Joan McKay Versus John McKay: Do Gender Stereotypes Bias Evaluations?

Abstract

Examines research using a classic, influential experiment conducted by Goldberg (1968), showing that women were likely to rate male authors (e.g., John T. McKay) more favorably than female authors (e.g., Joan T. McKay) of identical articles. Although replications of this study have been inconclusive, Goldberg's research is still frequently cited as demonstrating an evaluative bias against women. A quantitative meta-analysis of research using Goldberg's experimental paradigm shows that the average difference between ratings of men and women is negligible. Furthermore, although the effect sizes are not homogeneous, the difference remains negligible when other factors such as sex of subject or year of publication are taken into consideration. Several explanations for the heterogeneity of effect sizes and the inconsistency of findings are discussed. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Full text

Bulletin
105,
No.
3.409-429
Copyright
1989
by
the
American
Psychological
Association,
Inc.
0033-2909/89/M0.75
Joan
McKay
Versus
John
McKay:
Do
Gender
Stereotypes
Bias
Evaluations?
Janet
Swim,
Eugene
Borgida,
and
Geoffrey
Maruyama
University
of
Minnesota
David
G.
Myers
Hope
College
Examines
research
using
a
classic,
influential
experiment
conducted
by
Goldberg
(1968),
showing
that
women
were
likely
to
rate
male
authors
(e.g.,
John
T.
McKay)
more
favorably
than
female
authors
(e.g.,
Joan
T.
McKay)
of
identical
articles.
Although
replications
of
this
study
have
been
inconclusive,
Goldberg's
research
is
still
frequently
cited
as
demonstrating
an
evaluative
bias
against
women.
A
quantitative
meta-analysis
of
research
using
Goldberg's
experimental
paradigm
shows
that
the
average
difference
between
ratings
of men and
women
is
negligible.
Furthermore,
although
the
effect
sizes
are not
homogeneous,
the
difference
remains
negligible
when
other
factors
such
as
sex
of
subject
or
year
of
publication
are
taken
into
consideration.
Several
explanations
for
the
hetero-
geneity
of
effect
sizes
and the
inconsistency
of findings are
discussed.
In
1968,
social psychologist Philip Goldberg published
his in-
fluential
experimental
study
on
prejudice against
women
(Goldberg,
1968).
The
study
was
simple
yet
compelling. Gold-
berg
gave
female
subjects identical
booklets
containing
six
different
articles.
For
each article,
however,
half
the
subjects
were
told that
the
author
was a
woman
(e.g.,
Joan
T.
McKay)
and
the
other
half
were
told that
the
author
was a man
(e.g.,
John
T.
McKay).
On a
number
of
ratings, such
as
evaluations
of
the
author's competence,
the
male author
was
given
higher
ratings
than
the
female
author.
Specifically,
Goldberg
reported
that
men
received higher rat-
ings
on 44
of
54
measures.
However,
after
forming
an
index
by
summing
across
the
dependent
variables,
he
reported
six t
tests,
only
three
of
which
were
significant.
Two
of
the
three
significant
findings
were
for
articles about traditional masculine
fields
(law
and
city
planning),
and one was for a
sex-neutral
field
(linguis-
tics).
Two of the
three nonsignificant
findings
were
for the
arti-
cles
on
traditionally
feminine
fields
(dietetics
and
education),
and
the
third
was
for
a
second sex-neutral
field (an
history).
In
effect,
then, Goldberg seemed
to
have
found
qualified
support
for
gender
bias
in
evaluations.
Perhaps
not
surprisingly, Goldberg's article
is
frequently
Janet
Swim,
who
is
now
in the
Department
of
Psychology
at
Pennsyl-
vania
Slate
University,
was
supported
in
part
by a
fellowship
from
the
Graduate
School
of the
University
of
Minnesota.
The
authors
are
grate-
ful
to
John
Claypool
for his
research
assistance
and to
Nancy
Brekke
for
her
insights
during
the
formative
stages
of the research.
