Академический Документы
Профессиональный Документы
Культура Документы
REVIEWS
Further
CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Term inolog y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . .
Overv iew o f the Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Emp ir ical re searc h. . . . . .. . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Theo ry . . . . . . . . . . . . .. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . .
Me thodology .... . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . .. . . . . . . . . .. . . . . . . . . .. .. . . . ... . ..
Soc ia l prob lem s. . . . . . . ..... . . .. . .. . . . . . . . . . . . . . . . .. . . . . . .. . . . . . . . . .. . . . . ... . . . ..
...
S cope an d Roun dar ies of Coverag e . . . . . ....
.
. .
. .
. . .
SEX COMPARISONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
Cogn it ive S kill s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mat hema t ical ab il it y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Spa tia l ab il ity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
V erbal ab il it y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Per sonality Tra its an d D ispo sition s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A chievemen t . . . . . .. . . . . . .. . . . .. . . . . . . . . . . . . . . . . . . . . . . . . ... . . .. . . . . . . . . . . .... . . . . . . ... . ... .. . . ..
Ma scul in it y an dfem in in it y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mo ral development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Soc ial Be hav io rs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A ggre ssion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Co nfo rmit y an d soc ia l iriflu enc e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nonverbal be hav io r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cen tr al I ssues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mean in gfulne ss of diffe rence s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Comparison of labo rato ry an dfiel d stu dies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cause s of se x differences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
50
51
51
51
52
53
53
54
54
55
55
56
56
56
57
58
59
60
60
61
61
62
62
63
64
65
65
66
66
69
70
70
71
72
72
CONCLUDING THOUGHTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
49'
0066-4308/85/0201-0049$02.00
50
DEAUX
INTRODUCTION
A chapter entitled "Sex and Gender" has never before appeared in the Annual
Review of Psychology; nonetheless, there are important predecessors to this
piece. In 1 975, Mednick and Weissman authored a review of selected topics in
the psychology of women. Volumes in other Annual Review series have
included chapters on sex roles (Lipman-Blumen & Tickamyer 1 975, Miller &
Garrison 1 982) , sex differences and language (Philips 1 980), women's status
(Quinn 1 977) , and family structure (Yanagisako 1 979) . The most recent edition
of the Handbook of Social Psychology includes a chapter on sex roles (Spence
et alI985), the first such chapter since Catherine Cox Miles ( 1935) contributed
to the original Murchison volumes. The dates of these reviews provide some
indication of the resurgence of this topic area, a currency that can be further
documented in a variety of ways.
Restricting consideration to the psychological literature for the moment, one
finds two journals that have been established within the past 10 years: Sex Roles
and Psychology of Women Quarterly. first published in 1 975 and 1 976 respec
tively. Virtually all other psychology journals have shown an increase in the
number of articles dealing with sex and gender during the same period. The
extent of this activity is evidenced by the number of articles included in
Psychological Abstracts between 1 967 and 1 982 that are indexed under human
sex differences (N = 1 2,689) , sex role attitudes (N = 1 765), and sex roles
(N=362 1 ) . Supplementing the flow of empirical articles are a number of
influential books , including those by Maccoby & Jacklin ( 1 974) , Spence &
Helmreich ( 1 978), and Gilligan ( 1 982) . Edited collections such as those by
Lloyd & Archer ( 1 976) , Mayo & Henley ( 1 9 8 1 ) , and O'Leary et al ( 1984)
further stock the field.
Although this quantity certainly prompts the most dedicated reviewer to cry
"Enough!", analysis of sex and gender cannot in good conscience be restricted
to the psychological literature. The interdisciplinary women's studies journal
Signs has been a rich source of material related to gender, including periodic
reviews of the state of the art in psychology (Henley 1 984; Parlee 1 975, 1 979;
Vaughter 1 976) and other fields. Disciplinary journals in fields such as sociolo
gy, anthropology, and history have also included many articles pertinent for the
psychologist interested in gender. Similarly, a wide variety of books by
sociologists, anthropologists, and other social scientists have informed the
analysis, including those by Bernard ( 1 98 1 ) , Chodorow ( 1 978), and Sanday
( 1 98 1 ) . This magnitude mandates selectivity-a selectivity that is inevitably
biased by the viewpoint of the author. To render that viewpoint something less
than opaque, let me draw with rough strokes some of the themes that will be
developed in this chapter: first, general terminology; second, a brief but critical
overview of the field; and third, the specific issues that will be addressed.
51
Terminology
To avoid later confusion, I should stipulate how I will use some terms.
Terminology in this area has provided a continuing forum for debate (cf M.
Gould 1980, Unger 1 979), and my attempt to define terms here is intended less
as an ultimate resolution of controversy than as a clarification for the reader. By
sex, I will be referring to the biologically based categories of male and female.
In the use of gender, I refer to the psychological features frequently associated
with these biological states, assigned either by an observer or by the individual
subject. Thus studies that select two groups of subjects based on their biological
characteristics will be considered appropriate for use of the word sex. In this
context, one is studying sex differences rather than gender differences. In
contrast, if judgments are made about nonbiological characteristics or social
categories, then gender will be used as a referent; hence the appearance of such
terms as gender identity, gender stereotypes , and gender roles.
In avoiding use of the term "sex role," I concur with the contention of the late
Carolyn Sherif: the term "has become a boxcar carrying an assortment of
sociological and psychological data along with an explosive mixture of myth
and untested assumptions" ( 1 982, p. 392). Too frequently this label is used as
explanation, deflecting more careful consideration of the behaviors of interest.
In general, more consistent use of terms would clarify many of the discus
sions in this area. The confusion, however, is not merely an issue of semantics.
