Sex, Schemas, and Success
What’s Keeping Women Back?
By Virginia Valian
Virginia Valian is a professor of psychology and linguistics at the City
University of New York’s Hunter College and CUNY’s Graduate Center.
This article, which appeared in a slightly longer form in the September-
October, 1998 issue of Academe (Journal of the American Association
of University Professors) and is reprinted here with permission of the
author and journal, is based on her most recent book, Why So Slow?
The Advancement of Women (MIT Press).
The term glass ceiling has become popular
as a way of referring to women’s lack of
representation at the top levels of organizations.
The term suggests that invisible factors — as much as or more so than overt discrimination — keep women from rising to the top.
It also suggests that those hidden factors
will probably not simply disappear with
time. It implies, moreover, that women’s
performance is at least equal to that of
their male peers, for a ceiling keeps people
down despite their competence.
Although disputes about the implications
of the glass ceiling continue, solid
data from social and cognitive psychology, sociology,
and economics show that men and women
receive unequal returns for equal investments.
More important, the evidence reveals the perceptions
and practices that create and maintain
inequality. To move forward, we must understand
how our cognitive processes unconsciously distort
our judgments about men and women and
thereby perpetuate the inequities that we have
long been trying to overcome. Such an understanding
will allow us to determine remedies for
the present impasse, ranging from affirmative
action to better methods for evaluating job applicants
Discrepancies between men’s and women’s
salaries occur both in business and in academia.
In 1991 economists Mary Lou Egan and Marc
Bendick conducted a survey of U.S. professionals
working in occupations with an international
focus. The men and women in the study resembled
one another in many ways, such as years of
work experience, range of specialties, and hours
worked each week. But factors that benefited
men did not help women to the same degree.
Women’s achievements and qualifications
appeared to be worth less than men’s.
For example, a bachelor’s degree contributed
$28,000 to a man’s annual salary but only
$9,000 to a woman’s. And not constraining
one’s career for one’s spouse added $21,900 to
men’s yearly income but only $1,700 to
women’s. Some factors that added to men’s
salaries subtracted from women’s. Having lived
outside the United States added $9,200 a year
for men but subtracted $7,700 for women.
Speaking a language other than English added
$2,600 for men but cost women $5,100. Of the
seventeen factors Egan and Bendick examined,
fourteen helped men more than women (see
table, next page).
This study is typical of others in the literature.
Women tend to benefit less from their
qualifications than men do. In many cases,
women’s human capital — their training, years
of job experience, and so on — is less than
men’s. But even when men and women are
equal in human capital, or when their differences
are statistically equalized, men get more
from their investments than women do.
Men’s advantage above and beyond their
greater human capital is often termed discrimination.
Those who argue that the residual unexplained
disparity between men and women is
evidence of discrimination have been criticized
for incorrectly assuming that all relevant factors
have been measured and that the single variable
of discrimination accounts for the remaining
unexplained differences. Thus, the criticism continues,
discrimination could appear to be taking
place only because of a failure to specify all the
relevant sex differences.
Others have made the opposite criticism,
arguing that some economic studies have erred
by including variables that may themselves be
the consequence of discrimination. For example,
lesser work skills may be the result of less
opportunity to acquire skills. While both criticisms
point out potential pitfalls, the studies to
date appear neither to overlook major factors
contributing to disparity nor systematically to
under- or overestimate discrimination.
In academia men and women now start out
with equal salaries, but they do not progress at
the same pace. Data from the National Science
Foundation (NSF) for 1993 showed that fulltime
academic male and female scientists were
close to parity in their salaries one to two years
after they received their Ph.D.s. But three to eight years after completing the Ph.D., women
earned 92 percent of men’s salaries, and at nine
to thirteen years afterward, women earned only
90 percent of their male counterparts’ salaries
(NSF, unpublished data). Similar data hold for
male and female humanists.
In medicine, as well, the pattern is early parity
followed by a gap. Income data for 1990 for
physicians under forty-five years of age with
two to five years of experience showed equal
earnings (once human capital differences were
controlled for). But for physicians with six to
nine years of experience, women earned 96 percent
as much as men. The story is the same in
field after field; initial salaries are now close to
equal for similarly trained young men and
women. But disparities develop quickly.
The data reveal that women rise too slowly
through the professions, and their credentials
appear to be worth less than men’s. To understand
why that is so, I developed an explanation that
relies on two key concepts: gender schemas and
the accumulation of advantage. Our unarticulated
beliefs about men and women — gender schemas— make it harder for women (and easier for men)
to accumulate advantage and rise to the top.
Schemas are hypotheses that we use to interpret
social events. A schema resembles a stereotype,
but is more inclusive and neutral. Gender
schemas are hypotheses that we all share, men
and women alike, about what it means to be
male or female. Schemas assign different psychological
traits to males and females. We see boys
and men as capable of independent action, as
agents; they are task-oriented and instrumental.