Allen
Omoto,
Mark
Snyder,
Cynthia
Thomsen,
and
especially
Alice
Eagly
provided
helpful
comments
on an
earlier
version
of
the
manuscript.
Por-
tions
of
this
research
were
presented
at the
1986
meetings
of the
Ameri-
can
Psychological
Association,
Washington,
DC.
Correspondence
concerning
this
article
should
be
addressed
either
to
Janet
Swim,
Department
of
Psychology,
515
Moore
Building,
Pennsyl-
vania
State
University,
University
Park,
Pennsylvania
16802,
or to Eu-
gene
Borgida,
Department
of
Psychology,
University
of
Minnesota,
75
East
River
Road,
Minneapolis,
Minnesota
S5455.
cited
as
evidence
of
discrimination against
women
and of
women
holding
prejudicial
beliefs
about their
own
gender
(e.g.,
Cash
&
Trimer,
1984;
Gornick
&
Moran,
1971; Lips
&
Colwill,
1978;
Paludi
&
Strayer,
1985; Ruble
&
Ruble,
1982).
Although
it is
well
established that discrimination occurs
(e.g.,
Hewlett,
1986)
and
that Goldberg's
findings
certainly seem
to
illustrate
sex
bias,
many
authors
in the
scientific
and
popular literature
who
cite
his
article
in
fact
misrepresent
the
strength
of the
re-
sults Goldberg reported.
In
addition,
the
robustness
of
Gold-
berg's
findings is
uncertain because
of
numerous inconsistent
conceptual replications.
One
frequent
misrepresentation
of
Goldberg's
study
has
been
to
distort
the
nonsignificant
results.
For
instance, Paludi
and
Bauer
(1983)
stated,
"Results
indicated that
women
rated
the
articles
(even
those
in fields
considered
sex
appropriate
for
women)
more
favorably
when
they
were
attributed
to a
male
rather than
a
female
author"
(p.
387).
In
fact,
Goldberg does
not
report
any
significant
differences
on
ratings
of the
authors
on
the
traditionally
feminine
fields.
Further,
Goldberg
did not
demonstrate discrimination
in all of the
comparisons
tested.
Other authors
do not
misstate
the
conclusion
but
implicitly
distort
the findings by
failing
to
mention
the
nonsignificant
findings.
For
instance, Lips
and
Colwill
(1978)
stated,
"This
ste-
reotype
is
seen
in
action
in
studies that look
at the way
people
evaluate
male
and
female
performance. Goldberg
(1968)
showed
that
female
college
students evaluated
articles
suppos-
edly
written
by
women
lower
than
the
identical articles
attrib-
uted
to
male
authors"
(p.
188).
In a
more recent
review
paper,
Wallston
and
O'Leary
(1981)
also distorted
the findings by
stat-
ing,
"In the
landmark study
of
competency bias
favoring
men
(Goldberg,
1968),
college
women
rated professional articles
for
value,
persuasiveness,
profundity,
writing
style
and
compe-
tence. Higher ratings
were
given
to
identical papers
when
the
author
of the
article
was
portrayed
as a
male rather
than
a fe-
male"
(p.
19).
Omitting
the
nonsignificant
effects
creates
the
409
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

410
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
impression that
the
Goldberg
effects
were more robust than
they actually were.
The
strength
of
Goldberg's
findings
have
also
been distorted
by
disregarding failures
to
replicate.
For
instance,
in his
well-
respected introductory psychology text,
Gleitman
(1981)
de-
scribed Goldberg's study
but did not
mention
any of the
con-
ceptual
replications.
In a
review paper,
Unger
(1976)
referenced
both Goldberg
(1968)
and
Pheterson,
Kiesler,
and
Goldberg
(1971)
as
evidence
for
"the persistent tendency
of
females
to
devalue
the
work
of
other females"
(p. 5),
thus leaving
the im-
pression that Goldberg's
findings
have been supported
by
repli-
cations.
More recent reviews
of
this
literature,
such
as
those
by
Wall-
ston
and
O'Leary
(1981)
and
Basow
(1986),
have
acknowledged
the
presence
of
studies that
fail
to
replicate
Goldberg's
findings.
