1. science
2. medicine
3. psychology and psychiatry

Copyright 1991 by the American Psychological Association, Inc.
0033-2909/91/i3.00
Psychological Bulletin
1991, Vol. L10, No. 1,109-128
What Is Beautiful Is Good, But. . .: A Meta-Anatytic Review of Research
on the Physical Attractiveness Stereotype
Alice H. Eagjy
Purdue University
Richard D. Ashmore
Rutgers—The State University of New Jersey
Mona G. Makhijani
Purdue University
Rutgers—The State University of New Jersey
Laura C. Longo
This review demonstrates that the physical attractiveness stereotype established by studies of person perception is not as strong or general as suggested by the often-used summary phrase what is
beautiful is good. Although subjects in these studies ascribed more favorable personality traits and
more successful life outcomes to attractive than unattractive targets, the average magnitude of this
beauty-is-good effect was moderate, and the strength of the effect varied considerably from study to
study Consistent with our implicit personality theory framework, a substantial portion of this
variation was explained by the specific content of the inferences that subjects were asked to make:
The differences in subjects* perception of attractive and unattractive targets were largest for indexes
of social competence; intermediate for potency, adjustment, and intellectual competence; and near
zero for integrity and concern for others. The strength of the physical attractiveness stereotype also
varied as a function of other attributes of the studies, including the presence of individuating
information.
One of the most widely cited conclusions from research on
ality traits as well as to life outcomes such as marital happiness
physical attractiveness is summarized by Dion, Berscheid, and
and career success. Indeed, subjects ascribed more favorable
Walster's (1972) claim that, in people's perceptions of others,
"what is beautiful is good" (p. 285). This statement linking
beauty and goodness suggests the existence of a stereotype
personality traits and more successful life outcomes to the pictured individuals to the extent that they were physically attractive.
The experimental paradigm introduced by Dion and her col-
whereby physically attractive individuals are believed to possess
a wide variety of positive personal qualities. In this article, we
integrate the available research on the physical attractiveness
leagues (1972) has guided many subsequent investigators, and
there is, thus, a relatively large body of research on the attractive-
stereotype to determine the extent to which the statement that
what is beautiful is good provides an accurate summary of peo-
ness stereotype. Most narrative reviewers of this literature have
agreed that the beauty-is-good stereotype is a strong and gen-
ple's inferences from cues that convey physical attractiveness.
In the classic study on the physical attractiveness stereotype,
Dion and her associates (1972) had subjects rate facial photo-
eral phenomenon (Adams, 1982; Alley & Hildebrandt, 1988;
Berscheid, 1981; Berscheid & Walster, 1974; Dion, 1981,1986;
Hatfield &Sprecher, 1986; Patzer, 1985). For example, Hatfield
graphs that had been selected on the basis of judges' agreement
and Sprecher wrote that "people believe good-looking people
that the pictured individuals were low, medium, or high in phys-
possess almost all the virtues known to humankind" (p. xix).
ical attractiveness. Subjects' ratings pertained to various person-
Most secondary sources, especially introductory social psychology textbooks, have likewise treated the beauty-is-good stereotype as a homogeneous, potent, and firmly established phenom-
This research was supported by National Science Foundation
Grants BNS-8616149 to Richard EX Ashmore and Frances K. Del
Boca, Co-Principal Investigators, and BNS-S605256 to Alice H. Eagl>;
Principal Investigator. A preliminary report of this research was reported at the annual meeting of the American Psychological Society,
June 1989. A table showing the effect sizes and study characteristics for
each study included in the meta-analysis is available from Alice H.
Eagly or Richard EX Ashmore.
We thank Thomas Alley, Ellen Berscheid, Linda Jackson, and three
anonymous reviewers for their comments on a draft of the article.
Correspondence concerning this article should be addressed to
Alice H. Eagly, Department of Psychological Sciences, Purdue University, West Lafayette, Indiana 47907-1364, or Richard D. Ashmore, Department of Psychology, Tillett Hall, Kilmer Campus, Rutgers University, New Brunswick, New Jersey 08903.
109
enon (e.g, see Baron & Byrne, 1987, p. 197; Deaux & Wrightsman, 1988, p. 250). In this article, we report the first use of
meta-analytic procedures to combine quantitatively the existing studies to examine the strength and generality of the attractiveness stereotype.
Although the general consensus in social psychology is that
an attractive target leads perceivers to make strong inferences of
personality goodness, reviewers have noted exceptions to this
phenomenon. Dion (1981,1986) suggested that the beauty-isgood effect is strongest for measures of social competence and
interpersonal ease, and Bassili (1981) reached a similar conclusion, arguing that the core of the physical attractiveness stereotype is social vitality or extroversion. Also, beginning with
Dermer and Thiel (1975), reviewers noted a dark side to the
110
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
stereotype. Cash and Janda (1984, p. 52) labeled this aspect the
"what is beautiful is self-centered" stereotype on the basis of
Therefore, the stereotype question can be regarded as more
their observation that attractive people may be thought to be
basic to social psychological analysis than the questions of
treatment and assessed characteristics. If the physical attractive-
vain and egotistical. What is needed at this point is a formal and
ness stereotype is not at least moderately strong and robust, it is
general scheme for understanding these qualifications and ex-
unlikely to underlie the effects of physical attractiveness on
treatment and people's characteristics.
Our decision to focus on the physical attractiveness stereo-
ceptions. As a step in this direction, we conceptualize the process underlying the attractiveness stereotype in terms of the
person-perception construct of implicit personality theory (see
type has two important implications—one methodological and
also Adams, 1982; McArthur, 1982). We then frame predictions
about variation in the beauty-is-good effect in terms of this
conceptual approach. These hypotheses are tested by means of
one theoretical. Methodologically, we targeted our meta-analy-
a quantitative review of the accumulated research. Before presenting the details of this review, we situate our discussion of the
beauty-is-good stereotype within the total body of research and
theory on physical attractiveness.
sis to include only studies that assess the physical attractiveness
stereotype in a relatively pure form, without the constraints of
particular role relationships such as those that regulate social
interaction between teachers and students or therapists and
patients. To examine stereotyping in a relatively role-free environment, we included in our review only those studies in which
subjects inferred the attributes of people to whom they were not
Social Psychology of Beauty
linked by particular social roles. Subjects' inferences from physical appearance in role-regulated environments should be con-
A large amount of research has accumulated on physical attractiveness (see Cash's, 1981, bibliography of almost 500 nonre-
sidered in future reviews that are conceptualized to take the
specifics of role relationships into account. In terms of theoreti-
dundant entries). Because this research area is multidisciplinary, the studies are methodologically diverse and pertain to a
cal implications, our focus on the physical attractiveness stereotype led us to formulate our review in terms of existing psycho-
variety of specific hypotheses. However, among studies formulated at a social psychological level of analysis, most research
logical research on impression formation and stereotyping. We
now turn to this formulation.
has addressed three broad questions: (a) Are attractive people
perceived differently than unattractive people? (b) Are attractive
people treated differently than unattractive people? (c) Do attractive people have different characteristics (i£., personality
traits, skills, behavioral tendencies) than unattractive people?
The first research area, the perception of beautiful people,
concerns stereotyping on the basis of physical appearance. The
second area, which concerns differential treatment as a function of physical attractiveness, examines looks-based discrimination, a phenomenon that depends on factors in addition to
stereotypes about physical attractiveness (e.g., the social roles
governing interaction). The third type of research focuses not
Physical Attractiveness Stereotype and Implicit
Personality Theory
Ashmore and Del Boca (1979; see also Ashmore, 1981; Ashmore, Del Boca, & Wohlers, 1986) argued that stereotypes can
be fruitfully conceptualized in terms of the concept of implicit
personality theory (D. J. Schneider, 1973). By treating stereotypes as part of ordinary social cognition (see also Hamilton,
1981), Ashmore and his co-workers placed stereotypes within
the larger set of knowledge structures that individuals use to
make sense of other people's behavior. More specifically, im-
on people's reactions to targets, but on the characteristics of
attractive and unattractive individuals, as measured by various
plicit personality theories are hypothetical cognitive structures
methods of psychological assessment (e.g., personality tests,
self-reports of behavior, observations of behavior). The findings
of this third type of research have often been interpreted in
ality traits) and inferential relations that specify the degree to
which these attributes covary. The link between such implicit
theories and stereotypes about group members becomes apparent when we regard group membership as one of the personal
terms of self-fulfilling
prophecies (e.g, Snyder, Tanke, & Ber-
scheid, 1977) or a kernel of truth that underlies the attractiveness
whose primary components are personal attributes (e.g., person-
Our review concerns only the first area of social psychologi-
attributes inferentially associated with other attributes in such
an implicit theory (Ashmore, 1981). In fact, the inferential relations between group membership and other personal attributes
cal research on attractiveness: the perception of attractive and
can be regarded as synonymous with the stereotype itself. Thus,
unattractive individuals. There are several reasons for this
translating the usual definition of stereotype as a set of beliefs
focus. One reason is practical: A thorough review of all three
aspects of physical attractiveness research would require many
about the characteristics of group members into the language of
stereotype (e.g., Alley & Hildebrandt, 1988).
years to complete and would produce an outcome too large for
presentation in an article format. A second reason is strategic:
Meta-analyses are best targeted to research literatures that are
not defined extremely broadly and that have at least a moderate
amount of methodological coherence (see Mullen, 1989). A
third reason is analytic: The beauty-is-good stereotype is generally regarded as providing the major mechanism underlying
differential treatment based on attractiveness and the development of differential assessed characteristics, at least insofar as
these characteristics are mediated by self-fulfilling prophecies.
implicit personality theory, Ashmore and Del Boca (1979) defined stereotype as "a structured set of inferential relations that
link a social category with personal attributes" (p. 225).
The utility of this conceptualization has been demonstrated
for gender stereotypes. Using the implicit personality theory
approach, Ashmore and his colleagues identified the inferential
relations linking the social categories of male and female to
various dimensions of personality perception (see Ashmore,
1981). Specifically, the distinction between male and female
was closely associated with the personality dimension of
strong-weak (or potency). Also men were thought to possess
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
positive intellectual qualities such as scientific and negative social attributes such as sternness. In contrast, women were associated with negative intellectual qualities such as naivete and
111
tellectual competence, concern for others, integrity, adjustment
(i£, psychological stability), and potency.
The core of the social competence category is those interper-
From this same perspective, research on the beauty-is-good
sonal skills and traits concerned with sociability (e.g., extraversion) as well as the outcomes of such skills and inclinations (e.g.,
stereotype can be viewed as examining the inferential relations
popularity). Social competence, as here construed, corresponds
between physical attractiveness and personal attributes. The
closely with Bassili's (1981, p. 242) "social vitality" and Dion's
(1986, p. 8) "social competence and interpersonal ease." The
positive social attributes such as helpfulness.
social categories of attractive and unattractive people should
thus be associated in individuals' cognitions with various dimensions of personality. This approach contrasts with those
intellectual competence category centers on intellectual and
task-relevant ability (e.g, intelligent) and the results of such abil-
account only inferences from attractiveness to the general fa-
ity and motivation (eg, career success); this category also encompasses a hard and rational mental style (e.g., practical). Con-
vorability of personality. This presumed inference from attrac-
cern for others, like social competence, implies interpersonal
tiveness to evaluation is precisely described by the statement
skills. The emphasis, however, is not on sociability and popularity, but concern for others' welfare and the associated quality of
analyses of the physical attractiveness stereotype that take into
that beauty is good. This one-dimensional evaluative model
treats the different trait and life-outcome dependent variables
used in attractiveness research as if they indexed a single, abstract good-bad continuum. Although the notion of a contentfree evaluative dimension has much conceptual and empirical
value (e.g., Osgood, Suci, & Tannenbaum, 1957), research on
implicit personality theory has demonstrated that the evaluative dimension can be partitioned by perceivers into contentspecific types of evaluative meaning. Support for this partitioning of evaluative meaning exists in Rosenberg and his colleagues' research using multidimensional scaling (Kim &
Rosenberg, 1980; Rosenberg, 1977; Rosenberg, Nelson, & Vivekananthan, 1968; Rosenberg & Sedlak, 1972). According to this
research, one particularly important distinction between types
of evaluative meaning, at least for young adults, is between social and intellectual competence (Rosenberg, 1977; Rosenberg et
al, 1968). In addition, on the basis of his own and others' re-
selflessness. The positive side of this category involves being
emotionally supportive as well as sensitive and generous to
others. The negative side of this category, which includes traits
such as vain and egotistical, is the dark side of the attractiveness
stereotype acknowledged by some reviewers (eg., Adams, 1982;
Cash & Janda, 1984). The crucial issue for defining the integrity
category is honesty. This quality can take the form of interpersonal actions such as telling the truth (vs. being a liar) or abiding
by laws and norms (vs. breaking them). The adjustment category, which is a slight enlargement of Rosenberg's (1977) psychological stability category, includes not only normal (vs. abnormal) psychological functioning, but also other indicators of positive adult adjustment, such as self-esteem and good health.