We see girls and women as nurturing, communal,
and expressive. In brief, men act; women
feel and express their feelings.
Women have more trouble than men in
reaching the top because our gender schemas
skew our perceptions and evaluations, causing
us to overrate men and underrate women.
Experimental and theoretical work in social and
cognitive psychology and sociology supports this
thesis. People are not perceived as people but as
males or females. Once gender schemas are
invoked, they work to the disadvantage of
women and the advantage of men by directing
and skewing perception.
Laboratory experiments that control for
variables that might affect people’s judgments
have illustrated how gender schemas operate.
The findings from such experiments complement
the statistical data offered in the preceding
paragraphs. Despite their artificiality, the experiments
allow us to isolate the factors that
account for the lag in women’s achievements.
Take, for example, a laboratory study conducted
by New York University psychologist
that asked different
groups of students
to evaluate a
for a managerial
job. The number
of other women
candidates in a
pool of eight people
varied for each
group of student
one group, the
female applicant was the only woman; for others,
she was one of two women, one of three, one of
four, or one of eight.
Composition of pool
When women made up 25 percent or less of
the applicant pool, the female candidate was evaluated more negatively than when women
made up 37.5 percent or more of the pool.
Being in a small minority made a female applicant
appear less qualified and less worth hiring.
Even more interesting were the assessments of
the woman’s personality. When women made up
25 percent or less of the applicant pool, the student
judges perceived the female applicant as
more stereotypically feminine — unambitious,
emotional, indecisive, and soft — than when
women accounted for 37.5 percent or more of
Such skewed perceptions pervade every evaluation
of men and women. Even for objective
characteristics such as height, people do not perceive
males and females accurately. In a compelling
laboratory experiment by University of
Kansas psychologist Monica Biernat and her colleagues,
college students were shown photographs
of other students and were asked to estimate
their height in feet and inches. The photos
always contained a reference item, such as a
desk or a doorway, so that height could be accurately
The experimenters matched the photographs
so that for every photograph of a man of a given
height, there was a woman of the same height.
Here, then, was an easily visible
characteristic to be measured in“objective” units. One might have
expected accurate evaluations. But
the evaluators’ knowledge that
men are on average taller than
women affected their judgment.
When exposed to a sample contrary
to the general rule, they perceived
the women as shorter and
the men as taller than they really
were. In this experiment, as is typical,
there were no differences in
how male and female observers
This experiment and others
suggest that if someone has a
schema about sex differences, that
schema affects the person’s judgments.
Observers perceive individuals
who diverge from schemas in light of their
own gender hypotheses. If the schema is accurate,
as it is for height differences, that will exacerbate
errors made about individuals. Evaluators
tend not to question their judgment, because it is
supported by a real overall difference.
The implication of schemas for judgments of
professional competence are clear. Evaluators
may be faced with men and women who are
matched on the qualities relevant to success. The
evaluators may sincerely believe that they are
judging the candidates objectively. Yet they are
likely to overestimate the men’s qualifications
and underestimate the women’s because of
schemas that represent men as more capable
Take, for example, data from a study of
postdoctoral fellowships awarded by the
Swedish Medical Research Council in 1995.
Women made up 46 percent of the applicant
pool but only 20 percent of the winners,
because panels of senior scientists rated women
as inferior to men in scientific competence. A
subsequent analysis used an “impact” index to
rate the candidates’ productivity and the prestige
of the journals in which they published. This
analysis showed that women with a hundred or
more impact points had been rated by the original
panels as equal in scientific competence to
men with twenty points. The evaluators no
doubt considered themselves to be objective and
impartial judges of scientific merit. But as these
and other findings on gender schemas suggest,
people tend to underestimate women and overestimate
men in ways ranging from height to
professional ability whenever they have antecedent
beliefs about sex differences — even when
those beliefs are unarticulated.
No credit for leadership
Gender schemas not only make it difficult for
women to be evaluated accurately; they also
make it difficult for women to reap the benefits
of their achievements and be recognized as leaders.
Consider a study in which college students
viewed slides of five people seated around a table
working together on a project. The students
were asked to identify the leader of the group. In
same-sex groups, the man or woman sitting at
the head of the table was always selected as the
leader. In mixed-sex groups, a man at the table’s
head was reliably picked out as the leader. But if
a woman sat at the head, she was not always
labeled as the leader; a man seated elsewhere was
chosen as the leader about as often. As in other
studies, there were no differences in the perceptions
of female and male participants.
Failing to label a woman seated at the head
of a table as a leader may have no discriminatory
impetus behind it. But a woman leader is nevertheless
prone to lose out compared with a man
in the same position, because she is less likely to
receive the automatic deference that subtle
marks of leadership confer on men. As a result,
the woman is objectively hurt even if observers
intend no hurt. She has to work harder to be
seen as a leader.