These studies examined additional independent variables (e.g.,
status
of
the
person),
different
samples (e.g., male subjects),
and
different
stimulus materials (e.g.,
job
applications).
The
conclu-
sions
of
these reviews suggest that these empirical inconsisten-
cies
can be
explained
by
various
qualifying
conditions.
For in-
stance, Basow
stated,
"From
the
vast amount
of
research
in
this
area since
1968,
it is
clear that prejudice still exists
in
both
men
and
women,
but it
usually exists
in
interaction with
situational
factors,
and
more
often
in
subtle,
as
opposed
to
obvious ways"
(pp.
234-235).
Reviews
of
this literature
to
date have primarily been qualita-
tive
reviews
(e.g.,
Basow,
1986;
Wallston
&
O'Leary,
1981);
such
reviews
are
constrained
by
reviewers'
biases
and
limitations
in
synthesizing
information. Qualitative reviews typically require
reviewers
to
subjectively
evaluate studies, which
can
lead
to
bias
resulting
from
the
reviewers' beliefs
or
expectations. Perhaps
more importantly, qualitative reviews
are
characterized
by the
inherent
difficulty
of
summarizing large numbers
of
studies
and
by
the
neglect
by
reviewers
of
large amounts
of
data
in the
origi-
nal
reports
(see Cooper
&
Rosenthal,
1980;
Glass,
1976).
As an
alternative approach, quantitative methods such
as
meta-analy-
sis
have been developed
to
examine results across
a
body
of
con-
ceptually
interrelated studies. These methods
can
eliminate
some
of the
biases
associated with qualitative reviews.
For in-
stance, quantitative reviews tend
to be
more objective
and are
more
likely
to
include most,
if not
all, relevant studies (see
Bangert-Drowns,
1986;
Green
&
Hall, 1984).
In
light
of the
questionable strength
of
Goldberg's original
findings
and
because
of the
limitations
associated
with qualita-
tive
reviews,
the
validity
of
even
qualified
conclusions, such
as
those
of
Basow (1986),
is
uncertain.
The
purpose
of our
study,
then,
is to use
quantitative methods
in the
form
of
meta-analysis
techniques
(a) to
assess
the
strength
of
the
tendency
to
discrimi-
nate against women
in
studies using variations
of the
paradigm
originally
used
by
Goldberg
and (b) to
identify
and
examine
variables that might influence
the
strength
of
this tendency.
In
the first
section,
hypotheses that have been tested
by
other
inves-
tigators
are
reviewed. These same hypotheses
are
also examined
in
the
quantitative meta-analysis presented later
in the
article.
Additional hypotheses about methodological
differences
across
studies
and
about
characteristics
associated
with
the
publica-
tion
of the
articles
are
presented
in the
second
and
third
sec-
tions,
respectively.
Variables
Examined
in
Studies
Subsequent
to
Goldberg's
Original
Study
Subject
characteristics.
Three subject characteristics tested
in
previous research
are
examined
in
this
review.
Although
Goldberg
(1968)
only
studied
female
subjects, other studies
fre-
quently tested
differences
between male
and
female
subjects.
In
these studies
it was
generally expected that male subjects would
be
more likely than
female
subjects
to
rate
female
target persons
lower
than male target persons.
A
related subject variable
is sex
role orientation.
It was
expected that masculine
or
feminine
subjects
would
be
more discriminatory than androgynous
sub-
jects.
A
third
subject
variable
is age of the
subjects. Like subjects
with
masculine
or
feminine
sex
role orientations, older people
were
expected
to
discriminate more than younger people.
Target
person
characteristics.
One
characteristic
of
the
target
person that
has
been studied
is his or her
physical attractiveness.
People
who are
more attractive
are
likely
to be
rated more
favor-
ably
than less attractive people
(Berscheid,
1985;
Dion,
Ber-
scheid
&
Walster,
1972).
However,
attractiveness
may be
more
important
for
women
than
for
men
(Bar-Tal
&
Saxe,
1976;
Wall-
ston
&
O'Leary,
1981).