Potency resembles Osgood and colleagues' (1957) potency or
strong versus weak factor. In interpersonal terms, potency implies dominance over other people (Ashmore, 1981; White,
1980).
search, Rosenberg identified the content-specific evaluative categories of maturity, concern for others, integrity, psychological
stability, and physical attractiveness. He also designated potency
Sources of the Physical Attractiveness Stereotype
as an important category that is sometimes relatively indepen-
Understanding how these types of evaluative meanings
dent of evaluation (Kim & Rosenberg, 1980; Rosenberg et al,
1968; see also Ashmore, 1981). In view of Rosenberg's inclusion
of physical attractiveness among these content-specific evalua-
might he mentally associated with physical attractiveness requires consideration of how people learn their implicit theories
of personality including their beliefs about what is associated
tive categories, analyzing the beauty-is-good stereotype within
with good looks. Two general types of inputs seem most impor-
the framework of implicit personality theory is especially appropriate.
tant: (a) direct observations of attractive and unattractive people
in oneTs social environment and (b) exposure to cultural repre-
This conceptual framework suggests that part of the varia-
sentations of attractive and unattractive people.1 With regard to
tion in the impact of physical attractiveness on subjects' infer-
direct observation, perceivers no doubt observe that better look-
ences about personality and life outcomes may be explained by
the specific content of these inferences (see also Bull & Rumsey,
1988, pp. 296-297). In particular, physical attractiveness may
be more strongly linked to some dimensions of implicit personality theories than to other dimensions. To test this possibility,
we developed a content-analysis scheme for classifying the de-
ing people receive more favorable reactions from others; for
example, attractive people are more popular with peers and
receive preferential treatment from others (see reviews by
Adams, 1982; Dion, 1981,1986; Feingold, 1990b; Patzer, 1985;
but see Bull & Rumsey, 1988; Langlois, 1986). Perceivers may
assume that these favorable reactions are elicited, not merely by
pendent variables used in studies of the physical attractiveness
stereotype into the components identified in research on implicit personality theory. This scheme takes Rosenbergls (1977)
evaluative categories as its starting point. Although Rosenberg's
maturity category could not be treated as a separate class because it was.seldom represented in research on the physical
attractiveness stereotype, dependent variables corresponding
to Rosenberg's other types were common enough to include
them as categories in the meta-analysis: social competence, in-
1
In addition, physical attractiveness may be aesthetically pleasing
and therefore may elicit positive affect that leads perceivers to infer,
perhaps to account for the affect, that attractive people have favorable
characteristics. Although there is little evidence that the sight of physically attractive people induces positive affect, researchers who manipulate affect by presenting photographic slides commonly include images
of physically attractive people among the stimuli that induce positive
affect (e.g., Larsen, Diener, & Cropanzano, 1987).
112
EAGLY, ASHMORE, MAKHUANI, AND LONGO
people's good looks, but by other positive aspects of people's
behavior and personality.
Depending in part on the impact that popularity and favorable treatment have on personality and behavior, perceivers
may observe that physical attractiveness is correlated with
various positive traits. Consistent with Feingold's (1990b) review, research on the true covariation between physical attractiveness and other personal attributes suggests that good looks
covary with social skills, social adjustment, and an absence of
shyness and social anxiety (e.g., Adams, 1977; Adams & Wareham, no date; Goldman & Lewis, 1977; Lerner & Lerner, 1977;
Pilkonis, 1977). Aside from this relation between physical attractiveness and social competence, evidence for true covariation between good looks and personal attributes is much less
convincing, although there is some suggestion of a weak positive correlation with adjustment (see Cash, 1985) and potency
(Adams, 1977; Chaiken, 1979,1986).
Turning to cultural messages about appearance, we suggest
that American culture associates beauty with good things and
ugliness with bad things. In children's television and books, the
wicked witch and evil giant are ugly and the heroic prince and
virtuous princess are attractive. In advertising, attractive models appear in positive settings (e.g, in happy crowds, as the object of admiration) and with valued possessions (e.g- fancy cars,
fashionable clothes). In addition, as shown by Downs and
Harrison's (1985) content analysis of the verbal content of television commercials, statements extolling the virtues of attractive appearance are quite common. \et in these cultural productions, beauty may be associated more often with some positive
qualities than others. Although appropriate content analyses of
media content are lacking, we suspect that the media most
often link good looks with sociability and popularity, especially
with the opposite sex. We doubt that the media consistently
associate physical attractiveness with integrity, concern for
others, adjustment, or potency. Cultural portrayals of individuals outstanding in these qualities are not clearly distinguished
by positive appearance, nor are beautiful people necessarily
exemplars of such qualities.
Despite the possibility that the media may portray attractive
people as successful in their careers, we doubt that physical
attractiveness is strongly linked with intellectual competence in
the popular culture. People depicted as outstanding in intellect
or accomplishment are not uniformly good-looking, nor are
good-looking people necessarily portrayed as smart and accomplished (e.g, those labeled as dumb blondes). A weaker impact of attractiveness on intellectual competence than on social
competence was supported by Ashmore, Tumia, and Schreier's
(1980) variation of the classic Asch (1946) experiment on the
effect of the adjectives warm or cold on impression formation.
Ashmore and colleagues substituted good-looking for warm and
unattractive for cold, and they found that this attractiveness
manipulation strongly influenced the ascription of social competence to the stimulus person but not the ascription of intellectual competence.
In understanding the content of the physical attractiveness
stereotype, we also need to take into account the inferences by
which perceivers may associate stereotypic attributes with other
personal attributes. In particular, given the strong link we have
demonstrated between attractiveness and social competence,
perceivers may infer on the basis of attractive people's presumed social competence that they have other favorable traits.
For example, given that popularity is valued, physical attractiveness may be viewed as a valued commodity and thereby as a
source of interpersonal power. As a consequence, perceivers
may associate beauty with interpersonal potency. In addition,
perceivers may assume that good looks are not only a source of
power but also major ingredients of happiness, self-esteem, and
the other aspects of adjustment. Yet, all these second-order inferences that people make from attractiveness may not lead to
positive impressions of beautiful targets. For example, concerning integrity, people might infer that attractive individuals are
exposed to more opportunities for interpersonal cheating (e.g,
married people having affairs; see Brigham, 1980). Also perceivers may assume that the popularity of attractive people
makes it unnecessary for them to develop nurturant qualities or
interpersonal sensitivity to maintain their friendships and other
relationships.
In conclusion, our reasoning suggests a number of principles
that might underlie the content and structure of the physical
attractiveness stereotype. Most important, good looks should
strongly imply social competence because of (a) the perception
that attractive individuals elicit positive reactions from others,
(b) the perception of true covariation between attractiveness
and social competence, and (c) the media portrayal of attractiveness as critical to heterosexual popularity and social attention.
In addition, good looks may imply a variety of other positive
qualities based on the inferential linkages we have noted and
possibly based to some extent on observations of true (or mediadepicted) covariation between attractiveness and other attributes. This analysis, therefore, does support the notion of a
general mental association between beauty and nonphysical
goodness. However, as our analysis has also implied, the tendency for good looks to be associated with positive qualities
may not be very strong overall and may vary considerably in
strength, depending on the particular dimension of personality
assessed.
Predictions for Meta-Analysis
Predictions about specific content of physical attractiveness
stereotype. Our primary predictions follow from the implicit
personality theory framework as well as from the additional
assumptions that we introduced about (a) social attention and
rewards received by beautiful people, (b) genuine covariation
between physical attractiveness and various attributes, (c) media content, and (d) perceivers' inferential linkages. These predictions concern the differential effects that physical attractiveness has on perceivers' inferences about the personality traits
that represent the various dimensions of their implicit theories
of personality. Although we predict that attractive targets
should be assigned more positive attributes than unattractive
targets, we hypothesize that this tendency will vary considerably as a function of type of evaluative meaning. Specifically for
the reasons we have indicated, information about physical attractiveness should have its strongest impact on inferences
about social competence. Bassili (1981) and Dion (1981,1986)
would probably agree with this prediction. Because of the likely
spillover that we described from social competence to potency
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
and adjustment, physical attractiveness should affect inferences
to these dimensions, albeit somewhat more weakly than inferences to social competence.
According to the reasoning we have presented, manipulations of physical attractiveness should have a relatively small
effect on inferences about intellectual competence and an even
weaker impact on integrity and concern for others. The impact
of attractiveness on concern for others may be especially weak
given that this category includes traits such as vanity and egotism. As we have already noted, reviewers have implicated such
traits in the dark side of the physical attractiveness stereotype
(Adams, 1982; Cash & Janda, 1984; Dermer & Thiel, 1975).2
These predictions suggest that the specific content of the
dependent variables used in studies of the physical attractiveness stereotype contributes to observed differences in the
strength of the stereotype. Studies that happened to include
many dependent variables assessing perceptions of social competence should show a stronger impact of physical attractiveness than studies containing higher proportions of dependent
variables selected from the other dimensions of personality.
The content of dependent variables should also account for variations within studies in the strength of subjects' inferences from
physical appearance to aspects of targets' personalities.
Other predictions. Because implicit theories of personality
comprise much more than beliefs about attractiveness, our
framework raises the issue of whether information that subjects
receive about attributes other than physical attractiveness may
also account for inconsistencies in the outcomes of studies. In
particular, receiving information that conveys attributes of the
target other than physical attractiveness may moderate inferences from physical attractiveness to personality traits.
In general, the beauty-is-good effect should become smaller
as the amount of other information about targets increases. This
prediction is consistent, first of all, with Bull and Rumsey's
(1988) survey of existing research. It is also congruent with general models describing how people combine items of information into an overall judgment. Specifically, averaging models
(see N. H. Anderson, 1974,1981) postulate that the weight or
importance of an item of information as a determinant of a
judgment becomes smaller as a perceiver takes more items of
information into account. Such combinatorial models predict
that the impact of physical attractiveness on judgments of personality lessens as more other information is received.
In analyzing the probable effect of giving subjects additional
information about the target, we must also consider the informativeness or diagnosticity of physical attractiveness compared
with other information that subjects may receive about targets.
In experiments on the physical attractiveness stereotype, subjects sometimes receive information about targets' personal history and personality, and, in videotaped or live presentations of
targets, subjects receive extensive information about targets'
personal style. This other information may very commonly be
more diagnostic than physical attractiveness in relation to the
traits and behaviors that subjects were asked to predict. This
argument is analogous to one made in relation to gender stereotypes, namely, that information other than sex may be more
strongly linked than sex to the various dimensions of personality that comprise perceivers' implicit theories. When perceivers
receive such information, their inferences from sex to personal-
113
ity traits should be weakened. This prediction has been repeatedly confirmed in research showing that perceivers' inferences
from sex to personality or behavior are reduced or even reversed by giving subjects other information about targets (e.g,
Deaux & Lewis, 1984; Del Boca & Ashmore, 1980; Eagry &
Steffen, 1984; Locksley, Borgida, Brekke, & Hepburn, 1980;
Locksley, Hepburn, & Ortiz, 1982).
The sex of the target is an item of information that subjects in
physical attractiveness experiments almost always possess. This
fact raises the question of whether the attractiveness stereotype
may be stronger for female targets than for male targets. Such a
trend would be consistent with reviewers' conclusions that attractiveness is more central to the gender role of women than of
men as well as to women's personal identity, self-esteem, and
interpersonal outcomes (see Bar-Tal & Saxe, 1976; Rodin, Silberstein, & Striegel-Moore, 1985). Moreover, the media more
often portray women than men as decorative and sexual objects
(see Courtney & Whipple, 1983; Roberts, 1982). In addition, in
television commercials explicit verbal statements linking
beauty and good outcomes are more commonly paired with
female actors than with male actors (Downs & Harrison, 1985),
and viewers of televised advertisements regard the women who
are depicted as more concerned with their looks than the men
are (Scheibe, 1979). Finally, Jackson, Sullivan, and Rostker
(1988) found that women, compared with men, rated physical
attractiveness as more important to them and reported engaging in more behaviors to improve their looks. Such findings
suggest that physical attractiveness may be more important in
perceiving women than in perceiving men.
Whether physical attractiveness is more important to perceivers of one sex than the other is not entirely clear. In view of
Feingold's (1990a) meta-analysis showing that men place greater
value on physical attractiveness than women do, especially
when romantic attraction is considered, ft is possible to argue
that physical attractiveness could be a more powerful dimension in men's than women's implicit theories of personality.
However, the research we review on the physical attractiveness
stereotype was not carried out in the context of romantic relationships. Moreover, the question addressed in research on the
physical attractiveness stereotype pertains to the meaning inferred from attractiveness cues, not the importance of attractiveness. Men and women may differ in the importance they accord to attractiveness even though they infer similar meaning
(e.g., social competence) from attractiveness cues. Particularly
in view of evidence that the content of gender stereotypes
differs little for male and female subjects (e.g., Deaux & Lewis,
1983,1984; Eagly & Steffen, 1984), we suspect that the physical
attractiveness stereotype is widely shared in American culture
and is therefore little affected by variation in subject characteristics, including subject sex. To the extent that some studies on
the physical attractiveness stereotype used male subjects and
2
One reason we do not predict a negative relation between attractiveness and concern for others is that the findings cited by some narrative
reviewers to provide evidence for the dark side of the physical attractiveness stereotype may not be representative of the total domain of findings on egotism and vanity. In addition, our concern for others class of
attributes is somewhat broader than egotism and vanity because ft
includes the positive facets of concern for others and selflessness.
114
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
others used female subjects, our meta-analytic method allows
us to approach this issue empirically by examining subjects' sex
as a predictor of their inferences about targets' personalities.
Method
Sample of Studies
Computer-based information searches were conducted using the following keywords: physical attractiveness, unattractive, beautiful, ugly,
ugliness, facial feature, and physical appearance. These keywords were
searched in the following data bases: PsycINFO (Psychological Abstract^, 1967 to 1987; ERIC (Educational Resources Information
Center), 1966 to 1987; Dissertation Abstracts Online, 1861 to 1987; ABI/
INFORM (a worldwide business and management data base), 1971 to
1987; Sociological Abstracts, 1963 to 1987; Mental Health Abstracts,
1969 to 1987; MEDLINE (Medical Subject Headings), 1966 to 1987;and
AARP (American Association of Retired Persons), 1985 to 1987. The
Social SciSearch data base was also searched to locate articles that
cited Dion and associates (1972) or Dermer and Thiel (1975). Each
search ended with the latest information available in the summer of
1987.
We also consulted Cash's (1981) bibliography, which contains almost
500 abstracts obtained through computer-assisted searches using the
keyword physical attractiveness. Finally, we searched through the reference lists of all located studies and carefully read numerous review
articles (e.g., BerscheidA Walster, 1974), books (eg., Patzer, 1985), and
chapters of books (e.g., Adams, 1982).