A real-life example from a prestigious university,
circa 1990, shows gender schemas in action. A new female faculty member in a science
department at a prestigious university has a
conference with her department chair about
what courses she will teach. She is eager to teach
a large introductory lecture course. The chair
refuses, saying that the students will not accept a
woman in that format. The woman presses a bit,
saying she thinks she can do it and would like to
try. The chair doesn’t want to take a chance and
instead gives the lecture course to a new male
We can ask two questions about his decision:
why does he make it and how does it affect the
woman’s future? The chair makes that decision
because gender schemas influence how he perceives
and evaluates the scientist. He sees her
not just as a scientist but as a female scientist. As
such, she is probably unable to handle a large
group of people. She lacks the authority a male
automatically has by virtue of his sex.
We might be tempted to dismiss the incident.
We might be tempted to tell the woman not to
make a mountain out of a molehill. But the
woman ends up teaching a laboratory course
that requires much more work, giving her less
time for research and publishing and putting her
at a disadvantage relative to her male colleague
who teaches the lecture course. She also has had
a small failure she didn’t deserve, giving her a
small psychological disadvantage, because she
has something to worry about that her male colleague
Accumulation of advantage
Although a single course assignment is a
small thing, small things add up. Success is
largely the accumulation of advantage, the parlaying
of small gains into larger ones. Mountains
are molehills, piled one on top of the other.
A computer simulation demonstrates how
the accumulation of advantage and disadvantage
can work. Psychologist Richard Martell and his
colleagues at Teachers College of Columbia
University created a model eight-level hierarchical
institution, staffed initially by equal numbers
of men and women. Their model assumed that
over time a certain percentage of incumbents
would be promoted from one level to the next.
It also assumed a tiny bias in favor of promoting
men, a bias accounting for only 1 percent of the
variability in promotion. The researchers ran the
simulation through a series of promotions. After
many runs, the highest level in the institution
was 65 percent male. In the long run, a molehill
of bias creates a mountain of disadvantage.
Our gender schemas cause us systematically
to overrate men and underrate women. Our
doing so culminates in lower salaries and slower
rates of promotion for women. Knowing how
these gender schemas work can help us understand
why women in fields such as international
business gain less advantage from their credentials
than their male colleagues do. When men
learn another language and live outside the
United States, they are seen as preparing for
their careers, engaging in those activities not
because they enjoy them but because they expect
an economic return. Women, in contrast, are
perceived as choosing such activities for pleasure.
Men accumulate advantage more easily than
women because men are seen as
more professional than women.
One school’s success story
Fortunately, the situation is not
hopeless. We can improve
women’s status in different ways,
institutionally and personally. The
Johns Hopkins University School
of Medicine has shown what can
be done to address the problem of
lower salaries and slower promotion
rates for women.
In 1990 the university’s
Department of Medicine had only
four women associate professors. A
task force found that women were
put up for promotion later than
their male peers, both because
evaluators failed to identify qualified
women and because women did not realize
what was required for promotion. Each female
faculty member (and later, each male faculty
member) began to receive annual evaluations
that gave her explicit information about her
progress. The women also obtained concrete
information in monthly meetings on how to
develop their careers and how to handle different
problems that would arise. On top of that,
senior faculty members learned how to mentor
their junior colleagues, so that disparities in
treatment between junior men and women could
The monthly meetings and mentoring
addressed serious problems in the department’s
treatment of junior faculty members. Mentors gave
junior men more guidance and help than they gave
junior women. For example, mentors invited junior
men to serve as chairs at conferences six times
as often as they invited junior women to do so.
The junior men thus received more public exposure
than their female colleagues.
Within five years, the program was a success.
By 1995, with no change in the criteria for promotion,
the department had twenty-six women
associate professors. More subtle aspects of the
women’s experience also improved. In 1990, 38 percent of the women said that the institution
welcomed them, while 74 percent of the men
said so. In 1993, 53 percent of the women felt
welcome — a dramatic improvement within a
short period of time, albeit one that fell short of
equity. The Johns Hopkins experience demonstrates
that institutions willing to dedicate
resources to improving the status of their female
employees can do so.
Affirmative action is another
institutional tool that can counteract
the effects of gender schemas.
Designed to guarantee representation
of women and minorities in the
work force according to their numbers
and qualifications, affirmative
action policies implicitly acknowledge
the social and psychological
realities that I have just described.