Frequently underlying the debate on the use of sex versus gender, for example,
are assumptions about the determinants of differences between men and
women, whereby sex often implies biological causes while gender invokes
explanations based on socialization. Resolution of these controversies is prob
ably not imminent, although recognition of some of the sources may hasten the
process.
Overview
of the Area
To characterize research and theory in the area of sex and gender, recognizing
both the breadth of topics and the numerosity of reports, is a difficult task and
the reader must be willing to accept some generalizations that undoubtedly
have their exceptions . With that caveat in mind, let me offer a brief analysis of
the area in terms of empirical work, theoretical contributions, methodological
concerns , and problem-oriented efforts .
Empirical studies have certainly dominated the activ
EMPIRICAL RESEARCH
ity within psychology . After years of relative neglect, questions related to sex
and gender have become "hot topics," and many previous vacuums have been
filled. The scope of this empirical work is broad. Comparisons of the sexes in
virtually every form of behavior imaginable have been reported, and an
52
DEAUX
53
theory. [Two additional review papers that could be considered relevant to the
broadly defined topic, but which are outside the range of the present coverage,
are those by Storms ( 1 98 1 ) on the development of erotic orientation and by
Rasmussen ( 1 98 1 ) on pair bonding.]
Although these early attempts have whetted the theoretical appetite, one
hungers for more nourishing conceptual work (also cf Wittig 1984). Questions
of context and process have not yet been thoroughly addressed, and the
potential contributions of other disciplines have not been given their due. It is
not the function of this chapter to create theory. However I will, in reviewing
the research, attempt to identify those pieces that might contribute most to such
efforts, as well as to the missing pieces that need to be found.
Following on the heels of demonstrations that the annals of
psychological research show numerous signs of bias, ranging from selection of
subjects to interpretation of findings, many questions have been raised about
the appropriate ways to do research relating to scx and gender (e. g . Roberts
1 98 1 , Unger 1 983, Wallston & Grady 1 984) . Some critics have posed broad
challenges, suggesting the operation of a double standard in even the newer
research formulations (Eichler 1980). Others have more specifically analyzed
problems in design and interpretation (e. g. Lloyd 1 976, Sherman 1978, Grady
1981 , Parlee 1 9 8 1 ) . Still others have argued that the study of gender requires
the use of more descriptive methods such as ethnomethodology (Kessler &
McKenna 1978, Atkinson 1982).
In many respects , these arguments parallel those raised in the fields of
personality and social psychology in general, reflecting suspicions that a
dedicated commitment to laboratory experiments may not be sufficient (and
some would argue not even necessary) to develop a full understanding of
human behavior (Mishler 1 979 , Sherif 1979, Rorer & Widiger 1 983, Wallston
1 983) . For sex and gender, these critiques become particularly salient. The
social reality of gender cannot be removed from or fully controlled by the
laboratory experiment, and recognition of this reality mandates both concern
and caution (also cf Unger 198 1 ).
METHODOLOGY
54
DEAUX
SEX COMPARISONS
The longest history and the most extensive list of references within the area of
sex and gender in all probability belong to the investigation of sex differences
(cf Rosenberg 1 982). The accepted terminology itself suggests a belief in the
existence of such differences, and failures to find differences are less often
regarded as evidence of sex similarities than as a state of confusion or uncer
tainty. As suggested by the number of references accumulated within the past.
1 5 years, the areas in which such potential differcnces have been sought are
numerous, and only a highly selective sampling will be attempted here.
Maccoby & Jacklin's ( 1 974) compendium has served as the lightning rod for
most investigations during the past 1 0 years, as investigators have tried to
verify, modify, and extend the findings of this pair. Relying primarily on child
data collected prior to 1 970, Maccoby and Jacklin concluded that convincing
evidence for sex differences could be found in only four areas: verbal ability,
mathematical ability, visual-spatial ability, and aggression. (For a critical
55
Cognitive Skills
A fascination with the distribution of intellectual ability has been a hallmark of
American psychological inquiry, perhaps reflecting a belief that a meritocracy
depends on the ability to answer the question, "Who is very smart?". Questions
of possible differences between the sexes in i ntellectual capacity have a history
nearly as long, with a philosophical underpinning that extends further decades
back. I Numerous accounts of these early debates are available (Shields 1975,
1982; Sherman 1978; Rosenberg 1982).
The more recent comparisons of the sexes have, in accord with the zeitgeist
of intclligcnce research, focused on specific aspects of cognitive ability rather
than a global concept of general intelligence. Most typically, following on the
heels of Maccoby & Jacklin (1974), investigators have looked for sex differ
ences in verbal, spatial, and mathematical ability. The data are numerous at this
point, and the debates as to their meaning have often been acrimonious-an
acrimony reflecting the emotional values that come to the fore so readily when
issues of nature vs nurture emerge as they do in analyzing intellectual capabil
ity. (See Sherman 1978, Wittig & Petersen 1979 for reviews of these posi
tions.)
MATHEMATICAL ABILITY
The question of differences in mathematical abil
ity has been perhaps the most lively area of debate. Maccoby & Jacklin's
(1974) conclusion that boys are superior to girls in mathematical performance,
but that the differences primarily occur from the period of adolescence onward,
was followed by meta-analyses suggesting that only 1 % of the variance is
explained by this difference (Hyde 1981). More recently, the work of Benbow
& Stanley (1980, 1983) has provided new data documenting sex differences in
mathematical aptitude, differences that they suggest are evident prior to
lit is worth noting in this context that Gustave Le Bon, often considered a founder of social
psychology as a discipline, was one of the most virulent in his pronouncements of sex differences.