Selecting studies that focus on physical attractiveness stereotype.
These search procedures yielded over 600 possibly relevant reports.
Given our focus on the physical attractiveness stereotype (and not on
the treatment of attractive and unattractive people or their actual characteristics), we initially pared down the sample to include only studies
in which subjects rated attractive and unattractive targets on evaluative
attributes. The resulting sample was reduced further to include in the
final sample only studies that assessed the beauty-is-good stereotype in
situations unconstrained by particular role relationships (eg., teacherstudent). To select those studies that assessed the attractiveness stereotype in this relatively pure form, we implemented a number of additional exclusion rules:
1. Studies were included in the sample if the general physical attractiveness of target persons was one of its independent variables. Studies
were excluded if they instead examined the impact of specific physical
characteristics of target persons, such as height, weight, body type, or
disfiguring facial characteristics (e.g., Dibiase & Hjelle, 1968).
2. Studies were included only if physical attractiveness was defined
independently of the subjects who rated the targets' characteristics.
Attractiveness was in fact manipulated in almost all these studies. In a
few of the included studies, high- and low-attractiveness targets were
identified only after the subjects had participated, by attractiveness
ratings that were independent of subjects' own perceptions of targets'
physical attractiveness (e.g., Barnes & Rosemhal. 1985). We excluded
studies in which the attractiveness stereotype was defined only in
terms of subjects' own perceptions of targets* physical attractiveness;
that is, the stereotype was assessed by correlations between subjects'
own ratings of targets' physical attractiveness and their ratings of these
targets on traits and life outcomes (e.g., P. A. Schneider, Conger, &
Firth. 1986). Such studies are causally ambiguous and could illustrate
the "what is good is beautiful" principle (Gross & Crofton, 1977) if
subjects inferred good looks from positive personality characteristics.
3. Studies were included only if one or more of their dependent
variables was a rating of stimulus persons' personality traits (e.g.,
warmth, intelligence, trustworthiness) or life outcomes (eg., career success, marital success) on a dimension that could be considered evalua-
tive (i.e.. one pole could be identified as more positive than the other).
Consistent with our focus on partitioning evaluative meaning into its
components, we excluded studies if their dependent measures were not
clearly evaluative. For example, we eliminated studies that examined
only ratings of masculinity and femininity because they concern a
different hypothesis, what is beautiful is sex-typed (e.g. Cash & Janda,
1984), rather than what is beautiful is good. Similarly, we did not consider ascriptions of attitudes or group membership to targets (e.g, being a "feminist"; Jacobson & Koch, 1978) or causal attributions for
targets' success and failure (e.g.. Turkat & Dawson, 1976). We also excluded studies if their dependent measures assessed subjects' personal
reactions to targets (eg., ratings of whether the subject would like to
have the target as a date; Touhey, 1979) rather than the ascription of
evaluative characteristics to targets. Because the meta-analysis did not
concern the treatment of attractive and unattractive targets, we also
excluded studies using only behavioral measures (e.g., shocking a target; Miller, Gillen, Schenker, & Radlove, 1974). In addition, we eliminated the following types of dependent variables reported in studies
accepted for the sample: (a) ratings of targets' productions (eg., essays;
Bull & Stevens, 1979), (b) ratings of targets' effectiveness in a specific
occupational role (eg., counselor; Cash & Kehr, 1978), and (c) ratings of
targets' behavior (e.g., audibility of voice; Jackman, 1979). These eliminations reflect our interest in the personality stereotype rather than
perceptions of people's products or role enactments. For studies reporting more than one dependent variable, the meta-analysis included only
those measures that met these criteria for allowable dependent measures (eg., behavioral measures were excluded).
.
4. Consistent with our focus on the physical attractiveness stereotype, we included studies only if, following the precedent of Dion and
co-workers' (1972) seminal experiment, subjects judged personality
traits and life outcomes of target persons with whom they had no prior
role relationship. Because the review is not concerned with the effects
of specialized role relationships on the impact of physical attractiveness, we excluded studies in which subjects and targets had an ongoing
role relationship, for example, studies of students' ratings of their
teachers (e.g., O'Reilly, 1987) or of therapists' ratings of their patients
(eg., Nordholm, 1980). We also excluded studies in which subjects roleplayed a relationship with one or more targets, for example, studies in
which subjects were instructed to rate a counselor from the perspective
of a client (e.g., Cash & Kehr, 1978). In addition, studies were deleted if
the target persons were described as paired by means of a role relationship (eg., Hartnett & Secord's, 1983, targets were described as married
couples).
5. Several other criteria excluded studies examining specialized
types of target persons. Although studies presenting targets in specific
job categories (eg., counselor) were included as long as the subjects
were not in an actual or hypothetical role relationship with these targets, studies were eliminated if the targets were mentally ill (eg., mental hospital patients in the Martin, Friedmeyer, & Moore, 1977, study),
victimized (eg., rape victims in the Jacobson & Popovich, 1983, study),
or specialized in certain other ways (eg., described as having had premarital sexual intercourse in the Hocking, Walker, & Fink, 1982,
study). Studies with targets of these types were excluded because physical attractiveness would probably have specialized implications in
these contexts (e.g, it would not imply good social skills in mentally ill
targets; it might have atypical moral and sexual implications for rape
victims or unmarried persons described as sexually active).
6. To confine our review to the presumably fully developed stereotype among late adolescents and adults, studies were included only if
the subjects (and targets) were over the age of 14 years. The developmental aspects of physical attractiveness stereotyping could well be the
object of a separate meta-analysis.
7. Finally, numerous studies were excluded because they lacked sufficient information for the computation of at least one effect size (eg.,
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
R. Anderson &Nida, 1978). Such reports either lacked statistical detail
or presented statistical information in a form that does not allow the
computation of effect sizes (e.g., a multivariate analysis of variance
[MANOVA]; a regression with physical attractiveness included along
with other predictors).
In summary, our selection criteria narrowed our sample to studies
that examined the effects that targets' general physical attractiveness
has on perceivers' inferences about evaluative personal attributes in
settings that eliminate specialized roles and personal relationships
linking perceivers and targets. The resulting sample (see Appendix)
consisted of 69 documents reporting 76 studies.3
Variables Coded From Each Study
The following information was coded from each report: (a) date of
publication; (b) number of observations;' (c) number of judgments aggregated into each observation underlying the study-level effect sizes;5
(d) publication form (journal article, dissertation or unpublished document); (e) source of subject population (college undergraduates, other);
(f) nationality of subject population (American, Canadian); (g) sex of
subjects (male, female, male and female, unknown); (h) instructional
set (accuracy, i.e., of subject's judgment in relation to a presumed criterion [see Dion et al., 1972], impression formation, other, none or unclear); (i) type of attractiveness variation (attractive and unattractive
targets selected, changed appearance of same target, other);' (j) evidence that attractiveness varied as intended (preselection judgments,
postexperimental ratings, both, none); (k) control for confounds between attractiveness and other attributes of targets (multiple targets at
each attractiveness level, changed appearance of same target, other
controls, no such control or unknown); (1) within versus between variation of attractiveness (within, between); (m) stimulus modality (photograph, videotape, live, written description, other or more than one); (n)
part of target's body viewed (face, head and shoulders, full body, unclear, mixed or irrelevant); (o) stimuli in color (yes, no, unknown,
mixed or irrelevant); (p) source of stimuli (college yearbook, researcher
took own photographs or videotapes, other, unknown, irrelevant or
more than one); (q) basis of stimuli selection (means and variances of
judges' ratings, means of judges' ratings, consensual meaning of
words, unknown or intuitive, mixed or irrelevant); (r) sex of targets
(male, female, male and female); (s) race of targets (whites only, other
exclusion, not restricted, unknown, mixed or irrelevant); (t) restrictions on target's appearance (eyeglasses excluded, other exclusion,
more than one exclusion, no or unknown exclusion or appearance manipulated, mixed or irrelevant); (u) presence of individuating information along with attractiveness information (present, absent); (v) independent variables other than physical attractiveness in design, regardless of whether they provided individuating information (yes, no); and
(w) unusual features of statistical analysis (yes, no).
The categories we listed for some of these variables were simplified
from an initially more detailed set because (a) some of these more
detailed categories had few or no entries and (b) prediction of the effect
sizes was not improved by the more detailed categories. The variables
for which the initial coding was more detailed are the following: publication form, instructional set, type of attractiveness variation, control
for confounds, and presence of individuating information.
These variables were independently coded by Alice H. Eagly and
Mona G. Makhijani, with a median agreement of 95% (estimated *
[kappa] = .90); "other independent variables in design" yielded the
lowest agreement: 85% If = .39). Disagreements were resolved by dis-
Computation and Analysis of Effect
Sizes
The effect size calculated is g, the difference between the means of
two groups, divided by the relevant standard deviation.7 The sign of the
115
difference was positive when attractive targets were rated more favorably than unattractive targets and negative when unattractive targets
were rated more favorably. The computation of g was based on (a)
means and standard deviations or error terms for 446 effect sizes, (b) Fs
and K for 110 effect sizes, (c) correlations or chi-squares for 26, and (d)
level p values (e.g, p < .05) for 14. Thus, 596 effect sizes were computed.
To reduce computational error, these effect sizes were calculated
independently by Alice H. Eagly and Mona G. Makhijani. When the
data for a study were reported separately for male and female targets,
effect sizes were calculated separately for each sex. These additional
reports, separated by sex of target, supplemented the reports combined
over sex of target, which were used in most analyses.
The 596 effect sizes that we computed reported data on 104 different
individual rating scales as well as on 100 different composite measures
that authors of the studies had computed by aggregating two or more
ratings. To facilitate the analysis of these extremely diverse measures,
we aggregated them in various ways before analyzing them. To provide
a single measure for each study, thus creating effect sizes that satisfy the
independence assumption of meta-analytic statistics (Hedges & Olkin,
1985), we computed a single effect size for each study, combining all
separately calculated effect sizes. This study-level effect size was calculated using Rosenthal and Rubin's (1986) suggested formula and assum-
3
A list of the studies excluded by these criteria is available from
either Alice H. Eagly or Richard D. Ashmore.
4
The number of observations is the n taken into account by the
statistical analysis for the effects of physical attractiveness.
5
Study-level effect sizes were aggregated across all of the dependent
variables (see Computation and Analysis of Effect Sizes). For example,
each observation might represent a subject's ratings of 4 target persons
on 10 items, for a total of 40 judgments. To the extent that measures
were based on multiple judgments, they should yield more reliable
estimates of the physical attractiveness stereotype, in the manner that
the number of items in a test relates to the reliability of the total test
(e.g, Ghiselli, 1964).
6
When a study's design included more than two levels of attractiveness, we compared only the highest and lowest levels, if the report was
sufficient to yield this comparison. This rule maximized comparability of effect sizes across the studies because the typical study presented
targets preselected to be extremely attractive and extremely unattractive. The following variations were coded as "other" but included as
separate categories in preliminary analyses: high versus medium (when
only these conditions were present in experiment), changed verbal description, median split on attractiveness, and attractiveness treated as a
continuous variable.
7
When physical attractiveness was a between-subjects variable, this
standard deviation was computed separately within the high- and lowattractiveness conditions and pooled. When physical attractiveness
was a within-subjects variable, the standard deviation was the standard
deviation of the differences between the paired observations. These
standard deviations were estimated, whenever possible, only from the
portion of each study's data entering into the effect size.
Some statisticians have recommended that, when estimating standard deviations from the mean square error of an ANOVA, meta-analysts reconstitute this error term by adding into the sum of squares
error all other sums of squares except that for the independent variable
of interest (see Glass, McGaw, & Smith, 1981; Hedges & Becker, 1986).
This strategy was not chosen for this meta-analysis because physical
attractiveness was crossed with manipulated variables that in some
experiments were quite powerful. Consequently, adding sums of
squares for manipulated variables to the sum of squares error would
have had differing impact on these error terms across the studies. In
addition, statistical information was typically not extensive enough to
allow us to reconstitute the error terms.
116
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
ing that the correlation between measures was .48. This correlation was
estimated by averaging the correlations between dependent variables
(a) reported in the studies or (b) derived from coefficient alphas
reported for multiple-kern dependent variables used in the metaanalysis.
Categorization of dependent variables. To enable us to test our hypothesis that physical attractiveness has more impact on some types of
evaluative beliefs than others, we also classified each study's variables
into the following six categories: (a) social competence, including interpersonal skills and traits (sociable, fun loving) and successful outcomes
of such skills (popular, likable); (b) intellectual competence, including
intellectual ability (intelligent, skillful), rational mental style (rational,
scientific), high task motivation (ambitious, hard working), and successful outcomes of task endeavors (receiving good grades, achieving
career success); (c) concern for others, including social sensitivity (sensitive, empathic), nurturance (compassionate, generous), and lack of egotism (modest, not egotistic); (d) integrity, including honesty (trustworthy, honest) and respect for norms (faithfulness to spouse); (e) adjustment, including good mental health (normal, well adjusted),
happiness (satisfied, happy), high self-esteem (confident, positive selfregard), and well-being (maturity, good health); and (f) potency including power (strong, self-assertive) and dominance (dominant, acting as
leader). A composite measure reported in a study was placed into one
of these categories only if 60% or more of its component ratings fit into
the category.
In addition, some measures were placed into one of two other categories. The first of these categories is general evaluation, which was chosen for single-item measures whose evaluative meaning was completely
nonspecific (e.g., good) and for composite measures whose component
scales fit two or more of our specific categories and failed to attain the
criterion of at least 60% fitting any one category. The second of these
additional categories is other, which was chosen for measures that neither fit into any of the six specific categories nor qualified as measures
of general evaluation.
The measures from a study were thus placed into one or more of
these eight categories. This classification was accomplished by the four
authors' independent categorizations of the measures; three of the four
authors placed 80% of the single-item dependent variables and 72% of
composite measures in the same category. Disagreements were resolved by discussion between Richard D. Ashmore and Laura C.