Thus, more discrimination might
be
predicted
for
unattractive women,
in
comparison with unat-
tractive men.
A
second characteristic
of the
target person
is his or her
race.
To
the
extent that evaluations
of
women
reflect
general
impres-
sions based upon their
lower
status
in
society
(Unger,
1976),
other visible
and
salient status variables such
as
race
should
also
affect
evaluations. Insofar
as
effects
of
status variables
are cu-
mulative,
minority women
may be
judged particularly
unfavor-
ably.
Moreover,
the
interplay between race
and sex
cues
may
be
more complex, resulting
in
greater
differences
between non-
White women
and men
than between White women
and
men.
Thus
it
seems important
to
examine directly status cues based
on
race
in
order
to
understand better
how
evaluations
of men
and
women
may
differ.
Two
opposing conclusions
have
been proposed regarding
a
third characteristic
of the
target person, that
of
competency
or
success
of the
target person. First, because being
a
woman
is
typically
seen
as a
lower marker
of
status than
is
being
a man
(Unger,
1976),
competent
or
successful women
would
violate
gender
stereotypes.
If
so,
competent
or
successful women might
receive
greater discriminatory treatment
by
evaluators
than
would
incompetent
or
unsuccessful women.
In
support
of
this
perspective,
Nieva
and
Gutek
(1980)
stated,
"While females
are
evaluated less
favorably
than males when they
are
highly quali-
fied or
perform
well,
females
are
evaluated more
favorably
than
males
when both
are not
well
qualified
or are
unsuccessful per-
formers"
(pp.
273-274).
On the
other hand, biases
may
exist
when
women have
not
displayed competency
or
success,
but
biases
may not
exist when women
have
displayed competency
or
success.
For
instance, Wallston
and
O'Leary
(1981)
stated,
"The results indicate that women
are
likely
to be
evaluated
as
being
as
competent
as men
when
their performance
is
deemed
exceptional"
(p.
20). These opposite conclusions
may not be
inconsistent
but
rather
may be a
function
of an
additional vari-
able,
namely,
whether
the
competency
or
success
was
displayed
in a
traditionally
female
or
male occupation (see
Basow,
1986;
Wallston
&
O'Leary, 1981).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
411
Stimulus
material
characteristics.
Some
studies
have
varied
characteristics
of the
stimulus materials
in
addition
to
different
characteristics
of the
target person. Quality
of the
stimulus
ma-
terial
has
been manipulated
in a
manner similar
to the way
competency
or
success
of
a
target person
has
been manipulated.
For
instance, some studies varied
the
success
of the
target per-
son by
indicating whether
he or she had a
college degree,
and
others varied
the
quality
of the
stimulus material
by
indicating
whether
the
article being rated
had
been accepted
for
publica-
tion.
Thus, opposing
predictions
could
be
made
for
quality
of
the
stimulus material
in
the
same
way
as
they
are
made
for
com-
petency
or
success
of
a
target person.
A
second characteristic
of the
stimulus material
is its sex
role
orientation.
As
Goldberg (1968) concluded, biases against
women
may be
stronger
in fields
incongruent
with their
gender.
Methodological
Differences
Across
Studies
Research
design.
Strength
of findings in
studies
may
vary
as a
result
of
different
characteristics
of the
study designs.
For
instance,
whether subjects
rated
both
male
and
female
target
persons
or
just
a
male
or
just
a
female
target person
may
affect
ratings. That
is,
rating
both
male
and
female
target persons
might
sensitize subjects
to the
purpose
of
the
study,
which could
decrease
the
likelihood that
they
would
differentially
rate
the
target
persons.
On the
other hand, rating both male
and
female
target persons
may
cause subjects
to
contrast
the
target persons
and
thus perceive greater
differences
between male
and
female
target
persons.
Stimulus
material.
A
second methodological characteristic
is
the
amount
of
information
given
to the
subjects about
the
target person.