Affirmative action recognizes that
gender-blind policies are impossible
to implement because there are no
Affirmative action procedures
acknowledge that people are not
hired simply on the basis of their
qualifications. Those who have an
unfair advantage because of their
membership in a particular group
receive preferential treatment according to characteristics
irrelevant to the jobs they seek. Those
irrelevant characteristics have prevented women
and minorities from getting their fair share of
Although affirmative action has been misperceived
as making employers hire a woman or
minority candidate over a more qualified white
man, it in fact ensures the hiring of female and
minority candidates who are more qualified than
their white male competitors. It also gives hiring
preference to female or minority candidates who
are as qualified as white male applicants. The
goal is a work force in which no group is overrepresented.
The misunderstanding about affirmative
action stems in part from our belief that hiring
procedures are meritocratic and that the best
person gets the job. Even though we all have
ample evidence that the “best person” (if such a
notion can be sensibly defined) does not always
get the job, we cling to the idea of a “just
world” in which the deserving are rewarded and
the unrewarded are undeserving. We rely on
principles of meritocracy and fair play to justify
decisions that we make about others. Our
explicit commitment to equality makes it difficult
for us to perceive the extent to which we
make unfair, nonmeritocratic evaluations and
decisions based on gender and race schemas.
Those schemas are themselves the other
source of our misconceptions about affirmative
action. From the outset, nonwhite, nonmale job
candidates are perceived as having fewer qualifications
than white male applicants. Such persons,
it is assumed, need affirmative action to
get a job. In reality, however, affirmative action
helps to counteract the continuing, if unwitting,
overvaluation of white males.
Besides implementing institutional reforms
to eliminate the inequities that gender schemas
encourage, people can learn to reason better.
The findings of cognitive psychology can help us
avoid mistakes in judging other people. Even
without the influence of schemas, evaluators are
prone to errors in reasoning. They tend to give
too much weight to extreme examples, ignore
information about how frequently different
events occur, and overestimate the value of their
personal experience. Social schemas intensify
A common error is the failure to appreciate
covariation, the phenomenon in which two factors
vary together. For example, University of
British Columbia psychologist Mark Schaller
and his colleagues asked college students to
assess the leadership potential of men and
women in a fictitious company in which most
executives were men and most office workers
women. Within each group, the same percentages
of men and women showed leadership ability.
In this example, leadership ability and status
within the company covaried. The covariation
misled the male student judges, who erroneously
inferred that the male workers had more leadership
ability than the females. Those students saw
only that, overall, more men than women
showed high leadership ability; they neglected
the fact that most executives were men. Followup
studies demonstrated that students were less
likely to make such gender errors after receiving
training in the logic of covariation. People can
be trained to reason in a way that will minimize
the effects of gender schemas.
A similar reasoning error is the blocking of
relevant hypotheses. If people have a hypothesis
that explains a regularity, they tend not to entertain
other valid hypotheses. That is, they often
fail to perceive causes that might contribute to a
person’s performance if a prior hypothesis —
such as a gender schema — independently predicts
An experiment by University of Utah psychologist
David Sanbonmatsu and colleagues
demonstrated how blocking works. Participants in the experiment learned a number of facts
about fictitious students who had taken a welding
course. Many of the facts were irrelevant to
the students’ success or failure, but one piece of
information — about course load — was important.
Students with a light course load passed,
and those with a heavy course load failed.
Participants also received information about the
students’ gender. One group learned that all the
passing students were male and all the failing
students were female. Another group learned
that half the students who passed were male,
and half the students who failed were male.
Participants were asked to evaluate why some
students had passed and others had failed.
The experimenters reasoned that participants
would expect males to be more likely than
females to pass a welding course. If the gender
information supported such an expectation, they
thought, the participants would not notice the
other characteristic that predicted performance,
namely course load. The division of success and
failure along gender lines would block students
from seeing that gender covaried with course
load. In contrast, participants given information
that did not support expectations based on gender
schemas would tend to see that course load
explained the students’ performance. The results
verified the predictions.
The welding experiment has obvious implications
for judgments about women in professional
settings. People who see a woman fail at a
task they expect her to fail at because of the
influence of gender schemas will probably not
perceive other possible causes of her failure.
They will attribute her failure to her sex instead
of looking for other reasons, even if those other
reasons actually caused her failure. They may
even feel that a search for other causes is a
search for excuses.
Evaluators can learn how to correct for
blocking in the same way that they can learn to
understand covariation. Once trained to reduce
errors in their reasoning, these people may then
be able to mitigate the effects of gender schemas
in their own judgments. Understanding that their
own gender-based expectations may affect their
assessment of other people, these evaluators will
thus judge others more fairly and accurately.
On balance, there is some reason for optimism.
Although women’s advancement is too
slow, gender schemas operate covertly, bias evaluations
show that small examples of bias add
up, and people’s reasoning is often flawed, we
can understand how these processes work and
do something about them. Relying on our
knowledge of how schemas work and how
advantage accumulates, we can make institutions
Back to January 1999 Contents
Back to STATUS Table of Contents