As one example from a lengthier diatribe, published in 1 879: "All psychologists who have studied
the intelligence of women, as well as poets and novelists, recognize today that they represent the
most inferior forms of human evolution and that they are closer to children and savages than to an
adult, civilized man" (quoted in S. J. Gould 1980).
56
DEAUX
adolescence. Using the SAT test as a criterion with a large sample of seventh
grade students , these authors find a mean difference of approximately 30 points
between boys and girls. Emphasied in their report is the difference between
males and females when one considers the sex ratio at higher levels of perform
ance. Thus at scores of 700 and above (a performance level achieved by 1 of
1 0,000 students) , the ratio of males to females is approximately 1 3 to 1.
Subsequent analyses of these data suggest the male-female difference occurs
primarily on algebraic items and is not evident on arithmetical or geometric
problems (Becker 1983).
SPATIAL ABILITY
Spatial abilities have also been subject to scrutiny, follow
ing Maccoby and Jacklin's report of sex differences and Hyde's (1981) assess
ment of 4 . 5 % of the variance explained by this factor. In part, this area has been
of interest because of the presumed association between spatial skills and
achievement in mathematics and science, an association that has not been
supported by recent analyses (Meece et al 1982, Linn & Petersen 1983). The
sex difference in spatial skill appears to emerge prior to adolescence but is
limited to specific types of skills. Specifically, males are found to be superior to
females in measures of mental rotation and in tests of horizontality-verticality
(such as the rod and frame test) , but there are no apparent differences in spatial
visualization tasks that require a more analytic, sequential strategy and that are
in fact closer to the demands of mathematical and scientific reasoning (Linn &
Petersen 1983). Further, as in the case of mathematical ability, there is
substantial evidence that training can significantly alter the performance of both
women and men (Connor et al 1978, Newcombe et al 1983).
57
58
DEAUX
model is the assumption that settings, and particularly the roles and positions
defined within those settings , may influence a particular stylistic choice.
The sequence by which aptitude and motivation interact to produce achieve
ment in academic settings has been studied extensively by Eccles and her
colleagues [Eccles (Parsons) 1 983a,b; Eccles (Parsons) et a1 1 984; Meece et al
1 982]. Beginning with the structure of a traditional expectancy-value model of
achievement, Eccles specifies a number of cognitive elements including self
schemata, perceptions of task value, and a child's perceptions of the parents '
attitudes and expectations. On the basis of both cross-sectional and longitudinal
data, Eccles rejects deficit models of female achievement patterns and points to
the importance of subjective task value as a critical predictor of both male and
female choices in achievement domains.
Much of this recent work on achievement motivation and behavior offers
considerable promise for advances in understanding . Analysis of the multiple
components of achievement motivation, longitudinal studies , a focus on choice
and attainment in academic and career settings, and a concern with process
all of these strategies represent improvements to models that relied on static
states and sex-linked deficits . Further, as the work of Eccles and others
demonstrates, models that allow for a test of possible sex differences without
postulating different mechanisms may lead to a better understanding of sex and
gender as well as of the substantive processes of concern.
MASCULINITY AND FEMININITY
Garnering a great deal of attention during
the past decade have been the concepts of masculinity and femininity . Constan
tinople's ( 1 973) critique of traditional approaches to this topic, with their
assumption of a unidimensional bipolar scale of femininity-masculinity,
proved to be a watershed for research in this area. In quick succession, a
number of investigators developed separate scales to assess masculine and
feminine characteristics and demonstrated that the two scales are orthogonal to
one another (Bern 1 974, Spence et al 1974, Heilbrun 1 976, Berzins et aI 1 978).
The greatest excitement was generated by Bern's ( 1 974) introduction of the
concept of androgyny, used to refer to those men and women who possess both
masculine and feminine qualities in relatively equal proportion.
Although most investigators would now agree on the separate and orthogonal
nature of the two dimensions, there has been considerable controversy as to
what the measures assess and what significance various combinations of scores
on the two measures have. Whereas proponents hailed the androgyny concept
as a significant move away from traditional emphases on sex differences ( and
more than occasionally, as a prescription for social change), critics claimed
flaws in both methodology and conceptualization. The full scope of this debate
is too extensive to cover here , and interested readers can consult some of the
following sources: Kelly & Worell 1977, Sampson 1 977, Bern 1 979, Lenney
59
1979, Locksley & Colten 1979, Pedhazur & Tetenbaum 1979, Spence &
Helmreich 1979, Lou 1981, Myers & Gonda 1982, Lubinski et al 1983.
At this time, the most appropriate assessment of this area of investigation is
conservative. Masculinity and femininity, it is increasingly recognized, are
broad multidimensional concepts that are not readily captured in either one- or
two-dimensional questionnaires (Constantinople 1973, Spence 1983b, 1984,
Deaux 1984b, Feather 1984). The scales on the most commonly used instru
ments, the BSRI (Bern 1974) and the PAQ (Spence et aI1974), are best viewed
as measures of dominance and self-assertion on the one hand and nurturance
and interpersonal warmth on the other. (Agency and communion, terms pro
posed by Bakan 1966, are equally appropriate as descriptors.) As such, these
measures show good predictability for behaviors that require either assertive or
nurturant behavior, respectively. Less convincing evidence has been offered to
support the assumption that these trait measures are substantially related to
other classes of gender-related attributes and behaviors associated with the
broader concepts of masculinity and femininity (Spence & Helmreich 1980).