Longo. The 596 effect sizes calculated from the studies' reported findings were aggregated into category-level effect sizes for each study for as
many of the eight categories as the study represented. These aggregations were calculated under the assumption that the correlation between measures placed in the same category was .48. Each study thus
yielded, in addition to a study-level effect size, one or more of these
eight category-level effect sizes.
Analysis of effect sizes. The gs were converted to ds by correcting
them for bias (i.e., g's overestimate of the population effect size, which
occurs especially for small samples; see Hedges, 1981; Hedges & Olkin,
1985). Then the study outcomes were combined by averaging thet/s. To
determine whether each set of ds shared a common effect size (i.e., was
consistent across the studies), we calculated a homogeneity statistic, Q,
which has an approximate chi-square distribution with k — 1 degrees of
freedom, where k is the number of effect sizes (Hedges, 1981; Hedges &
Olkin, 1985).
In the absence of homogeneity, we accounted for variability in heterogeneous effect sizes by relating them to the attributes of the studies. To
determine the relation between these study characteristics and the
magnitude of the effect sizes, both categorical and continuous models
were tested (Hedges, 1982a, 1982b; Hedges & Olkin, 1985). Categorical
models, which are analogous to ANOVAs, may show that heterogeneous effect sizes are homogeneous within the subgroups established
by dividing studies into classes based on study characteristics. The
techniques for calculating categorical models provide a betweenclasses effect (analogous to a main effect in an ANOVA) and a test of the
homogeneity of the effect sizes within each class. The between-classes
effect is estimated by QB, which has an approximate chi-square distribution with p—1 degrees of freedom, where pis the number of classes.
The homogeneity of the effect sizes within each class is estimated by
<2w, which has an approximate chi-square distribution with m - \
degrees of freedom, where m is the number of effect sizes in the class.
The tables reporting tests of categorical models also include the mean
weighted effect size for each class, calculated with each effect size
weighted by the reciprocal of its variance, and an indication of whether
this mean differed significantly from the 0.00 value that indicates exactly no effect of physical attractiveness.
The continuous models are least squares simple linear regressions,
calculated with each effect size weighted by the reciprocal of its variance. Each such model yields a test of the significance of a predictor as
well as a test of model specification, which evaluates whether significant systematic variation remains unexplained in the regression model
(Hedges, 1982b; Hedges & Olkin, 1985). The sum of squares error statistic, QEr which provides this test of model specification, has an approximate chi-square distribution with k - p -1 degrees of freedom,
where k is the number of effect sizes and p is the number of predictors
(not including the intercept). If correctly specified models are not
achieved when implementing continuous models (or homogeneity is
not achieved within the classes when implementing categorical models), the results of these analyses cannot be interpreted as confidently as
they would otherwise be.
As a supplementary analysis, we attained homogeneity by identifying outliers among the effect sizes and sequentially removing those that
reduced the homogeneity statistic by the largest amount (see Hedges &
Olkin, 1985). Using such a procedure, Hedges (1987) found for several
meta-analyses on psychological topics that the removal of up to 20% of
the outliers in a group of heterogeneous effect sizes usually resulted in a
high degree of homogeneity. Inspection of the percentage of effect sizes
removed to attain homogeneity allows one to determine whether the
effect sizes are homogeneous aside from the presence of relatively few
aberrant values. The mean attained after removal of such outliers may
better represent the distribution of effect sizes than would the mean
based on all of the effect sizes.
Results
Characteristics of the Studies
Before considering the findings reported in research on the
physical attractiveness stereotype, we examined the characteristics of the studies from which conclusions about this research
will be drawn. Table 1 shows many of these study characteristics.
As shown by the central tendencies of the characteristics
listed in Table 1, studies (a) were published relatively recently, (b)
were published as journal articles, (c) involved a moderate number of observations, and (d) aggregated a moderate number of
judgments into each observation. The studies used subjects
who were college undergraduates, Americans, and both male
and female. In addition, the studies generally (a) established an
impression-formation instructional set; (b) compared perceptions of targets selected to be attractive and unattractive; (c)
presented both preselection and postexperimental evidence
that attractiveness varied as intended; (d) controlled confounds
by means of multiple targets at each attractiveness level, within
each level of target sex; (e) varied attractiveness on a betweensubjects basis; (f) presented targets by means of photographs or
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
117
Table 1
Summary of Study Characteristics
Variable and class
Median publication year
Publication form
Journal article
Dissertation or unpublished document
Median number of observations
Median number of judgments aggregated into each
observation
1980
53
23
12
70
6
Nationality of subjects
American
Canadian
70
6
60
8
5
I
2
Part of target's body viewed
Face
Head and shoulders
Full body
Unclear
Mixed or irrelevant
11
37
15
11
2
Yes
No
Unknown
Mixed or irrelevant
9
8
58
1
Instructional set
Accuracy
Impression formation
Other
None or unclear
19
38
9
10
Type of attractiveness variation
Attractive and unattractive targets selected
Changed appearance of same target
Other
57
10
9
Within versus between variation of attractiveness
Within
Between
Stimulus modality
Photograph or slide
Videotape
Live
Written description
Other or more than one
Stimuli in color
Sex of subjects
Male
Female
Male and female
Unknown
Control for confounds
Multiple targets
Changed appearance of same target
Other controls'
No such control or unknown
Value
108
Source of subject population
College undergraduates
Other
Evidence that attractiveness varied as intended
Preselection judgments
Postexperimental ratings
Both
None
Variable and class
Value
19
8
46
3
52
9
3
12
18
58
19
18
37
2
Source of stimuli
College yearbook
Researcher took own photographs or videotapes
Other
Unknown
Irrelevant or more than one
22
21
6
20
7
Basis of stimuli selection
Means and variances of judges' ratings
Means of judges' ratings
Consensual meaning of words
Unknown or intuitive
Mixed or irrelevant
35
28
1
11
1
Sex of targets
Male
Female
Male and female
8
30
38
Race of targets
Whites only
Other exclusion
Not restricted
Unknown
Mixed or irrelevant
21
1
1
51
2
Restrictions on target's appearance
Eyeglasses excluded
Other exclusion
More than one exclusion
No or unknown exclusion or appearance changed
Mixed or irrelevant
2
12
13
47
2
Presence of individuating information
Present
Absent
48
28
Note. For categorical variables, numbers in tables represent frequencies of reports in each class.
* Includes changed written description and use of both multiple targets and changed appearance in same study
slides; (g) used head-and-shoulders poses of targets; (h) did not
source of stimuli; (j) selected stimuli based on means and vari-
mention whether stimuli were presented in color or black and
ances of judges' ratings or on means only; (k) used female tar-
white; (i) used college yearbook photographs, or photographs or
gets only or both male and female targets; (1) did not report on
videotapes taken by the researcher, or did not mention the
the race of the targets; (m) did not report on or had no restric-
118
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
tions on targets' appearance or used targets whose appearance
women. This stereotype was inconsistent in magnitude across
was changed; and (n) presented individuating information
the studies and somewhat smaller once outlying effect sizes
were deleted.
along with the attractiveness information.
Summary of Study-Level Effect
Sizes
The summary of the study-level effect sizes given in Table 2
Physical Attractiveness Stereotype and Type of Attributes
Ascribed to Targets
allows one to determine whether more favorable characteristics
were ascribed to attractive individuals than to unattractive indi-
The categorical model in Table 3 tests our prediction that the
strength of the physical attractiveness stereotype depends on
viduals based on the available reports. An overall difference in
the particular dimension of meaning that was assessed. This
the stereotypic direction is shown by a positive mean effect size
that differed significantly from the 0.00 value that indicates
model produced a highly significant between-classes effect. As
exactly no difference (i*, by a confidence interval that did not
include 0.00). As expected, evaluations were more favorable for
of social competence and the smallest effect sizes for concern
attractive targets than for unattractive targets, as shown by the
dimensions of meaning (intellectual competence, adjustment,
potency) produced effect sizes that were in between these
weighted mean of the effect sizes8 as well as by the unweighted
mean and the median. In addition, a large proportion (.91) of
these study-level effect sizes favored the attractive targets over
expected, the largest mean effect size was produced for ratings
for others and integrity. Ratings on the remaining three specific
classes of effect sizes. Consistent with this pattern, the general
evaluation category produced effect sizes in the middle range of
the unattractive targets. This proportion differed significantly
from .50, the proportion expected under the null hypothesis
the means for the six specific types of meaning, as did the
(/K.001).
set of ratings that did not fit into the six specific categories that
As also shown in Table 2, the effect sizes were not homogeneous, and homogeneity was not attained until a relatively large
we established. We evaluated the significance of the differences
between these classes of effect sizes by means of post hoc com-
proportion of the effect sizes C33) was removed. The removal of
outliers produced a somewhat smaller weighted mean that also
parisons, which are represented by the subscripts attached to
the mean effect sizes given in Table 3 (see Hedges & Becker,
1986; Hedges & Olkin, 1985).'
differed significantly from 0.00.
The various indexes of the central tendency of the effect sizes
thus substantiate the existence of an attractiveness stereotype,
as documented by the ascription of more favorable characteristics to attractive men and women than to unattractive men and
category labeled other, which contained an extremely diverse
Relations Between Other Study Attributes and the
Physical Attractiveness Stereotype
The relations between other study attributes and the physical
attractiveness stereotype were evaluated on the study-level effect sizes. The categorical models that produced significant be-
Table 2
Summary of Study-Level Effect
tween-classes effects are given in Table 4. The first model indicates a significant between-classes effect for publication form;
Sizes
Criterion
Value
journal articles produced a stronger stereotype than dissertations and unpublished articles. The second model shows a significant effect for instructional set; the accuracy set yielded a
76
0.58
0.54/0.62
450.62*
0.61
0.50/0.72
0.57
69/76 (.91)
weaker stereotype. Given Dion's (1981,1986) logic that the accuracy set would reduce stereotyping and encourage careful think-
Including outliers
«
Mean weighted effect size (dj95%CIforrf +
Homogeneity (Q) of effect sizes comprising
Mean effect size (mean d)
95% CI for mean d
Median effect size
Differences favoring attractive targets"
rf+"
sons between the accuracy class of effect sizes and the three
other classes seem appropriate. By such comparisons, the
stereotype was significantly weaker with the accuracy set than
each of the other three sets (ps < .05).
Although the physical attractiveness stereotype related to the
Excluding outliers
n
n removed outliers
Mean weighted effect size (d+)
95% CI for d+
Homogeneity (Q) of effect sizes comprising rf+
ing about targets' personal characteristics, a priori compari-
type of method researchers used to vary attractiveness, post
51
25 (.33)
0.49
0.44/0.54
69.11
hoc contrasts show only that the quite diverse category of other
manipulations produced a stronger stereotype than the more
typical manipulation that compared stimuli preselected to be
high or low in attractiveness ( p < .001). The type of control that
researchers established for confounds of the physical attractive-
Note. Effect sizes are positive for more favorable evaluation of attractive than unattractive targets and negative for more favorable evaluation of unattractive targets. CI = confidence interval.
" Effect sizes were weighted by the reciprocal of the variance. b Significance indicates rejection of the hypothesis of homogeneity. ° Frequencies are the number of study-level effect sizes favoring the more
attractive targets divided by the total number of effect sizes. The proportion appears in parentheses.
* The weighted means were computed by weighting each effect size
by the reciprocal of its variance (Hedges & Olkin, 1985), a procedure
that gives more weight to effect sizes that are more reliably estimated.
9
Although our dimensions differed somewhat from those used by
Feingold (1990b), the ordering of the dimensions was substantially similar.
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
119
Table3
Categorical Model for Type of Attributes Ascribed to Targets
95% CI for <k
Attribute type
Between-classes
effect (G>)
n
Mean weighted
efiect size (d,+)
Lower
Upper
Homogeneity within
each class (G^
35
18
14
38
11
22
39
13
0.68.
0.52.J
0.49^
0.46t
0.13,
0.01,
0.57^
0.54^
0.62
0.45
0.39
0.41
0.03
-0.06
0.52
0.44
0.73
0.60
0.58
0.51
0.23
0.09
0.62
0.64
213.60*
239.59*
36.67*
178.48*
80.84*
284.65*
414.59*
154.00*
279.97*
Social competence
Adjustment
Potency
Intellectual competence
Integrity
Concern for others
General evaluation
Other
Note. Effect sizes are positive for more favorable evaluation of attractive than unattractive targets and negative for more favorable evaluation of
unattractive targets. Differences between means that do not have a subscript in common are significant at the .05 level or beyond. CI = confidence
interval.
* Significance indicates rejection of the hypothesis of homogeneity.
*p<.001.
ness variable also produced a significant between-classes effect.
Yet, using post hoc contrasts, the only significant comparison
The number of judgments aggregated into each observation
was the only continuous variable that produced a significant
among the four classes showed that the stereotype was stronger
model. Thus, the physical attractiveness stereotype was
stronger when a greater number of judgments (i£., items) were
when no control was used or when the nature of any controls
was unknown than it was when the typical multiple-target technique was used (p < .05). In addition, consistent with the
greater control of extraneous variables and the more precise
error term in within-subjects designs, within-subjects variations of attractiveness produced a stronger stereotype than between-subjects variations.