Nieva
and
Gutek
(1980)
stated,
"The
more task-
related
information
provided about
the
'evaluatee'
and the
greater
the
clarity about
the
criteria
to be
used
in the
evaluation
situation, therefore,
the
less
likely
it is
that
'actuarial
prejudice'
will
operate"
(p.
273;
see
also
Futoran
&
Wyer,
1986). Similarly,
Locksley,
Borgida,
Brekke,
and
Hepburn
(1980)
found
that
sex
stereotypes,
conceptualized
as
probabilistic
base
rates,
were less
likely
to be
used
when
specific information
in
addition
to
gender
was
presented. Thus, more information about
the
target person
might
yield less discriminatory ratings.
The
different
behaviors
or
products
that
were
used
as
stimu-
lus
material
can be
placed into general categories such
as
behav-
iors
or
applications
for
jobs.
Differences
in
ratings
of men and
women
may be
more pronounced
in one
category than
in an-
other.
For
instance, subjects
may be
less cautious about using
their
gender
stereotypes
when
the
stimulus materials
are
essays
than
when
they
are job
applications.
Dependent
variables.
The
extent
to
which stereotypes
are
used
when
making judgments
may
differ
depending
on the
par-
ticular
dependent variables
in use
(e.g.,
Bodenhausen
&
Wyer,
1985).
For
instance,
Whitley
and
Frieze (1986)
found
differ-
ences
in
attributions
for
achievement based
on
question word-
ing.
One
difference
in
question wording
in the
literature
re-
viewed
here involves whether subjects rated
the
target person
or
the
stimulus materials. Rating
the
target person might make
the
gender
of the
target person more salient than would rating
the
stimulus material.
In
the
same manner that
having
less
informa-
tion
about
a
target
may
result
in
more
stereotypic
responses,
increasing
the
salience
of the
target
person's
gender might yield
a
greater tendency
to
discriminate
against women.
On the
other
hand, greater
saliency
might
increase
the
tendency
to
give
a
so-
cially
desirable response
and
therefore decrease
the
likelihood
of
a
discriminatory response.
Quality
of
research.
A
more general methodological
differ-
ence
in the
studies that might
influence
the
strength
of the find-
ings
is the
quality
of
the
research.
Meta-analyses
have
been
criti-
cized
for
giving
equal
weight
to
both
well-conducted
and
poorly
conducted studies
(e.g.,
Eysenck,
1978;
Slavin,
1986).
Like
all
reviews,
meta-analysis
results
are
influenced
by the
quality
of
the
primary
sources.
For
instance, previous
meta-analyses
have
shown
that
the
strength
of the
summary statistics
is
influenced
by
the
inclusion
or
exclusion
of
studies that
only
allow
the
re-
viewer
to
assign nominal levels
of
significance
to
effects
(e.g.,
Eagly
&
Carli,
1981).
Other deficiencies
in
reporting—such
as
completeness, accuracy,
and
clarity—have
also been
shown
to
influence
meta-analyses
(Orwin
&
Cordray,
1985).
Publication
of
Articles
Sex
of
author.
The
last
set of
variables that might
influence
the
strength
of
discriminatory responses involve
characteristics
of
the
articles themselves.
One
variable
associated
with
the
arti-
cles
is the sex of the
authors.
Eagly
(1978)
and
Eagly
and
Carli
(1981),
in
reviews
of
gender
differences
in
influenceability,
con-
cluded that male authors
were
more
likely
than
female
authors
to find
that women
were
more easily
influenced
than men. This
conclusion
has
not,
however,
been reached
in
other
meta-analy-
ses of
gender
differences
(e.g.,
Eagly
&
Crowley,
1986;
Eagly
&
Steffen,
1986).
Publication
year.
A
second
difference
in the
articles
is the
year
in
which
they
were
published. Publication year
has
been
shown
to be
related
to findings
regarding gender
differences
in
cognitive
abilities
(Rosenthal
&
Rubin, 1986),
influenceability
(Eagly
&
Carli,
1981),
and
helping behavior
(Eagly
&
Crowley,
1986).