Nor is there evidence that androgyny, as a specific combination of scores on the
two measures (most typically, high scores on both scales arbitrarily defined by
the group median), has any unique predictive power (Taylor & Hall 1982,
Lubinsky et al 1983). In short, although the visibility of androgyny has been
great and the concept has undoubtedly been influential in thought, its empirical
status is questionable and it shows little hope of becoming the panacea that
many anticipated.
Whereas the proponents of androgyny attempted to
MORAL DEVELOPMENT
minimize the significance of sex differences, arguing that measured masculin
ity and femininity obviate the importance of biological sex, the work of
Gilligan (1982) on moral development has given increased attention to the
position that males and females differ in significant ways. Using Kohlberg's
cognitive-developmental approach to moral development but criticizing its
masculine bias, Gilligan argues that there are two distinctly different sequences
and principles of moral development. One, identified in Kohlberg's earlier
research, emphasizes separation and individuation, while the other is con
cerned with attachment and caring. Although not claiming that these patterns of
moral belief are sexually dimorphic, Gilligan for the most part describes the
two in terms of a masculine and feminine voice, respectively. Differences in
both the antecedents and the consequences of these two styles are believed to be
substantial, in an argument that is consistent with the theorizing of Chodorow
(1978). Support for this theory is provided primarily by the descriptive pro
tocols of women and men who responded to moral dilemmas (Gilligan 1982).
Additional evidence consistent with the thesis is reported by Pollak & Gilligan
(1982), who find that men show more violence imagery in response to affilia-
60
DEAUX
tion themes, while achievement situations are more likely to elicit violence
imagery in women . The latter results are being debated (Benton et al 1983,
Pollak & Gilligan 1983, Weiner et aI 1983).
Other recent work in moral development suggests that sex differences may
be more subtle. Lifton (1984), for example, finds no sex differences in level of
moral reasoning, using three different theoretical approaches, but notes that the
patterns of interrelationships among personality variables differ for men and
women within the same morality level . Work on moral development will
undoubtedly continue, prompted by Gilligan's broad-ranging thesis. Among
the challenges for investigators in this area is a linkage between moral reason
ing and moral behavior, an issue that has long concerned personality theorists
working on this problem.
Social Behaviors
As in the case of personality dispositions, the range of social behaviors
explored for possible evidence of sex differences has been broader than the
present coverage allows . Selected for discussion, primarily on the basis of
activity level, are aggression, social influence, and nonverbal behavior.
AGGRESSION
Aggressive behavior, the only social behavior identified by
Maccoby & Jacklin (1974) as furnishing clear evidence of a sex difference, has
continued to be analyzed during the past decade . The issue has not been
resolved to everyone' s satisfaction. In a reanalysis of the Maccoby-Jacklin data
base, Hyde (1982) found that sex differences in aggression accounted for
approximately 5% of the variance. Frodi et al (1977) , taking a more careful
look at situational variations, reported that sex differences are characteristic of
only certain types of situations. They are less likely to be found when aggres
sion is in response to provocation, for example, than when it is initiated by the
experimental subject. It has also been suggested that different types of situa
tions may elicit anger in women and men, leading to situationally dependent
patterns of aggression (Frodi et al 1977, Towson & Zanna 1982).
In counterpoint to these analyses of laboratory studies are a bevy of statistics
suggesting much greater aggression by males than by females in the society.
Crime rates for both homicide and rape, for example, show a consistent and
wide differential between women and men. Studies of family violence testify to
the aggressive behavior of both women and men, although the target of the
attacks often varies (cf Breines & Gordon 1983) . The disparity between
findings obtained in the laboratory versus the field is perhaps more marked in
this area of research than in many others, and results from the latter could do
much to inform the former type of investigation .
Of more dispute than the existence of sex differences is the possible cause of
such differences. Tieger (1980) contested the Maccoby-J acklin suggestion of a
61
NONVERBAL BEHAVIOR
Sex differences in nonverbal behavior have been
observed in nearly every area studied-personal space, touching, gaze, and
posture, to mention only a few. The superiority of women in both encoding and
decoding nonverbal cues has also been documented (Hall 1977, 1978, 1979),
although explained variance is again relatively small. Female superiority in
encoding is particularly marked in the case of visual cues and less marked for
vocal expressions (Hall 1979). As decoders , women show less advantage when
cues are presented very briefly or when the cues are transmitted through the
"leakier" channels of tone or body position as opposed to facial expression.
Rosenthal & DePaulo (1979) explain the latter finding in terms of an interper
sonal accommodation hypothesis, suggesting that women are more polite in
their nonverbal interactions, willing to ignore more revealing cues for the sake
of interpersonal harmony.
Other explanations for sex differences in nonverbal communicative skills
have included both differential attention (motivated by a need for the less
powerful group to be able to predict the behavior of the more powerful group)
as well as biological predispositions that may have evolutionary significance
(Hall 1978, 1979; Hall & Halberstadt 1981). Within many areas of nonverbal
behavior, the parallels between gender and status have been considered (Hen
ley 1977).
Areas of social behavior in which sex differences could be explored are
limited only by the ingenuity of investigators and the range of human behavior.
Many topics are being ignored in this review, several of which show promise
62
DEAUX
for future developments in theory. Among areas that have been reviewed
recently by others are leadership (Hollander 1985), reward allocation (Major &
Deaux 1982), causal attribution (Deaux 1976, Frieze et al 1982), helping
behavior (Piliavin & Unger 1984), and group interaction (Dion 1984). In each
case, careful analysis has suggested the substantial influence of situational
factors and the need to interpret the effects of sex and gender within a broader
theoretical network.