As also shown in Table 4, several attributes of the target persons affected the strength of the physical attractiveness stereotype. The use of color (vs. black and white) stimuli related signifi-
aggregated into each of the observations underlying the studylevel effect sizes (b = 0.0017, tf = .11, p < .025; QE[74] =
445.38, p < .001). This analysis raises the question of whether
the various dimensions of evaluative meaning shown in Table 3
differed with respect to the number of judgments aggregated
into each observation. Although some differences were observed, they showed no relation to the strength of the evaluative
stereotype (Le, to the means shown in Table 3).
cantly to the effect sizes. Post hoc comparisons showed that
smaller efiect sizes were produced by stimuli that were known
to be black and white than by stimuli that were in color (p <
.005) or for which color was unknown (p < .001). The next
model given in Table 4 failed to confirm our hypothesis that the
physical attractiveness stereotype is stronger for female targets
Discussion
Strength of the Physical Attractiveness Stereotype
Although the consensus among social psychologists has been
than for male targets. This hypothesis was tested on the 98
that the beauty-is-good stereotype is extremely strong and robust, our implicit personality theory framework led us to ex-
effect sizes produced when studies reporting separate data for
pect that the tendency for people to ascribe more favorable
male and female targets were represented by these separate
effect sizes and the other studies were represented by study-level
personal attributes to attractive targets than to unattractive tar-
effect sizes. Although the variable of target sex produced a sig-
across studies. Relevant to the overall magnitude of the physical
attractiveness stereotype are the several measures of efiect size
nificant categorical model, the attractiveness stereotype was
stronger for reports combined over male and female targets (i.e,
reports that could not be partitioned by sex of target) than for
reports on either male targets (p < .01) or female targets (p <
gets would be somewhat modest in size and quite variable
central tendency reported in Table 2. These measures varied
from a high of 0.61 for the unweighted mean to a low of 0.49 for
.001). Contrary to hypothesis, the stereotype did not differ for
the weighted mean with outliers excluded. These figures indicate approximately one-half standard deviation in the direction
male and female targets. Moreover, we calculated the sex-oftarget model separately within each of the dimensions shown in
of more positive personality traits and life outcomes ascribed to
attractive compared with unattractive targets.
Table 3 and obtained no evidence of a stronger stereotype for
Methodological features, both of our meta-analysis and of
the research paradigms of the studies we reviewed, should be
female than male targets within any of these dimensions. Also,
sex of subjects did not produce a significant model. Finally,
consistent with our hypothesis that individuating information
taken into account in interpreting the magnitude of the effect
sizes we obtained. Among the important features of our meta-
would attenuate inferences made on the basis of physical attractiveness, the stereotype was weaker when such information was
analytic methods are the rules we followed for including studies
present rather than absent.
if we could compute at least one effect size, and among the
in the present sample. In particular, studies were included only
120
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
Table 4
Categorical Models on Study-Level Effect
Sizes
95% CI for rf,+
Variable and class
Publication form
Journal article
Dissertation or unpublished document
Instructional set
Accuracy
Impression formation
Other
None or unclear
Type of attractiveness variation
Attractive and unattractive targets selected
Between-classes
effect (Q,)
Upper
Homogeneity within
each class (Gw,)'
53
23
0.64
0.45
0.60
0.39
0.68
0.51
258.01***
168.64***
19
38
9
10
0.50
0.60
0.62
0.64
0.43
0.54
0.47
0.57
0.56
0.65
0.77
0.71
156.86***
227.00***
21.90**
35.36***
57
10
9
0.54
0.66
0.73
0.50
0.52
0.64
0.58
0.81
0.82
60.30***
23.14**
51.16***
52
9
3
12
0.55
0.66
0.54
0.69
0.51
0.51
0.35
0.61
0.59
0.81
0.73
0.78
382.92***
23.14**
4.36
30.00**
18
58
0.66
0.53
0.60
0.48
0.72
0.57
126.87***
310.72***
19
18
37
0.63
0.42
0.63
0.55
0.35
0.58
0.72
0.50
0.68
112.12***
137.89***
172.99***
30
52
16
0.57
0.50
0.72
0.51
0.45
0.65
0.64
0.56
0.79
154.49***
280.24***
52.80***
48
28
0.46
0.68
0.41
0.63
0.51
0.73
172.96***
241.03***
9.20*
16.03***
10.18**
Multiple targets
Changed appearance of same target
Other controls
No control or unknown
Within versus between variation of attractiveness
Within
Between
13.03***
Stimuli in color"
Yes
No
Unknown
23.15***
Sex of targets
24.23***
Male
Female
Both
Presence of individuating information
Present
Absent
Lower
23.98***
Changed appearance of same target
Other
Control for confounds
«
Mean weighted
effect size (dH)
36.63***
Note. Effect sizes are positive for more favorable evaluation of attractive than unattractive targets and negative for more favorable evaluation of
unattractive targets. Cl = confidence interval.
"Significance indicates rejection of the hypothesis of homogeneity. b The category "mixed or irrelevant*1 (n = 2) was omitted from this analysis.
*p<.05. **p<.01. ***p<.001.
included studies, only effect sizes that we could compute entered into our aggregations. Although some meta-analysts have
assigned the value of 0.00 to unknown effect sizes, we were
reluctant to compromise the accuracy of our data in this way,
especially in view of the fact that our computed effect sizes were
then aggregated into study-level effect sizes. Nonetheless, our
procedure probably increased the magnitude of our average
effect sizes because of the tendency for minimally reported findings to be nonsignificant and generally relatively small.
An additional consideration in evaluating meta-analytic
methods is the extent to which the magnitude of the physical
attractiveness stereotype may have been exaggerated in our review because of a publication bias in favor of significant findings (Greenwald, 1975; Lane & Dunlap, 1978). Indeed, in our
data the effect sizes were larger in published than unpublished
research (see Table 4), a typical result in meta-analyses (see
Glass et al., 1981). Although publication bias must therefore be
considered an influence, our inclusion of all available unpublished findings diluted the impact of this bias on our aggregated
effect sizes. Moreover, the mean effect size in the unpublished
studies remained substantial (see Table 4).
One of the methodological features of physical attractiveness
studies relevant to interpreting the magnitude of our effect sizes
is the degree to which extraneous variables were controlled in
typical experiments. Such control reduces the standard deviations that serve as the denominator of the effect sizes. Compared with many other areas of research in social and personality psychology (e.g., prosocial and antisocial behavior, group
process), research on the physical attractiveness stereotype is
relatively controlled, because carefully selected target stimuli
are presented to subjects in a laboratory setting that usually
precludes social interaction between the subjects and targets. In
addition, because dependent variables that aggregate subjects'
responses ordinarily strengthen effects (Abelson, 1985; Ajzen,
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
121
1988), the fact that the observations analyzed in the studies we
20% or less of effect sizes that Hedges (1987) removed to attain
reviewed typically represented a moderate number of judgments should also be taken into account (see Table 1). Thus, two
homogeneity among studies on several diverse psychological
important methodological features of this research literature—
psychological topics have achieved homogeneity by removal of
the relatively high level of experimental control and the moderate amount of aggregation of responses—favored the produc-
smaller proportions of effect sizes than the 33% removed in the
tion of large effect sizes.
1989). Thus, the effect sizes produced by studies on the physical
Several additional considerations are relevant to evaluating
the magnitude of the average ds produced by our meta-analysis
attractiveness stereotype should be viewed as quite inconsistent
hypotheses. Furthermore, other recent meta-analyses on social
present review (see Eagly & Johnson, 1990; Johnson & Eagly,
in magnitude.
(see Eagly, 1987). One set of guidelines for the d metric has been
suggested by Cohen (1977), who proposed that 0.20 can be
described as small, 0.50 as medium, and 0.80 as large. Cohen
based these standards on his impressions of effect sizes in the
behavioral sciences and maintained that phenomena represented by medium effect sizes would ordinarily be noticed in
daily life. Cohen's characterization of medium effect sizes as
Impact of Type of Attributes on the Physical
Attractiveness Stereotype
Congruent with our hypotheses, which built on the excep-
discriminable is consistent with the transformation of our
tions and qualifications noted in some narrative reviews (e.g.,
mean weighted effect size of 0.58 into the metric of Rosenthal
Adams, 1982; Alley & Hildebrandt, 1988; Bassili, 1981; Bull &
and Rubin's (1982) binomial effect-size display. By this transformation, 64% of attractive people but only 36% of unattractive
Rumsey, 1988; Cash & Janda, 1984; Dermer & Thiel, 1975;
people were perceived as having above-average personalities.
This difference of 28% suggests a substantial tendency, despite
strength of the beauty-is-good effect from study to study and
the fact that a d of 0.5 8 can be interpreted as signifying that only
this variation was explained by the type of dependent variable
8% of the variability in subjects' inferences about targets' personalities was explained by their physical attractiveness.
researchers had used. The fact that physical attractiveness had
Another basis for interpreting the average effect sizes for the
physical attractiveness stereotype is to compare them with aver-
Dion, 1981, 1986), there was considerable variation in the
from measure to measure within studies. As expected, some of
its strongest impact on social competence supports our contention that the core of the physical attractiveness stereotype is
sociability, popularity, and similar attributes. Other evaluative
age effect sizes typically produced by meta-analyses on other
hypotheses in social and personality psychology. Eagly's (1987)
dimensions varied considerably in how closely they were psychologically connected to good looks versus bad looks. Physical
overview of such effect sizes suggested that they ranged from
roughly 0.00 to 1.20. For example, mean </s for sex differences
in social behavior ranged from a low of 0.13 for helping behav-
attractiveness had little impact on integrity and concern for
others; potency, adjustment, and intellectual
competence
showed intermediate impact.
ior to a high of 1.19 for the use of pause fillers in vocal behavior.
Another such comparative standard is provided by Cooper and
tive people perceived less favorably than unattractive people,
Findley's (1982) computation of a mean d of 0.92 for the find-
although attractive people possessed no advantage in our con-
ings cited in a sample of social psychology textbooks.10 Because
studies cited in textbooks are likely to provide the best demon-
cern for others class. Because the argument that the attractive-
strations of those phenomena that social psychologists believe
are genuine, this figure no doubt considerably overestimates the
strength of typical findings for these phenomena.
Dermer & Thiel, 1975) suggests that attractive people are perceived to be more likely than unattractive people to possess the
In summary, the magnitude of the physical attractiveness
stereotype in our meta-analytic findings was enhanced by (a)
our exclusion of studies for which we could not calculate an
effect size, (b) an apparent publication bias in favor of positive
findings, and (c) experimental designs featuring a high degree of
experimental control and moderate aggregation of judgments.
In view of these considerations as well as the numeric guidelines commonly used for evaluating the magnitude of effect
sizes, the beauty-is-good effect should be considered moderate
in magnitude but certainly not extremely strong.
Finally, in relation to the robustness of the beauty-is-good
stereotype, it is worth commenting on the consistency of the
findings across the studies. The direction of the effect proved
quite consistent, as shown by the fact that 91% of the study-level
effect sizes favored the attractive targets (see Table 2). Nonetheless, as shown by the large value of the Q statistic (see Table 2),
the findings were quite inconsistent in magnitude. Moreover, to
achieve a homogeneous set of findings, 33% of the study-level
effect sizes had to be removed. This figure compares with the
In none of our categories of evaluative meaning were attrac-
ness stereotype has a dark side (e.g. Cash & Janda, 1984;
negative attributes of vanity and selfishness, we performed separate analyses to determine whether ratings of modesty versus
vanity (« = 6) or egotism versus selflessness (n = 6)—two components of our concern for others class—would show significant
reversals of the usual tendency to ascribe more favorable qualities to beautiful people. Indeed, both means differed significantly from 0.00 in the predicted negative direction: d+ — —0.67,
CI = -0.51/-0.84 for modesty versus vanity; d< = -0.25, CI =
—0.08/-0.41 for egotism versus selflessness. However, the
mean for egotism became positive when one highly negative
outlying effect size was removed. Thus, we conclude only that
attractive people are perceived more negatively than unattractive people in one respect: They are thought to be more vain
and less modest. The present results, then, serve to clarify the
dark side of the physical attractiveness stereotype. It includes
10
This mean effect size aggregates comparisons involving (tests, F
tests, and correlation coefficients.
122
EAGLY, ASHMORE, MAKHUANI, AND LONGO
vanity but perhaps not the apparently similar attribute of egotism."
The heterogeneity of the attractiveness stereotype across dimensions of evaluative meaning has important implications for
interpreting research in this area. These meta-analytic findings
raise questions about the common practice of aggregating subjects' responses into composite indexes. To take the Social Desirability Index used by Dion and her collaborators (Dion et al,
1972) as an example, we placed 7 of the 14 items in this composite in our social competence category, 3 in potency, 3 in concern
for others, and 1 in other. Because this composite emphasized
social competence, its distribution of items across the dimensions of personality resulted in a relatively strong stereotype, a
much stronger stereotype than would have been obtained by
choosing, for example, a large number of items representing
integrity and concern for others. Our meta-analytic results thus
suggest that researchers should select items systematically by
designing indexes that represent the content-specific dimensions of subjects' implicit theories of personality. Aggregate
measures consisting of items that were selected through more
arbitrary methods produce unreliable estimates of the strength
of the physical attractiveness stereotype and obscure the subtleties of people's stereotypic thinking about looks.
Impact of Other Variables on the Physical
Attractiveness Stereotype
Individuating information. The implicit personality theory
framework guiding the present analysis also suggested that the
strength of the physical attractiveness stereotype would be affected by information about the targets in addition to their
physical attractiveness. As predicted, the beauty-is-good effect
was weaker when subjects received individuating information
along with attractiveness cues. This finding, which is consistent
with the prior research we noted early in this article, suggests
that the evaluative implications of looks may be somewhat weak
in many natural settings in which perceivers have extensive information about other people in addition to their physical appearance. Looks should thus be relatively less important in perceptions of friends, acquaintances, family members, and coworkers than in perceptions of strangers.
Sex of targets. The prediction that the attractiveness stereotype is stronger for female targets than for male targets was not
supported.12 In seeking to explain why, we offer two quite different lines of reasoning. One possibility is that this hypothesis
may be valid under circumstances not present in the research
paradigms represented in this meta-analysis. In particular, the
greater weight of attractiveness in perceivers' judgments of
women may occur in the context of face-to-face interaction or,
more generaUy, in situations in which perceivers are likely to
interact with targets. If women take more interest in and responsibility for the interpersonal aspects of daily experience (see
Eagly, 1987; Eagly & Wood, in press), physical attractiveness
may have more impact on judgments of women than men in
natural settings where people engage in social interaction, and
social competence therefore has important consequences. The
greater importance of attractiveness to perceptions of women
may not be revealed in the typical stereotype study, in which
subjects merely rate photographs of targets whom they do not
expect to meet.
