, if any at all [a much larger role].
Almost everyone has used a diet or an exercise regime at some point or
other to control their weight. But why don’t these strategies work all of the
time? At a simple level, weight is only lost when net energy expenditure
exceeds net energy intake over a prolonged period of time. However, humans
are hardwired with a strong drive to eat, and as a direct result of our lifestyle,
we engage in much less physical activity than did our evolutionary ancestors.
In addition, as a person loses weight, that person’s metabolism slows down. One
effect of this is that people on diets extract more calories from food than they do
normally. This makes weight loss difficult. In fact, many people gain more
weight than they lose as a result of dieting. The balance between eating and
exercise, and changes in humans’ metabolism while dieting explain why obesity
is such a difficult condition to change. [Obesity often results from an imbalance
between diet and exercise and can be cured by addressing this balance.
Organizations like Weight Watchers® re-educate obese people about healthy
eating habits and exercise, and many obese people who join such programs
have managed to lose weight and to keep it off. In other words, when obese
people manage to consistently eat less and to consistently exercise more,
obesity can be successfully treated.]
Note. Controllable condition items appear in brackets.
1044 HEGARTY AND GOLDEN

Appendix B

Comprehension Manipulation Check Items: Obesity Conditions

Common Items Across Texts

1. Over half of the UK population is overweight, and nearly 20% are

clinically obese. (True)
2. Obesity is a condition of insufficient body fat. (False)
3. Being obese puts you at increased risk of heart disease, cancer,
diabetes, and other serious health conditions. (True)
4. People who are obese often face discrimination and prejudice.
(True)
5. Studies have shown that adopted children are as likely to become
obese as their adoptive parents. (False)
6. Life experience plays a large role in the development of obesity.
(Controllable)

Controllability-Specific Items

1. Studies show significant differences in activity between the brains of

obese people and non-obese people in response to food stimuli.
(Uncontrollable)
2. Most scientists agree that obesity is a genetically caused condition.
(Uncontrollable)
3. Weight is only lost when net energy expenditure exceeds net energy
intake over a prolonged period of time. (True)
4. Obesity is treatable and can be cured. (Controllable)
5. Obesity can be successfully treated when obese people manage to
consistently eat less and exercise more. (Controllable)
6. People on diets extract more calories from food than they do nor-
mally. (Uncontrollable)