Central Issues
As a general statement, one might say that evidence of clear and consistent
differences is less prominent in the case of social behaviors than it is when
differences in cognitive skills are considered. In the former, one immediately
confronts the potent influence of situational variables, an influence that readily
alters behaviors and makes simple main effect conclusions much less likely
than when the focus is on the single individual acting alone, reacting to
cognitive tasks. This greater variability may act as a source of frustration or as
an important cue to the investigator. Eschewing frustration and beliefs that sex
and gender can never be cleanly depicted, the investigator might do well to
acknowledge the variability and ponder just what this malleability of behavior
signifies.
There are at least three issues that should be considered when evaluating the
research on sex comparisons. Briefly stated, these three issues are: (a) the
meaningfulness of observed differences, (b) the distinction between findings
obtained in laboratory and field studies, and (c) the possible causes of observed
sex differences (cf also Deaux 1984a, Jacklin 1981).
Meta-analytic techniques provide a
way of calculating the magnitude of difference when a number of diverse
studies are combined, obtaining, in effect, an average of the observed differ
ences. Application of such techniques to studies of sex differences has typically
yielded relatively low estimates of effect size, generally accounting for less
than 5% of the variance. Such analyses are a useful corrective to the somewhat
insidious tendency to assume that a reported sex difference implies a bimodal
distribution. Further caution is suggested by evidence that these sex differences
are not necessarily constant over time, as the Rosenthal & Rubin (1982)
analysis of cognitive skills has indicated. At the same time, it should be
recognized that even when mean differences between the sexes are small,
differences in the proportions of males and females occurring at the extremes of
the distribution or even the percentages above the median may be quite
substantial and thus have some practical significance (Rosenthal & Rubin
1982, Huston 1983).
There are, of course, some difficulties in combining studies that vary widely
in quality, procedures, settings, and a number of other characteristics. Many of
MEANINGFULNESS OF DIFFERENCES
63
the meta-analytic studies have attempted to account for some of these potential
ly confounding variables, and often investigators have been able to point to
significant variables other than sex of subject. Eagly et al ( 1983), for example,
suggest that the particular task is critical in predicting sex effects in prosocial
behavior. Among other task attributes, they find that situations which are
perceived to involve greater danger for females than males are most likely to
elicit sex differences favoring men, while less threatening situations are more
apt to show sex similarities. Investigators who have used less quantitative
methods of summarizing studies have also pointed to the importance of task
characteristics in areas as diverse as aggression (Frodi et al 1977), reward
allocation (Major & Deaux 1982), and self-confidence and expectations for
performance (Deaux 1976, Lenney 1977).
Interpretation of the importance of sex differences in behavior is also
tempered by some reports of a relationship between the sex of the investigator
and the direction of findings. Thus Eagly & Carli ( 1981) report that in their
analysis of the social influence literature, as well as in a reanalysis of Hall's
( 1978) review of nonverbal decoding, sex of author accounts for some variance
in the findings. Specifically, both sexes are apparently more likely to report sex
differences that are favorable to their own sex, males more likely to report male
independence and females more likely to report female superiority in nonverbal
decoding. Carli ( 1983) has observed a similar effect in studies of group
behavior and reward allocation. The generality of this pattern is as yet un
known, but it clearly poses a problem for interpretation of specific sex differ
ences and a question for more general issues of scientific procedure.
Evidence from many
laboratory studies suggests that sex differences, although not insignificant, are
relatively small. Such evidence has led many to conclude that recognition of
sex similarities should outweigh an emphasis on sex differences. Such conclu
sions are often met with disbelief among those whose reference point is outside
of the psychological laboratory, where differences between women and men
are immediately apparent in such areas as occupation, household division of
labor, child-caring responsibilities, and the widely heralded "gender gap" in
political opinion.
The differences between these two settings in the kinds of information that
they can provide need to be more clearly recognized. Within the laboratory, it is
possible to vary the degree to which gender is salient. Under some conditions,
the role of subject may be most salient and one can conduct a relatively
"gender-free" test of sex differences in capability-the ability to perform
specific tasks selected by the experimenter (cf also Eagly 1978, Unger 1981).
For the most part, this assessment indicates limited sex differences in capability
for a wide range of tasks. Alternatively, it is possible to invoke gender
considerations quite strongly in the lab through, for example, choice of task,
COMPARISON OF LABORATORY AND FIELD STUDIES
64
DEAUX
65
66
DEAUX
67
categories . From this vantage point, questions have been raised not only about
the content of gender stereotypes (what might be termed the "old look"), but
also about the structure of these categories and the processes by which they
operate.
The reliance on trait characteristics to define gender stereotypes, reflected in
the original Rosenkrantz et al ( 1968) methodology and updated by Spence et al
( 1974), has been questioned in recent years (Ashmore & Del Boca 1979,
1984b; Deaux & Lewis 1983). Deaux & Lewis ( 1983, 1984) have identified a
set of components, including role behaviors, physical characteristics, and
occupation in addition to the more typically used traits that are associated with
gender stereotypes. These components, although bearing some relationship to
one another, nonetheless function with some independence, suggesting a more
complex composition than was previously considered. Such findings are con
sistent with the suggestion of Ashmore & Del Boca ( 1979) that there are various
kinds of attributes associated with gender, including genital and biological
features (defining attributes), physical appearance cues (identifying attributes),
and the more commonly used trait descriptors (ascribed attributes) .