In the
case
of
helping
behavior,
however,
publication year
was
found
to be
correlated
with
a
number
of
other study charac-
teristics
(Eagly
&
Crowley,
1986),
and for
aggressive
behavior,
the
correlation
with
publication year
was not
significant
(Eagly
&
Steffen,
1986). Given
the
apparent advances
in
women's
rights
and
the
decrease
in
people's willingness
to
endorse
un-
equal treatment
of men and
women,
it
seems important
to ex-
amine
the
relationship between ratings
of
target people
and
year
of
publication
for
the
present research literature
as
well.
In
summary, variables examined
in
previous research
were
reviewed,
and for
each variable, pertinent hypotheses
were
identified
for
purposes
of
this meta-analysis. Furthermore,
methodological factors that bear
on the
robustness
of
the
Gold-
berg
(1968)
phenomenon were reviewed
and are
examined
in
the
meta-analysis
presented
in the
next section. Finally,
differ-
ent
characteristics
of the
articles
were
identified
as
possible
sources
of
variation
in the
strength
of the
effects.
Method
Literature
Search
The
data
base
of
articles
was
compiled
by
locating
articles
referenced
in
the
Social
Science
Citation
Index,
by a
computer
search
of the
Psy-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

412
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
chological
Abstracts
(PsycINFO),
and by
cross-referencing
various arti-
cles
and
reviews
of the
literature
on sex
discrimination. From 1968
to
1985,
the
Social Science
Citation
Indices contained
322
citations
that
referenced
P. or P. A.
Goldberg (1968), making this study
a
"citation
classic"
by
Current
Content's
standards.
The
PsychlNFO
computer
search
of
abstracts
from
1968
to
1985, using
the
document
title
Gold-
berg
Replication,
yielded
103
abstracts.
Clearly, this
is a
well-known
and
influential
study.
The
articles
and
chapters
that
reviewed
the
experimen-
tal
literature were
Arvey
(1979),
Nieva
and
Gutek
(1980),
Wallston
and
O'Leary
(1981),
Basow
(1986),
and
Tbsi
and
Einbender
(1985). Addi-
tionally,
the
reference
sections
of the
articles included
in
this
meta-anal-
ysis
were
examined.
The
following
selection criteria were used
in
order
to
form
a
relatively
homogeneous sample that represented conceptual replications
of
Gold-
berg's
original study:
1.
The
vast
majority
of
empirical papers were published
in
North
American
English-language journals. Thus, only studies conducted
in
North America
and
published
from
1968
to
1985
were
included.
2.
Only published
articles
were included. Although this
may
produce
a
biased sample, this limitation
was
imposed primarily because most
reviews
indicate that unpublished articles
are
less
likely
to
report sig-
nificant
findings
(Glass,
McGaw,
&
Smith, 1981; Green
&
Hall, 1984;
Greenwald,
!975;Lane&Dunlap,
1978). This tendency
was
supported
in
the
unpublished papers
we
obtained through correspondence with
a
few
investigators. Thus, exclusion
of
unpublished studies
would
yield
findings
that maximize
the
size
of
effects
in the
present
meta-analysis.
Alternatively,
it
could
be
argued that there
was a
bias against publishing
differential
evaluations
of
men
and
women.
However,
such
a
publication
bias seemed
unlikely
for
this topic because articles
and
reviews
pub-
lished
during
the
time period covered tended
to
emphasize
significant
findings.
3.
Only studies
in
which subjects rated hypothetical adult target per-
sons
were
included. This excluded studies
in
which subjects rated chil-
dren, actual adults such
as
subjects' supervisors
or
teachers, occupa-
tional categories,
sex
stereotypical
traits,
or men and
women
"in
gen-
eral,"
and
excluded salary comparisons,
differential
rates
of
publications,
and
self-ratings.
In
addition
to
keeping
the
sample homo-
geneous,
this also insured that there
were
not
actual
differences
among
the
target persons.