A second, more radical interpretation is that our hypothesis
about the greater weight of attractiveness in judgments of
women is incorrect. If so, how can this conclusion be reconciled
with research suggesting that physical attractiveness is more
important to the female gender role than to the male gender
role and to women's self-esteem and personal identity (see summary presented early in this article)? To resolve this apparent
paradox, we suggest framing the effects of physical attractiveness in terms of two stages: (a) In a first stage, perceivers infer
from appearance cues that targets have certain personality
characteristics; and (b) in a second stage, the implications of
these inferences unfold during social interaction. In the first
stage, physical attractiveness may cause perceivers to draw the
same conclusions about women's and menls personalities (iff., it
conveys the same information about social competence and
other attributes). However, in the second stage, conclusions
based on attractiveness may be more consequential for women
than men. Specifically, as social interaction proceeds, the gains
in perceived social competence enjoyed by good-looking people may have more important consequences for women because the structuring of interaction by means of gender roles
assigns women more responsibility for the maintenance of positive and friendly relationships (see Eagly, 1987). Therefore,
women may work harder to be perceived as socially competent,
accounting for women's greater interest in their personal appearance and the greater centrality of looks in their selfconcepts. Also if marriage and family relationships are more
central to women than men (see Wood, Rhodes, & Whelan,
1989), the implications of looks for heterosexual attraction may
be more important to women than men.
Within- versus between-subjects manipulation of attractiveness. The finding that the attractiveness stereotype was
stronger for within-subjects manipulations than for betweensubjects manipulations of appearance is amenable to two possible explanations. One interpretation is that the within-subjects
procedure provides a more precise error term because subjects
served as their own controls (see also footnote 7). A second
interpretation is that the context in which stimuli are perceived
in within-subjects experiments—namelji exposure to targets of
widely differing levels of physical attractiveness—may induce a
perceptual contrast effect whereby attractive targets are seen as
more attractive and unattractive targets as less attractive than
they would otherwise be. Such contrast effects are well documented in research on judgments of physical attractiveness (see
Patzer, 1985, for a summary). The more extreme judgments of
attractiveness that result from such a contrast effect could induce more extreme judgments of personality.
Color versus black-and-white stimuli. Another finding of our
" Our conclusion about egotism is somewhat tentative because five
of the six egotism effect sizes were based on composite measures of
egotism that included some attributes in addition to egotism and its
synonyms.
12
Yet Feingold (1990b) showed that physical attractiveness led to
stronger inferences of sexual warmth for female targets than for male
targets.
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
meta-analysis is that the beauty-is-good effect was more pronounced for color stimuli than for black-and-white stimuli. Interpretation of this finding must remain tentative because, as a
result of the correlational nature of meta-analytic findings, the
color versus black-and-white aspect of the stimuli was confounded to some extent with other features of the experiments
(e.g., the studies with color stimuli encompassed all of those
with face-to-face interaction). Nonetheless, if we assume that
color stimuli are more vivid than black-and-white photographs,
this finding can be interpreted in terms of the greater psychological impact of vivid (vs. pallid) stimuli. Vivid information is
presumed to attract and hold people's attention because it is, for
example, concrete and imagery provoking, emotionally interesting, or proximal in a sensory, temporal, or spatial way (Nisbett
& Ross, 1980). Although empirical support for the hypothesis
that vividly presented information has more impact on judgments than pallid information has been mixed, many of the
failures to confirm vividness hypotheses may have been due to
researchers' failures to specify exactly what was made vivid by
their particular vividness manipulations (see Chaiken & Eagly,
1983; Taylor & Thompson, 1982). This ambiguity was for the
most part not present in this research because photographs
were typically presented, making the physical appearance of
the target vivid. Thus, the typical experimental paradigm for
research on the physical attractiveness stereotype may be especially well adapted to demonstrating vividness effects, and primary research within this paradigm might well explore these
effects.
Number of judgments aggregated into dependent variable.
The finding that the beauty-is-good effect was stronger when a
larger number of judgments was aggregated into the dependent
variable is explicable in terms of ordinary principles of reliability and aggregation: Dependent variables show stronger relations to appropriate predictors to the extent that measures of
these variables are more reliable because they are based on
aggregations of correlated responses (e.g., Ajzen, 1988). This
relation between the strength of findings and response aggregation has been obtained in other meta-analyses (Eagly & Johnson, 1990; Eagly & Steffen, 1986), and it is appropriate to routinely track the effects of aggregation in quantitative reviews.
Other findings. The relation that the physical attractiveness
stereotype showed to instructional set, type of attractiveness
variation, and control for confounds are of somewhat less interest because it is difficult to interpret these findings in any but
the most tentative way. \fet the tendency for the accuracy instructional set to produce a weaker stereotype than other instructional sets is interesting and consistent with Dion and colleagues' (1972) logic in choosing this particular set (see Dion,
1981,1986). It is also worth noting the tendency for the stereotype to be more pronounced in studies that apparently did not
(vs. did) implement any control for confounds of attractiveness
with other features of target stimuli. Because such studies presented only two individuals, one attractive person and one unattractive person (of each sex if both sexes of targets were included), attractiveness could have been confounded with other
aspects of targets' physical appearance. The fact that this lack of
control was associated with a stronger beauty-is-good effect suggests that confounds favored researchers' hypothesis.
123
Limitations of This Meta-Analysis
Our integration of research on the physical attractiveness stereotype addressed only a limited number of questions, and future meta-analyses of this research literature could well consider several other important issues. In particular, the beauty-isgood hypothesis might be compared with an ugly-is-bad
hypothesis (Hatfield & Sprecher, 1986; Patzer, 1985). Such a
meta-analysis would compare the differences in judgments of
high- and medium-attractiveness targets with the differences in
judgments of medium- and low-attractiveness targets. We did
not compute effect sizes comparing medium-attractiveness targets with high- or low-attractiveness targets because the design
comparing high- and low-attractiveness targets is the prototype
for research on this topic. In fact, 71% of the studies in our
sample did not include a middle or medium level of attractiveness.
Another limitation of our meta-analysis is that we had no
satisfactory way to gauge the strength of researchers' manipulations of physical attractiveness. Some of the between-study variability in findings was no doubt due to variation in the strength
of the attractiveness manipulation. Most researchers used
judges' ratings to select their target stimuli, but rules for selecting good- and bad-looking stimuli varied (e.g, some investigators excluded targets that were rated as extremely attractive or
unattractive, whereas others did not; see Wallston & O*Leary's,
1981, discussion of this point). Although manipulation checks
were often present, data were generally not reported in sufficient detail to allow the computation of effect sizes representing
the perceived difference in the physical attractiveness of the
high- and low-attractiveness targets. Moreover, norms of attractiveness probably changed over time so that the features of
highly attractive faces varied somewhat from study to study.
These variations in looks may have influenced the particular
inferences that perceivers made.
The conclusions of our meta-analysis are also limited by the
characteristics of the existing paradigm for assessing the attractiveness stereotype. Many researchers from Dion and coworkers (1972) onward have used the same basic laboratory
experimental procedure. In the interest of experimental control, Dion and associates' paradigm presented only one type of
stimulus in a particular way: namely, college students depicted
in head-and-shoulders photographs. In this study and in most
subsequent research, the targets were young adults with the
physical appearance idiosyncracies of people who attend college. This stimulus domain reduced the likelihood that goodlooking but bad types (e.g, dumb blonde, whore/tramp, femme
fatale, sleazy guy, gigolo) were represented. For example, the
prototypic femme fatale is generally depicted in the media as
older than the modal college woman and identified by clothes,
makeup, and jewelry that are unlikely to be worn by college
women posing for yearbook photos.
The most important limitation of the present meta-analysis
follows directly from our goal of evaluating the strength and
generality of the physical attractiveness stereotype. Because of
our decision to study only the stereotype itself, we included only
studies that assessed evaluative perceptions of targets in situations that did not involve personal or role relationships between
perceiver and target. Although this emphasis allowed us to do
124
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
the best possible job of documenting the overall size of the
attractiveness stereotype and its variability, especially as a function of the specific type of inference, it did not permit us to
probe the impact that the beauty-is-good stereotype has on action. We have thus not examined the impact of attractiveness
stereotyping in ongoing social interaction and everyday role
relationships. Quantitative summaries of the large research literature addressing such issues can now proceed given the relatively clear description that the present review provides of the
physical attractiveness stereotype itself.
Conclusion
This meta-anatysis indicates that narrative reviewers were
correct in concluding that, in terms of perceivers' inferences
from cues that convey physical attractiveness, what is beautiful
is good. At the same time, our findings reveal that this effect is
highly variable, and that the implicit personality theory framework successfully identifies factors that help account for this
variation. Thus, our results indicate that the beauty-is-good effect depends crucially on the type of inference the perceiver is
asked to make. Good looks induce strong inferences about social competence and weaker inferences about potency, adjustment, and intellectual competence, but have little impact on
beliefs about integrity and concern for others. One important
direction for future primary research is to design explicit tests
of the effects we demonstrated of the type of attributes contained in dependent measures. Such research would require the
development of reliable and valid indexes of the content-specific dimensions of personality represented in people's implicit
theories. Indeed, some researchers have already taken steps in
this direction. Specifically, Bassili (1981), who also proposed
that the attractiveness stereotype is not as general as implied by
the phrase beauty is good (see discussion early in this article),
used factor analysis to demonstrate that Dion and colleagues'
(1972) Social Desirability Index contains several contentspecific types of evaluative meaning. He also offered a more
general instrument for measuring dimensions of personality
that are affected by positive versus negative appearance.
Other parts of the social psychological research literature on
attractiveness also provide fruitful domains for quantitative reviewing. The most pressing next step is to integrate the results
of the large number of studies that have examined the effects of
physical appearance on how people are treated (e.g, in educational settings, Adams & Cohen, 1974; in judicial proceedings,
Stewart, 1980). Our finding that the strength of the beauty-isgood stereotype depends on the type of inference perceivers
make yields some predictions about treatment effects. For example, targets' physical attractiveness should have greater impact on how people treat them in the general domain of social
competence (eg., peer popularity achieved) than intellectual
competence (e.g., school grades received), integrity (e.g, sentences assigned for legal violations), or concern for others (e.g.,
opportunities provided for prosocial behavior). In addition, by
extending our logic about the dampening effect of individuating information, we expect a weakening of the overall effect of
beauty as one moves from stereotype studies to research assessing behavior in natural settings and even more reduction when
these natural settings require that people interact within the
constraints imposed by social roles and by organizational
norms and hierarchies.
References
Abelson, R. P. (1985). A variance explanation paradox: When a little is
a lot. Psychological Bulletin, 97,129-133.
Adams, G. R. (1977). Physical attractiveness research: Toward a developmental social psychology of beauty. Human Development, 20,
217-239.
Adams, G. R. (1982). Physical attractiveness. In A. G. Miller (Ed), In
the eye of the beholder: Contemporary issues in stereotyping(pp. 253304). New York: Praeger.
Adams, G. R., & Cohen, A. S. (1974). Children's physical and interpersonal characteristics that affect student-teacher interactions. Journal
of Experimental Education, 43,1-5.
Adams, G. R., & Wareham, J. (no date). Beautiful is good: Mechanism of
self and others'perceptions. Unpublished manuscript, Utah State University; Logan, UT.
Ajzen, I. (1988). Attitudes, personality, and behavior. Chicago: Dorsey
Press.
Allen, B. P. (1978). Social behavior: Facts and falsehoods about common
sense, hypnotism, obedience, altruism, beauty, racism, and sexism.
Chicago: Nelson Hall.
Alley, T. R, & Hildebrandt, K. A. (1988). Determinants and consequences of facial aesthetics. In T. R. Alley (Ed), Social and applied
aspects of perceiving faces (pp. 101-140). Hillsdale, NJ: Erlbaum.
Anderson, N. H. (1974). Cognitive algebra: Integration theory applied
to social attribution. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 7, pp. 1-101). New York: Academic Press.
Anderson, N. H. (1981). Integration theory applied to cognitive responses and attitudes. In R. E. Petty, T. M. Ostrom, & T. C. Brock
(Eds.), Cognitive responses in persuasion (pp. 361-397). Hillsdale,
NJ: Erlbaum.
Anderson, R., & Nida, S. A. (1978). Effect of physical attractiveness on
opposite- and same-sex evaluations. Journal of Personality, 46,401413.
Asch, S. E. (1946). Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258-290.
Ashmore, R. D. (1981). Sex stereotypes and implicit personality theory.
In D. L. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup behavior (pp. 37-81). Hillsdale, NJ: Erlbaum.
Ashmore, R. D., & Del Boca, F. K. (1979). Sex stereotypes and implicit
personality theory: Toward a cognitive-social psychological conceptualization. Sex Roles, 5, 219-248.
Ashmore, R. D., Del Boca, E K., & Wohlers, A. J. (1986). Genderstereotypes. In R. D Ashmore & F. K. Del Boca (Eds.), The social psychology of female-male relations: A critical analysis of central concepts
(pp. 69-119). Orlando, FL: Academic Press.
Ashmore, R. D, Tumia, M, & Schreier, P. (1980). Forming impressions
of personality: A reanalysis of the central trait effect through multidimensional scaling. Unpublished manuscript, Rutgers University,
New Brunswick, NJ.
Barnes, M. L., & Rosenthal, R. (1985). Interpersonal effects of experimenter attractiveness, attire, and gender. Journal of Personality and
Social Psychology, 48, 435-446.
Baron, R. A, & Byrne, D. (1987). Social psychology: Understanding
human interaction (5th ed.). Boston: Allyn and Bacon.