Cross-cultural studies o f gender stereotypes, although less concerned with
the more detailed analysis of various components, have nonetheless provided
valuable information as to both the diversity and the generality of gender
stereotypes. In a study of 30 nations, Williams & Best ( 1982) found a consider
able degree of "pancultural generality," evidenced most clearly in the associa
tion of instrumental traits with males and expressive traits with females. At the
same time, some variations were evident when comparisons were made be
tween groups that differed in religious tradition or in national work-related
values. In more descriptive but less systematic explorations of other cultures,
anthropologists have also found evidence for both the commonality of in
strumentality and expressiveness, as well as for unique attributes that reflect
specific cultural contexts (e. g. Strathern 1 976, Dwyer 1978).
More analytic work on the structure of gender stereotypes has addressed two
questions: (a) the interrelationship among various components and their influ
ence relative to simple sex label, and (b) the level of categorization at which
gender stereotypy occurs. With regard to the first question, it is evident that
there are a variety of complex inference processes that operate across various
components , although the picture is not yet crystalline. Information about role
occupancy can influence the ascription of stereotypic traits, with people identi
fied as in positions of higher influence being accorded more instrumentality,
regardless of their sex (Eagly & Wood 1982, Eagly 1983, Eagly & Steffen
1984). The Deaux & Lewis ( 1984) results suggest that physical characteristics
may have some priority as an influence on other inference processes, a sugges
tion that is consistent with work by social psychologists indicating the general
importance of physical appearance cues (McArthur 1982).
68
DEAUX
It is also becoming apparent that people tend to view men and women in
tenns of opposites, despite the empirical overlap that they accord to various
components of the stereotypes. Thus just as early investigators of self-assessed
masculinity and femininity assumed a bipolar model, so do observers tend to
assume that what is male is "not female" and vice versa (Foushee et al 1979 ,
Major et al 1 98 1 , Deaux & Lewis 1984).
A debate has also surfaced as to the relative influence of gender label in the
face of other more diagnostic information (Locksley et al 1980; Grant &
Holmes 1981, 1982; Locksley et al 1982; Deaux & Lewis 1984). Ex
perimentally , one can negate the influence of gender label by providing
infonnation that is closely linked to the judgment being requested, thus sug
gesting that the influence of stereotypic beliefs is quite weak. However, the
network of associations linked together under the umbrella of gender
stereotypes has been shown to influence judgments even when the categorical
label loses its potency. [For example, information about gender-related physi
cal characteristics will influence judgments about traits and role behaviors not
necessarily related to those characteristics (cf Deaux & Lewis 1984).]
Recent work in cognition, most notably that of Rosch and her colleagues
(Rosch 1978, Mervis & Rosch 1 9 8 1 ) , has also prodded investigators to look
beyond the general labels of male and female to search for more specific gender
SUbtypes that may be part of stereotypic thought. In a pre-Roschian investiga
tion, Clifton, McGrath & Wick ( 1976) noted the existence of "housewife" and
"bunny" as distinct female stereotypes. Other investigators have also sought to
identify these more specific categories (Ashmore 1 981, Deaux et al 1984).
Although it is far from clear that gender stereotypes can be ordered in a neat
hierarchy in the way that some natural objects line up (cf Lingle et al 1984),
consideration of the diversity of stereotypes may yet prove useful. One possi
bility suggested by Deaux et al ( 1984) is that the range of female stereotypes is
much broader than that of males , showing greater distinctiveness and covering
a broader range of attributes .
Yet another line of investigation being pursued in the recent assault on
gender stereotypes is the question of individual differences. Whereas earlier
concerns with questions of content and assumptions of shared conceptions
precluded an analysis of individual variation, the recent social cognition
approach is far more amenable to consideration of this variety. Bern's ( 1981)
proposal of gender schema theory assumes that individuals differ in their
tendency to use gender as an organizing principle. According to this view,
people who are highly sex-typed (masculine males and feminine females , as
categorized by scores on the Bern Sex Role Inventory) are more likely to
process information and to make distinctions on the basis of gender than are
other people. [The relationship between androgeny and gender schema is
treated more explicitly by Bern ( 1984).] The conceptual basis of this formula-
69
tion has been challenged and its status is somewhat shaky (Spence & Helmreich
1 98 1 , Bern 198 1 , 1 982, Markus et a1 1 982, Crane & Markus 1 982, Mills &
Tyrrell 1 983). Nontheless, the view that individuals differ in their tendency to
stereotype is probably acceptable to most and is being studied by a number of
investigators (Spence et a1 1 975 , Deaux et aI 1 984). These latter investigations
suggest little relationship between endorsement of gender stereotypes and
self-described masculinity (agency) and femininity (communion) .
Research on gender stereotyping has moved from the simple demonstrations
of discrimination and unequal judgment to a more sophisticated analysis of the
components of gender belief systems and the processes by which information is
encoded, stored, and retrieved. It can be predicted that this area will generate
continued activity within the next decade, pursuing more sophisticated ques
tions about the structure of gender stereotypes (and, by extension, other
stereotypic structures) as well as posing new questions about the role of affect,
the sources of stereotypes, and the influence of stereotypes on subsequent
behavior.
70
DEAUX
Marital Roles
A functionalist view of division of labor within the household is most widely
associated with the work of Parsons & Bales ( 1955). Analyses of research
related to this thesis are available elsewhere (Miller & Garrison 1 982, Spence et
al 1 985). Within recent psychological research, perhaps the greatest emphasis
has been placed on the dual-career couple, exploring the ways in which this
state alters traditional division of labor within the household and the conflicts
that may arise between career and family roles (e.g. Bryson & Bryson 1 978,
Hall & Hall 1 979). Most often these analyses have focused on the middle-class
career couple, leaving many questions remaining with regard to the more
numerous dual-worker couples in the society. Nonetheless, this research makes
some contribution to an understanding of some of the broader factors that relate
to sex and gender, although its contribution to date has been somewhat limited
with respect to the more fundamental issues of sex and gender.