4. The
subject
had to
have
been
given
information about
a
behavior
produced
by the
target person, such
as a job
performance
or
articles
written
by the
target person. That
is,
studies
were
excluded
when
sub-
jects
rated
a
target person
on the
basis
of
only
a
name
or a
name
and an
occupational
title. If a
subject
was
given
a job
application, resume,
or
similar
set of
materials, these documents were assumed
to
include indi-
cators
of
past behaviors,
and so
these studies
were
included. Addition-
ally,
this criterion excluded studies that
had
subjects decide
if
hypotheti-
cal
psychotherapy clients
were
psychologically healthy.
However,
this
criterion
did not
exclude studies that
had
subjects evaluate hypothetical
counselors'
or
clinicians'
job
performances.
5.
Studies primarily exploring
different
topic
areas
other than
differ-
ential evaluations
of men and
women were excluded.
In
particular,
this
excluded persuasion, attribution,
and
attraction
studies
that
were
not
directly related
to our
focus
on the
Goldberg paradigm.
If a
study pri-
marily
dealt
with
evaluations
but
included
a
few
attributional
or
attrac-
tion-dependent
variables,
it was
included,
but the
dependent variables
measuring
attribution
or
attraction
were excluded.
The
search
and
selection criteria resulted
in
123
studies
from
106
articles published primarily between
1974
and
1979
(see Appendix
A).
The
trend
in
publishing
the
replications resembles
a
normal distribu-
tion
(see
Figure
1),
perhaps
reflecting
the
established
tendency for
cer-
tain
research themes
in
social
psychology
to
generate considerable
re-
search over
a
relatively brief period
of time
(Jones,
1985).
Coding
of
Studies
Both
main
effects
and
simple
effects
reflecting
the
comparisons
be-
tween
male
and
female
target persons
from
two-way
interactions
were
coded
for
each dependent variable. Higher order interactions
were
not
included. Mean
effect
sizes
(d) and
voting scores
(a
tally
of the
number
of
significant
and
insignificant
results
reported)
were calculated
for
both
the
main
effects
and the
simple
effects.
(Appendix
B
contains
a
more
detailed description
of our
assumptions,
and
Appendix
C
contains
a
more detailed description
of the
calculations.)
A
negative value
for the
effect
sizes
represents a
more
favorable
rating given
to
men,
and a
posi-
tive
value
represents a
more
favorable
rating
given
to
women.
Effect
sizes
were
weighted
inversely
by
their variances. This
gives
more
weight
to
studies with smaller variances
and
thus yields
a
more
precise estimate
of
(/(see
Hedges
&
Olkin,
1985).
Because
the
weighted
d
is
a
more precise estimate,
only
the
weighted
ds
are
reported.
However,
it
should
be
noted that because
the
majority
of the
effect
sizes
were
small, there
was
little
difference
between
the
weighted
and the un-
weighted
effect
size.
In
addition
to the
main
effects,
the
following
variables
were
coded:
Subject
variables.
These included
(a) sex of
subjects (women
only,
men
only,
or
both
men and
women),
(b) the
Bern
Sex
Role Inventory,
and (c) age of the
subjects (high
school,
undergraduates
or
graduates,
older than college age, more than
one age
group represented
in the
study).
If the
sample contained more than
one age
group
and
differences
between
age
groups
were
reported, separate results
were
recorded
for
each
age
group. These
two
results
were
coded
as two
different
studies.
If
the
differences
were
not
significant
between
age
groups
and if
separate
data
for the age
groups
were
not reported,
identical
effect
sizes
and
vot-
ing
scores
were
recorded
for
each
finding.
Target
persons,
(d)
Competence
of the
target person (high, medium,
or
low)
was
determined
by the
level
reported in the
article.
If the
study
only
had two
levels,
no
value
was
recorded
for
the
medium competence
level,
(e)
Attractiveness
of the
target person
was
determined
by the
level
reported
in
the
article.
If the
study only
had two
levels,
no
value
was
recorded
for the
medium attractiveness
level.
Although there
are no
absolute standards
of
competence
and
attractiveness,
it was
assumed
that
using
the
levels
reported
would yield distributions
of
high,
me-
dium,
and low
levels
whose
means
would
differ
in the
same
order,
(f)
Race
of
the
target person (White
or
non-White)
was
also coded.