Bar-Tal, D, & Saxe, L. (1976). Physical attractiveness and its relationship to sex-role stereotyping. Sex Roles, 2,123-133.
Bassili, J. N. (1981). The attractiveness stereotype: Goodness or glamour? Basic and Applied Social Psychology, 2, 235-252.
Berscheid, E. (1981). A review of the psychological effects of physical
attractiveness. In G. W Lucker, K. A. Ribbens, & J. A. McNamara
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
(Eds), Psychological aspects of facial form (pp. 1-23). Ann Arbor,
MI: Center for Human Growth.
Berscheid, E., & Walster, E. (1974). Physical attractiveness. In L.
Berkowitz (Ed), Advances in experimental social psychology (Vol. 7,
pp. 157-215). New York: Academic Press.
Brigham, J. C. (1980). Limiting conditions of the "physical attractiveness stereotype": Attributions about divorce. Journal of Research in
Personality, 14, 365-375.
Bull, R., & Rumsey, N. (1988). The social psychology of facial appearance. New York: Springer-Verlag.
Bull, R, & Stevens, J. (1979). The effects of attractiveness of writer and
penmenship on essay grades. Journal of Occupational Psychology.
52, 53-59.
Cash, T. F. (1981). Physical attractiveness: An annotated bibliography
of theory and research in the behavioral sciences. Psychological Documents, 11, 83. (Ms. No. 2370)
Cash, T. F. (1985). Physical appearance and mental health. In J. A.
Graham & A. M. Kligman (Eds.), The psychology of cosmetic treatments (pp. 196-216). New York: Praeger.
Cash, T. F, & Janda, L. H. (1984, December). The eye of the beholder.
Psychology Today, pp. 46-52.
Cash, T. F, & Kehr, J. (1978). Influence of nonprofessional counselors'
physical attractiveness and sex on perceptions of counselor behavior.
Journal of Counseling Psychology, 25, 336-342.
Chaiken, S. (1979). Communicator physical attractiveness and persuasion. Journal of Personality and Social Psychology, 37,1387-1397.
Chaiken, S. (1986). Physical appearance and social influence. In C. P.
Herman, M. P. Zanna, & E. T. Higgins (Eds), Physical appearance,
stigma, and social behavior: The Ontario symposium (Vol. 3, pp. 143177). Hillsdale, NJ: Erlbaum.
Chaiken, S., & Eagly, A. H. (1983). Communication modality as a determinant of persuasion: The role of communicator salience. Journal of
Personality and Social Psychology, 45, 241-256.
Cohen, S. (1977). Statistical power analysis for the behavioral sciences
(rev. ed). New York: Academic Press.
Cooper, H. M., & Findley, M. (1982). Expected effect sizes: Estimates
for statistical power analysis in social psychology. Personality and
Social Psychology Bulletin, 8,168-173.
Courtney, A. E., & Whipple, T. W (1983). Sex stereotyping in advertising. Lexington, MA: Lexington Books.
Deaux, K., & Lewis, L. L. (1983). Components of gender stereotypes.
Psychological Documents, 13. 25. (Ms. No. 2583)
Deaux, K., & Lewis, L. L. (1984). Structure of gender stereotypes:
Interrelationships among components and gender label. Journal of
Personality and Social Psychology, 46, 991-1004.
Deaux, K., & Wrightsman, L. S. (1988). Social psychology. Pacific
Grove, CA: Brooks/Cole.
Del Boca, F. K., & Ashmore, R. D. (1980). Sex stereotypes through the
life cycle. In L. Wheeler (Ed), Review of personality and social psychology (Vol. 1, pp. 163-192). Beverly Hills, CA: Sage.
Dernier, M., & Thiel, D. L. (1975). When beauty may fail. Journal of
Personality and Social Psychology, SI, 1168-1176.
Dibiase, W J., &Hjelle, L. A. (1968). Body-image stereotypes and bodytype preferences among male college students. Perceptual and Motor
Skills, 27,1143-1146.
Dion, K. K. (1981). Physical attractiveness, sex roles and heterosexual
attraction. In M. Cook (Ed), The bases of human sexual attraction
(pp. 3-22). London: Academic Press.
Dion, K. K. (1986). Stereotyping based on physical attractiveness: Issues and conceptual perspectives. In C. R Herman, M. P. Zanna, &
E. T. Higgins (Eds), Physical appearance, stigmaand social behavior:
The Ontario Symposium (Vol. 3, pp. 7-21). Hillsdale, NJ: Erlbaum.
Dion, K. K., Berscheid, E., & Walster, E. (1972). What is beautiful is
good. Journal of Personality and Social Psychology, 24, 285-290.
125
Downs, A. C, & Harrison, S. K. (1985). Embarrassing age spots or just
plain ugly? Physical attractiveness stereotyping as an instrument of
sexism on American television commercials. Sex Roles, 13,9-19.
Eagly, A. H. (1987). Sex differences in social behavior: A social-role
interpretation. Hillsdale, NJ: Erlbaum.
Eagly, A. H., & Johnson, B. T. (1990). Gender and leadership style: A
meta-analysis. Psychological Bulletin, 108, 233-256.
Eagly, A. H., & Steffen, V J. (1984). Gender stereotypes stem from the
distribution of women and men into social roles. Journal of Personality and Social Psychology, 46, 735-754.
Eagly, A. H., & Steffen, V J. (1986). Gender and aggressive behavior: A
meta-analytic review of the social psychological literature. Psychological Bulletin, 100, 309-330.
Eagty, A. H., & Wood, W (in press). Explaining sex differences in social
behavior: A meta-analytic perspective. Personality and Social Psychology Bulletin.
Feingold, A. (1990a). Gender differences in effects of physical attractiveness on romantic attraction: A comparison across five research
paradigms. Journal of Personality and Social Psychology, 59, 981993.
Feingold, A. (1990b). Good-looking people are not what we think. Unpublished manuscript.
Ghiselli, E. E. (1964). Theory of psychological measurement. New York:
McGraw-Hill.
Glass, G. V, McGaw, B., & Smith, M. L. (1981). Meta-anafysis in social
research. Beverly Hills, CA: Sage.
Goldman, W, & Lewis, P. (1977). Beautiful is good: Evidence that the
physically attractive are more socially skillful. Journal of Experimental Social Psychology, 13,125-130.
Greenwald, A. G. (1975). Consequences of prejudice against the null
hypothesis. Psychological Bulletin, 82,1-20.
Gross, A. E, & Crofton, C. (1977). What is good is beautiful. Sociometry, 40, 85-90.
Hamilton, D. L. (Ed). (1981). Cognitive processes in stereotyping and
intergroup behavior. Hillsdale, NJ: Erlbaum.
Hartnett, J. I, & Secord, G. (1983). Physical attraction and its effects on
the perception of extra-marital affairs. Perceptual and Motor Skills,
56, 310.
Hatfield, E., & Sprecher, S. (1986). Mirror, mirror: The importance of
looks in everyday life. Albany: State University of New York Press.
Hedges, L. V(1981). Distribution theory for Glass's estimator of effect
size and related estimators. Journal of Educational Statistics, 6,107128.
Hedges, L. V (1982a). Fitting categorical models to effect sizes from a
series of experiments. Journal of Educational Statistics, 7,119-137.
Hedges, L. V (1982b). Fitting continuous models to effect size data.
Journal of Educational Statistics, 7,245-270.
Hedges, L. V (1987). How hard is hard science, how soft is soft science?
The empirical cumulativeness of research. American Psychologist,
42, 443-455.
Hedges, L. V, & Becker, B. J. (1986). Statistical methods in the metaanalysis of research on gender differences. In J. Hyde & M. C. Linn
(Eds), The psychology of gender: Advances through meta-anafysis
(pp. 14-50). Baltimore, MD: Johns Hopkins University Press.
Hedges, L. V, & Olkin, I. (1985). Statistical methods for meta-anafysis.
Orlando, FL: Academic Press.
Hocking, J. E, Walker, B. A., & Fink, E. L. (1982). Physical attractiveness and judgments of morality following an "immoral" act. Psychological Reports, 51,111-116.
Jackman, D. M. (1979). Effects of physical attractiveness and sex on the
perception of assertiveness, aggressiveness, and submissiveness.
Dissertations Abstracts International, 39, 5648B. (University Microfilms No. 79-10272)
126
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
Jackson, L. A, Sullivan, L. A., & Rostker, R. (1988). Gender, gender
role, and body image. Sex Roles, 19,429-443.
Jacobson, M. B., & Koch, W (1978). Attributed reasons for support of
the feminist movement as a function of attractiveness. Sex Roles, 4,
169-174.
Jacobson, M. B, & Popovich, P. M. (1983). Victim attractiveness and
perceptions of responsibility in an ambiguous rape case. Psychology
ofWomen Quarterly, 8,100-104.
Johnson, B. T, & Eagly, A. H. (1989). The effects of involvement on
persuasion: A meta-analysis. Psychological Bulletin, 106, 290-314.
Kim, M. R, & Rosenberg, S. (1980). Comparison of two structural
models of implicit personality theory. Journal of Personality and Social Psychology, 38, 375-389.
Lane, D. M., & Dunlap, W P. (1978). Estimating effect size: Bias resulting from the significance criterion in editorial decisions. British
Journal of Mathematical and Statistical Psychology, 31,107-112.
Langlois, J. H. (1986). From the eye of the beholder to behavioral reality: Development of social behaviors and social relations as a function of physical attractiveness. In C. P. Herman, M. P. Zanna, & E. T.
Higgins (Eds.), Physical appearance, stigma, and social behavior:The
Ontario Symposium (Vol. 3, pp. 23-51). Hillsdale, NJ: Erlbaum.
Larsen. R. J., Diener, E., &Cropanzano, R. S. (1987). Cognitive operations associated with individual differences in affect intensity. Journal of Personality and Social Psychology, 53, 767-774.
Lerner, R., & Lerner, J. (1977). Effects of age, sex and physical attractiveness on child-peer relations, academic performance, and elementary school adjustment. Developmental Psychology, 13,585-590.
Locksley, A, Borgida, E., Brekke, N., & Hepburn, C. (1980). Sex stereotypes and social judgment. Journal of Personality and Social Psychology, 39.S21-S31.
Locksley, A, Hepburn, C, & Ortiz, Y (1982). Social stereotypes and
judgments of individuals: An instance of the base-rate fallacy. Journal of Experimental Social Psychology, 18, 23-42.
Martin, P. 1, Friedmeyer, M. H, & Moore, J. E. (1977). Pretty patient/
healthy patient? A study of physical attractiveness and psychopathology. Journal of Clinical Psychology, 33,990-994.
McArthur, L. Z. (1982). Judging a book by its cover: A cognitive analysis of the relationship between physical appearance and stereotyping. In A. H. Hastorf & A. M. Isen (Eds.), Cognitive social psychology
(pp. 149-211). New York: Elsevier.
Miller, A. G., Gillen, B., Schenker, C., & Radlove, S. (1974). The prediction and perception of obedience to authority. Journal of Personality,
42, 23-42.
Mullen, B. (1989). Advanced BASIC meta-analysis. Hillsdale, NJ: Erlbaum.
Nisbett, R. E.,& Ross. L. (1980). Human inference: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Prentice-Hall.
Nordholm, L. A. (1980). Beautiful patients are good patients: Evidence
for the physical attractiveness stereotype in first impressions of patients. Social Science and Medicine, 14A, 81-83.
O'Reilly, M. T. (1987). Physical attractiveness of professors in dental
education: Its relationship to students' ratings of teaching effectiveness (Doctoral dissertation, University of Pittsburgh, 1986). Dissertation Abstracts International, 47, 433A.
Osgood, C. E., Suci, G. J, & Tannenbaum, P. H. (1957). The measurement of meaning. Urbana: University of Illinois Press.
Patzer, G. L. (1985). The physical attractiveness phenomena. New York:
Plenum Press.
Pilkonis, P. A. (1977). The behavioral consequences of shyness. Journal
of Personality, 45, 596-611.
Roberts, E. J. (1982). Television and sexual learning in children. In D.
Pearl, L. Bouthilet, & J. Lazar (Eds), Television and behavior: Ten
years of scientific progress and implications for the eighties (Vol. 2, pp.
209-223). Washington, DC: US. Government Printing Office.
Rodin, J, Silberstein, L., & Striegel-Moore, R. (1985). Women and
weight: A normative discontent. In T. B. Sonderegger & R. A.
Dienstbier (Eds), Nebraska Symposium on Motivation 1984 (Vol. 33,
pp. 267-307). Lincoln: University of Nebraska Press.
Rosenberg, S. (1977). New approaches to the analysis of personal constructs in person perception. In A. L. Land & J. K. Cole (Eds.),
Nebraska Symposium on Motivation 1976 (Vol. 24, pp. 179-242).
Lincoln: University of Nebraska Press.
Rosenberg, S., Nelson, C, & Vivekananthan, P. S. (1968). A multidimensional approach to the structure of personality impressions.
Journal of Personality and Social Psychology, 9, 283-294.
Rosenberg, S., & Sedlak, A. (1972). Structural representations of implicit personality theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 6, pp. 235-297). New York: Academic
Press.
Rosenthal, R., & Rubin, D. B. (1982). A simple, general purpose display
of magnhudeof experimental effect. Journal of Educational Psychology, 74,166-169.
Rosenthal, R., & Rubin, D. B. (1986). Mela-analytic procedures for
combining studies with multiple effect sizes. Psychological Bulletin,
99, 400-406.
Scheibe, C. (1979). Sex roles in TV commercials. Journal of Advertising
Research, 19, 23-27.
Schneider, D. J. (1973). Implicit personality theory: A review. Psychological Bulletin, 79, 294-309.
Schneider, P. A, Conger, A, & Firth, E. A. (1986, April). Sex as a
moderator of the beauty-good relationship. Paper presented at the
meeting of the American Association for Counseling and Development, Los Angeles.
Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and
interpersonal behavior: On the self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666.