More general work has concerned the relationship between the husband-wife
balance of power and marital satisfaction. A recent review of this literature
suggests that marital satisfaction is lowest when the wife is dominant and
highest in egalitarian couples (Gray-Little & Burks 1 983). Although the use of
coercive control appears to be most closely associated with marital dissatisfac
tion , more research on the dynamics of the exchange process within the marital
relationship is still needed.
Theoretical advances in understanding marital roles have been relatively
few , and the research has typically not been linked to other areas of gender
related research. A promising corrective to this state, however, can be seen in
the recent work of Kelley and his colleagues (Kelley 1 979 , Kelley et aI 1 983) .
These investigators offer an ambitious and generative conceptual framework
for analyzing close relationships, incorporating both personal and social en
vironmental factors in a process-oriented analysis of dyadic interaction.
Although not limited to issues of gender, this theory may prove to be an
important framework for exploring such issues.
Occupational Structure
The increasing trends in female employment have prompted a surge of research
related to women's place in the occupational hierarchy. Apart from the eco
nomic statistics documenting the changing composition of the work force, there
are many accounts of the performance of women in various occupations and a
comparison of men and women on several dimensions believed to be related to
work performance and occupational choice. (For a summary of much of the
psychological literature, see Nieva & Gutek 1 98 1 ). The lack of comparability
71
of the typical male and female job has hampered many of these comparisons. In
fact, recent research suggests that occupational segregation is even more severe
than had previously been believed. In a comprehensive study of over 400
California work organizations, Bielby & Baron ( 1984) report that 59% of the
companies were perfectly segregated by sex, i.e. there was no overlap between
the job titles filled by one sex and those filled by the other. Fewer than 10% of
the workforce was in specific jobs that included both sexes. This uneven
distribution within the employment realm has important consequences, not
only for issues of wage and equity, typified in recent debates about comparable
worth, but also for the psychological processes that are activated by dispropor
tionate sex ratios.
Sex Ratios
Kanter ( 1977), in dissecting the structure of corporate organizations , illumi
nated the relevance of proportions to the study of sex and gender. In groups in
which one of two groups is in a decided majority, members of the minority
group can be considered "tokens . " Both the behavior of the token members and
the behavior of the majority members toward the tokens can be shown to be
channeled in systematic ways, independent of the particular identifying charac
teristics (e . g. sex, race) of the two groups. The consequences of this token
status have been explored by Kanter and others (Taylor & Fiske 1 978, McAr
thur 1981) and support the contention that both the behaviors of and the beliefs
about males and females may in some cases be a specific example of a more
general process that operates when group members are in unequal proportions.
Gutek & Morasch ( 1982) have also used a sex-ratio approach in their analysis
of sexual harassment in the workplace.
The potential influence of sex ratios on behavior has been considered further
in the work of Guttentag and Secord (Guttentag & Secord 1983; Secord 1982,
1983). Applying a social exchange perspective to the analysis of dyadic
relationships between men and women, these authors contend that the ratio of
males to females in a society affects the power balance between the sexes.
When the ratio of males to females is high, they predict that women are more
likely to assume traditional roles of wife and mother and the division of labor
would be strong. In contrast, when women are in "oversupply," Guttentag and
Secord see the likelihood of greater sexual libertarianism, lower values on
marriage and family, and an increased expression of feminist ideology. Focus
ing their analysis on age cohorts and implied marital opportunities, these
investigators note striking changes in the United States sex ratio from the 1960s
to the 1970s. Such changes, they argue, have important implications for
structural power and social exchange. Although the sex ratio must be consider
ed only one factor influencing male and female behavior, and perhaps is most
applicable within the marital system, the work of Guttentag and Secord is
72
DEAUX
73
suggests that there is much to be learned from a more careful analysis of the
process of gender presentation .
Goffman's ( 1 976) suggestive notions of gender enactment, emphasizing the
element of choice in people' s portrayal of gender, have found some confirma
tion in the literature, but the work has generally been scattered and unsystem
atic . Yet although this particular approach has been mined more shallowly than
some of the traditional perspectives, it demonstrates a distinctly psychological
contribution to an understanding of sex and gender while also giving more
attention to process rather than static elements .
Some investigators have chosen to pit internal dispositions against situation
al factors in exploring the fluidity of sex differences. For example, it has been
shown that fear of success is less predictive of performance than the attitudes of
a partner toward accomplishment (Jellison et al 1 975). Assessed measures of
dominance predict leadership assumption for women paired with other women
but not paired with men, at least when the task is masculine in its orientation
(Megargee 1969, Carbonell 1 984) . The influence of normative expectations on
the display of dominance behavior has also been shown by Klein & Willerman
( 1 979) , although in this case the dispositional tendency continued to exert an
effect. Women instructed to be as dominant as possible showed no differences
in their behavior toward a male or female partner, whereas the absence of
instructions resulted in lower displays of dominance toward a male partner.
Each of these studies attests to the flexibility that individuals have in displaying
behaviors that are either consistent with an internal disposition or are consistent
with the expectations (either implicit or explicit) of others (cf also Deaux 1977,
1 984a). It is also interesting to note that women have more often been the
subject of these investigations, perhaps reflecting an assumption that women
are more likely to be responsive to situational variations . The validity of this
assumption remains to be demonstrated. In fact, individuals of both sexes have
been shown to be highly responsive to the instrumental or expressive demands
of a situation, shaping their behavior in corresponding ways (La France &
Carmen 1 980, Putnam & McCallister 1 980).
Other investigations, while less concerned with specific personality disposi
tions, have also implicated situational factors as a determinant of gender
related behavior. Zanna and his colleagues have demonstrated that a woman' s
beliefs about her male partner' s attitudes toward women will lead to alterations
in self-presentation, both verbal and physical (Zanna & Pack 1 975, von Baeyer
et al 198 1 ) . Still other evidence testifies to the chain of events that may shape
gender behavior, wherein expectations of one person can alter the behavior of
another in the direction of stereotypically gender-related choices (Skrypnek &
Snyder 1 982). General models of expectancy confirmation sequences , such as
the one presented by Darley & Fazio ( 1 980), may prove valuable in further
elucidating the parameters of this process as it relates to gender.
74
DEAUX
CONCLUDING THOUGHTS
Sex and gender is an area of research whose time has come. The questions, of
course , have always existed , but until relatively recently the answers were
more likely to be offered by social commentators than by scientists . The very
obviousness and assumed natural order of males and females, while producing
a surplus of pronouncements on the way things were, may have dulled the
sensitivities of scientists to many basic issues. Activity within the past decade
or two has surely provided ample evidence that unanswered questions remain.
As an area of research, sex and gender is fraught with dilemmas and decision
points . The interface of ideology with the scientific enterprise, long a topic of
debate for social science research in general, supplies a tension that pervades
the area. What one may wish as a feminist is not necessarily what one sees as a
scientist. Yet a clear view of what is, and an understanding of why, become
underpinnings for the work of those who seek change. There is, in fact, a
perceptible shift in the ideological pendulum in recent years . Attempts to
"disprove" the existence of sex differences have given way to arguments, both
at the scientific and popular level , that differences do exist (cf Chodorow ,
Gilligan). Acknow ledgment of the existence of differences should not, howev
er, serve as a cap on efforts to understand the processes by which sex and
gender become influential in human behavior.
Psychology is not alone in its concern with questions of gender. Historians,
anthropologists , sociologists , biologists, and evolutionary theorists all have
knowledge and insights to offer to analyses of the questions. Informed by the
perspectives of these other disciplines, psychologists too have much to offer.
They also have much to gain. The issue of sex and gender is fundamental to an
understanding of human behavior, and increased consideration of its parame
ters and implications can only benefit psychology as a whole. From my own
perspective, the best is yet to come and I look forward to the next Annual
Review chapter that will document that progress .
ACKNOWLEDGMENTS
This chapter was prepared while I was a Fellow at the Center for Advanced
Study in the Behavioral Sciences. I am grateful for financial support provided
by the John D. and Catherine T. McArthur Foundation and by a grant from
the National Science Foundation (BNS-82 1 73 1 3) . The insights and sugges
tions of the following people, commenting on an earlier draft, are recog
nized with appreciation: Denise Bielby, William Bielby, Alice Eagly, Sheila
Kamerman , Marcia Linn, Eleanor Maccob y , Brenda Major, Charles
McClintock, Ellen Messer, Paul Rosenkrantz, Janet Spence, Rhoda Unger,
and Michelle Wittig.
75
Literature Cited
Albin, R. S. 1 977. Psychological studies of
rape. Signs 3 :423-35
Archer, D . , Iritani, B . , Kimes, D. D . , Barrios,
M. 1 983. Face-ism: Five studies of sex dif
ferences in facial prominence. J. Pers. Soc.
Psychol. 45:725-35
Ashmore, R. D. 1 98 1 . Sex stereotypes and
implicit personality theory. See Hamilton
1 98 1 , pp. 37-8 1
Ashmore, R. D . , Del Boca, F. K. 1979. Sex
stereotypes and implicit personality theory:
Toward a cognitive-social psychological
conception. Sex Roles 5:219-48
Ashmore, R. D . , Del Boca, F. K . , eds. 1984a.
The Social Psychology of Female-Male Re
lations: A Critical Analysis of Central Con
cepts. New York: Academic. In press
Ashmore, R. D . , Del Boca, F. K. 1 984b. Gen
der stereotypes. See Ashmore & Del Boca
1 984a
pp.
76
DEAUX
Cooper,
H. M.
1979.
Statistically combining
77
78
DEAUX
Roles 5:703- 1 9
chol. 4 1 :988- 1 00 1
and the
psychology of achievement: Implications for
personal and social change. In Women in the
Workforce, ed. J . Bernardin, pp. 48-69.
New York: Praeger. 242 pp.
Mednick, M. T. S . , Weissman, H . J . 1 975 .
The psychology of women-Selected topics.
Ann. Rev. Psychol. 26: 1 - 1 8
oriented
79
1 05
press
Parlee, M. B . 1 97 5 . Psychology. Signs 1 : 1 1 938
Parlee, M.
Free Press
Pedhazur, E. J . , Tetenbaum. T. J . 1 979. Bern
Sex Role Inventory: A theoretical and
methodological critique. J. Pers. Soc.
Psycho I. 37:996- 1 0 1 6
Peters, L. H . , Terborg , J . R . , Taynor, J . 1 974.
Women as managers scale: A measure of
attitudes toward women in management
positions. JSAS Cat. Sel. Doc. Psychol. 4:27
80
DEAUX
5 : 1 47-63
Spence, J . T . , Helmreich, R . L. 1 98 1 .
Androgyny versus gender schema: A com
ment on Bern's gender schema theory.
Psychol. Rev. 88:365-68
Spence, J. T . , Helmreich, R. L. 1983.
Achievement-related motives and behaviors.
See Spence 1983a, pp. 7-74
Spence , 1. T . , Helmreich, R . , Stapp,
1. 1974.
Unger, R . K. 1 98 1 .
81