Stewart, J. E. (1980). Defendant's attractiveness as a factor in the outcome of criminal trials: An observational study. Journal of Applied
Social Psychology, 10, 348-361.
Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive "vividness"
effect. Psychological Review, 89,155-181.
Touhey, J. C. (1979). Sex-role stereotyping and individual differences in
liking for the physically attractive. Social Psychology Quarterly, 42,
285-289.
Turkat, D., & Dawson, J. (1976). Attributions of responsibility for a
chance event as a function of sex and physical attractiveness of target
individual. Psychological Reports, 39, 275-279.
Wallston, B. S, & O"Leary, Y E. (1981). Sex makes a difference: Differential perceptionsof women and men. In L. Wheeler (Ed),/tevfewo/
personality and social psychology (Vol. 2, pp. 9-41). Beverly Hills,
CA: Sage.
White, G. M. (1980). Conceptual universal in interpersonal language.
American Anthropologist, 82, 759-781.
Wood, W, Rhodes, N., & Whelan, M. (1989). Sex differences in positive
well-being: A meta-analytic review of the effects associated with
marital status. Psychological Bulletin, 106, 249-264.
REVIEW OF PHYSICAL ATTRACTIVENESS STEREOTYPE
127
Appendix
References for Studies in Meta-Analysis
Adams, G. R., & Huston, T. L. (1975). Social perception of middle-aged
persons varying in physical attractiveness. Developmental Psychology, 11. 657-658.
Alicke, M. n, Smith, R. H., & Klotz, M. L. (1986). Judgments of physical attractiveness: The role of faces and bodies. Personality and Social Psychology Bulletin, 12, 381-389.
Anderson, S. M., & Bern, S. L. (1981). Sex typing and androgyny in
dyadic interaction: Individual differences in responsiveness to physical attractiveness. Journal of Personality and Social Psychology, 41,
74-86.
Armstrong, D P. (1979). The facial attractiveness riddle (Doctoral dissertation, Southern Illinois University at Carbondale, 1978). Dissertation Abstracts International, 39, 5127B.
Barnes, M. L, & Rosenthal, R. (1985). Interpersonal effects of experimenter attractiveness, attire, and gender. Journal of Personality and
Social Psychology, 48, 435-446.
Barocas, R., & Karoly, P. (1972). Effects of physical appearance on
social responsiveness. Psychological Reports, 31, 495-500.
Bassili, J. N. (1981). The attractiveness stereotype: Goodness or glamour? Basic and Applied Social Psychology, 2, 235-252.
Benassi, M. A. (1981). The effects of order of presentation and physical
attractiveness on attributions of ability (Doctoral dissertation, De
Paul University). Dissertation Abstracts International, 42, 2596B.
Black, H. K. (1974). Physical attractiveness and similarity of attitude in
interpersonal attraction. Psychological Reports, 35, 403-406.
Boor, M., & Zeis, E R. (1975). Effect of physical attractiveness on 1Q
estimation: A failure to extend results of prior research. Psychological Documents, 5, 234. (Ms. No. 929).
Byrne, D, London, O., & Reeves, K. (1968). The effects of physical
attractiveness, sex, and attitude similarity on interpersonal attraction. Journal of Personality, 36,259-271.
Carter, J. A. (1978). Impressions of counselors as a function of counselor physical attractiveness. Journal ofCou,:'eling Psychology, 25,
28-34.
Cash, T. E, Begley, P I, McCown, D. A, & Weise, B. C. (1975). When
counselors are heard but not seen: Initial impact of physical attractiveness. Journal of Counseling Psychology, 22, 273-279.
Cash, T. E, Kehr, J. A., Poryson, J., & Freeman, V (1977). Role of physical attractiveness in peer attribution of psychological disturbance.
Journal of Consulting and Clinical Psychology, 45, 987-993.
Covington, J. K. (1981). The communicator's physical attractiveness
and credibility as determinants of the effectiveness of a speech (Doctoral dissertation, Louisiana State University). Dissertation Abstracts International, 42, 437A.
Cunningham, M. R. (1986). Measuring the physical in physical attractiveness: Quasi-experiments on the sociobiology of female facial
beauty. Journal of Personality and Social Psychology, 50, 925-935.
Dermer, M. (1974). When beauty fails (Doctoral dissertation, University of Minnesota, 1973). Dissertation Abstracts International, 34,
4402A.
Dermer, M, & Thiel, D. L. (1975). When beauty may fail. Journal of
Personality and Social Psychology, J/,1168-1176.
Dion, K.., Berscheid, E, & Walster, E. (1972). What is beautiful is good.
Journal of Personality and Social Psychology, 24, 285-290.
Dushenko, T. W, Perry, R. P., Schilling, J., &Smolarski, S. (1978). Generality of the physical attractiveness stereotype forage and sex. Journal
of Social Psychology, 105, 303-304.
Gallucci, N. T. (1984). Effects of men's physical attractiveness on interpersonal attraction. Psychological Reports, 55, 935-938.
Gallucci, N. T, & Meyer, R. G. (1984). People can be too perfect:
Effects of subjects' and targets' attractiveness on interpersonal attraction. Psychological Reports, 55, 351-360.
Gillen, B. (1981). Physical attractiveness: A determinant of two types
of goodness. Personality and Social Psychology Bulletin, 7.277-281.
Gillen, B. (1975). Physical attractiveness and sex as determinants of
trait attributions. Dissertation Abstracts International, 36,1945B.
(University Microfilms No. 75-21,693)
Gillen, B., & Sherman, R. C. (1980). Physical attractiveness and sex as
determinants of trait attributions. Multivariate Behavioral Research,
15, 423^37.
Green, C. E, Cunningham, 1, & Yanico, B. J. (1986). Effects of counselor and subject race and counselor physical attractiveness on impressions and expectations of a female counselor. Journal of Counseling Psychology, 33, 349-352.
Guise, B. J., Pollans, C. H., & Turkat, I. D. (1982). Effects of physical
attractiveness on perception of social skill. Perceptual and Motor
Skills, 54,1039-1042.
Hailey, B. J. (1976). Selective attention as a factor in the maintenance of
physical attractiveness and sex role stereotypes (Doctoral dissertation, University of Georgia, 1975). Dissertation Abstracts International, 36, 4227B.
Havassy, V J. (1981). Reactions to professional women as a function of
attractiveness and level of competence (Doctoral dissertation, California School of Professional Psychology). Dissertation Abstracts International, 42, 2130B.
Heilman, M. E.,&Stopeck, M. H. (1985). Attractiveness and corporate
success: Different causal attributions for males and females. Journal
of Applied Psychology, 70, 379-388.
Hicks, R. A., Pellegrini, R. I, & Tomlinson, N. (1978). Attributions of
female college students to male photographs as a function of attractiveness and pupil size. Perceptual and Motor Skills, 47,1265-1266.
Hill, M. K., & Lando, H. A. (1976). Physical attractiveness and sex-role
stereotypes in impression formation. Perceptual and Motor Skills,
43,1251-1255.
Jackson, L. A. (1983). The perception of androgyny and physical attractiveness: Two is better than one. Personality and Social Psychology
Bulletin, 9, 405-413.
Janda, L. H., O'Grady, K. E., & Barnhart, S. A. (1981). Effects of sexual
attitudes and physical attractiveness on person perception of men
and women. Sex Roles, 7,189-199.
Joseph, W B. (1977). Effect of communicator physical attractiveness
and expertness on opinion change and information processing. Dissertation Abstracts International, 38, 5017A. (University Microfilms
No. 77-31,899)
Kaplan, R. M. (1978). Is beauty talent? Sex interaction in the attractiveness halo effect. Sex Roles, 4,195-204.
Kleinke, C. L, & Kahn, M. L. (1980). Perceptions of self-disclosers:
Effects of sex and physical attractiveness. Journal of Personality, 48,
190-205.
Kleinke, C. L., Staneski, R. A, & Pipp, S. L. (1975). Effects of gaze,
distance, and attractiveness on males' first impressions of females.
Representative Research in Social Psychology, 6, 7-12.
Kollar, M. M. (1973). The beautiful is rotten phenomenon: A negative
stereotype of physical attractiveness. Dissertation Abstracts International, 34, 4632B. (University Microfilms No. 74-07,769)
Landy, D., & Sigall, H. (1974). Beauty is talent: Task evaluation as a
functionof the performer's physical attractiveness. Journal of Personality and Social Psychology, 29, 299-304.
128
EAGLY, ASHMORE, MAKHIJANI, AND LONGO
La Voie, J. C, & Adams, G. R. (1978). Physical and interpersonal attractiveness of the model and imitation in adults. Journal of Social Psychology, 106,191-202.
Lee, L. J., Adams, G. R, & Dobson, W R. (1984). Male and female
attributions and social influence behavior towards a physically attractive female. Journal of Psychology, 117,97-103.
Lewis, K. N, & Walsh, W B. (1978). Physical attractiveness: Its impact
on the perception of a female counselor. Journal of Counseling Psychology, 25, 210-216.
Longtain, M. J. (1981). Status generalization and performance expectation effects of sex role stereotypes, physical attractiveness, and
expertise. Dissertation Abstracts International, 42, 3034B. (University Microfilms No. 81-28,655)
Lucker, G. W (1977). Physical attractiveness, individual differences
and personality stereotyping (Doctoral dissertation, University of
Texas at Austin, 1976). Dissertation Abstracts International, 37,
4219B.
Maddux, J. E, & Rogers, R. W (1980). Effects of source expertness,
physical attractiveness, and supporting arguments on persuasion: A
case of brains over beauty. Journal of Personality and Social Psychology, 39, 235-244.
Miller, A. G. (1970). Role of physical attractiveness in impression formation. Psychonomic Science, 19, 241-243.
Mims, P. R, Hartnett, J. J., & Nay, W R. (1975). Interpersonal attraction and help volunteering as a function of physical attractiveness.
Journal of Psychology, 89,125-131.
Molberg, A. N. (1977). The effects of interpersonal distance, physical
attractiveness, and sex on impression formation (Doctoral dissertation, University of Minnesota, 1976). Dissertation Abstracts International, 37, 6408B.
Moore, J. S., Graziano, W G., & Miller, M. G. (1987). Physical attractiveness, sex role orientation, and the evaluation of adults and children.
Personality and Social Psychology Bulletin, 13, 95-102.
Nielsen, J. P., & Kernaleguen, A. (1976). Influence of clothing and
physical attractiveness in person perception. Perceptual and Motor
Skills, 42, 775-780.
Osterkamp, M. B. (1980). Impressions of older persons in initial interactions (Doctoral dissertation, University of Kansas, 1979). Dissertation Abstracts International, 41,849A.
Patzer, G. L. (1983). Source credibility as a function of communicator
physical attractiveness. Journal of Business Research, 11, 229-241.
Pistilli.J. A. (1976). Stereotyped perceptionsof women asafunctionof
menstrual cycle phase and physical attractiveness (Doctoral dissertation, Kent State University, 1975). Dissertation Abstracts International, 36, 4760B.
Reeves, R. A. (1982). The effects of source physical attractiveness, similarity, and message type on persuasion: When beauty is not enough
(Doctoral dissertation, University of Georgia). Dissertation Abstracts International, 43, 2044B.
Saner, S. I. (1983). Beyond the halo effect: Physical attractiveness and
impression formation as a function of cue availability (Doctoral dissertation, University of Connecticut). Dissertation Abstracts International, 44,1253B.
Sanders, C. H, & Borowy, T. D. (1983). Physical attractiveness: Its influence on the perceptions of counselors (Report No. CG-017-587).
Knoxville: University of Tennessee. (ERIC Document Reproduction Service No. ED 246-345)
Saxe, L. (1976). Perceiver characteristics as determinants of judgments
of physically attractive and unattractive others (Doctoral dissertation, University of Pittsburgh, 1975). Dissertation Abstracts International, 37, 2577B.
Small-Weil, S. B. (1981). The relationship between gender, physical
attractiveness and occupational prestige (Doctoral dissertation,
Tufts University, 1980). Dissertation Abstracts International, 41,
3613B.
Snyder, M, &Rothbart, M. (1971). Communicator attractiveness and
opinion change. Canadian Journal of Behavioural Science, 3, 377387.
Solomon, S., & Saxe, L. (1977). What is intelligent, as well as attractive,
is good. Personality and Social Psychology Bulletin, 3, 670-673.
Stretch, R. H, & Figley, C. R. (1980). Beauty and the boast: Predictors
of interpersonal attraction in a dating experiment. Psychology: A
Quarterly Journal of Human Behavior, 17, 35-43.
Tanke, E. D. (1982). Dimensions of the physical attractiveness stereotype: A factor/analytic study. Journal of Psychology, 110, 63-74.
Thornton, B., & Linnstaedter, L. (1980). The influence of physical
attractiveness and sex-role congruence on interpersonal attraction.
Representative Research in Social Psychology, 11, 55-63.
Webster, M., & Driskell, J. E. (1983). Beauty as status. American Journal of Sociology, 89,140-165.
Widgery, R. N. (1974). Sex of receiver and physical attractiveness of
source as determinants of initial credibility perception. Western
Speech, 38,13-17.
Widgery, R. N., & Webster, B. (1969). The effects of physical attractiveness upon perceived initial credibility. Michigan Speech Journal, 4,
9-19.
Wiener, S., Saxe, L, & Bar-Tal, D. (1975). The effects of physical attractiveness on attributions of causality for success and failure (Report
No. CG-010-288). Pittsburgh, PA: University of Pittsburgh. (ERIC
Document Reproduction Service No. ED 117-588)
Wilson, M., Crocker, J., Brown, C. E., Johnson, D, Liotta, R, & Konat,
J. (1985). The attractive executive: Effects of sex of business associates on attributions of competence and social skills. Basic and Applied Social Psychology, 6, 13-23.
Received May 17,1989
Revision received November 20,1989
Accepted November 6,1990 •