Stimulus
materials,
(g)
Quality
of the
stimulus material
was
deter-
mined
by the
level
reported
in the
article
and
again
was
assumed
to
represent a
distribution
of
high,
medium,
and low
levels.
If the
study
only
had two
levels,
no
value
was
recorded
for the
medium quality
level.
(h)
Sex
role orientation
of the
stimulus material (feminine, masculine,
or
sex
neutral)
was
also
coded.
Methodological
characteristics.
These included
(i)
number
of
target
persons evaluated
and (j)
amount
of
information
about
the
target per-
son
given
to the
subjects
(only
name,
one
sentence,
one
paragraph,
more
than
one
paragraph,
or resume/application).
Also coded
was
whether
the
subjects
saw a
picture
of the
target person,
viewed
a film or
video
of
the
target person,
or
heard
a
tape recording
of the
target person.
In
the
analyses,
if
the
target person
was
represented
in
these
forms,
the
amount
of
information
about
the
target person
was
considered
to be
more
than
one
paragraph.
Additionally,
the
category
of
resume/application
was
collapsed with
the
category
of
more
than
one
paragraph.
Thus,
only
four
levels
of
amount
of
information about
the
target person were ana-
lyzed,
(k) The
number
of
independent variables manipulating
differ-
ences
in the
target person across conditions
was
also used
as a
second
method
of
approximating
the
amount
of
information about
the
target
person.
(1) Six
categories were
formed
for
choice
of the
stimulus mate-
rial: written
works
(e.g.,
an
article,
memo, speech,
or
artwork); behav-
iors;
job
applications
or
resumes;
biographical
descriptions;
applica-
tions
and
essays;
and
other,
(m)
Question wording
was
defined
as
whether
the
dependent variable asked
for
ratings
of the
target person
or
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER STEREOTYPES
413
ratings
of
the
stimulus material
or
both,
(n) The
\ar\able
journal
quality
was
used
as an
indirect measure
of the
quality
of the
research.
Journal
ratings were obtained
by
averaging ratings
of
these journals
by five
social
psychologists
at the
University
of
Minnesota.
Study
characteristics.
These included
(o) sex of the first
author,
(p)
percent
of
female
authors,
and (q)
publication year.
Analyses
Effect
sizes
were
treated
in
three
different
ways.
First,
the
mean
weighted
ds
were
calculated across
all the findings.
Second,
the
mean
weighted
d was
calculated
after
excluding studies
in
which
effect
sizes
and
voting scores
were
assigned
a
value
of
zero because
of
insufficient
information.
Third, mean
effect
sizes, including
all findings,
were
calcu-
lated within
a
study,
and
then
a
mean
of
these mean
effect
sizes
was
calculated.
Findings
with
insufficient
information
were
from
studies that
(a) did
not
mention
the
significance
of the finding or (b)
reported
the finding
as
nonsignificant
but did not
report
sufficient
information
to
calculate
the
d or to
estimate
the z
score. When included, these
findings are
as-
sumed
to be
randomly distributed around
a
mean
of
zero.
It
should
be
noted that studies that reported
findings as
significant
but did not
report
other information needed
for
calculations
were
included
in
this second
calculation (see Appendix
C). It
could
be
argued that including
all find-
ings
(i.e., those with
sufficient
and
insufficient
information) results
in a
bias toward nonsignificant
findings and
that
excluding
findings
assumed
to be
equal
to
zero results
in a
bias toward more
significant
findings.
Thus,
the
results including
findings
assumed
to be
equal
to
zero
can
be
considered
a
lower limit,
and
those excluding these
findings can be
considered
an
upper limit.
The first two
methods
of
combining
d&
accords more weight
to
studies
with
more dependent variables, whereas
the
third method weights stud-
ies
equally
by
weighting each
finding
inversely
by the
number
of findings
per
study.
The
unit
of
analysis
in the first two
approaches
is findings,
and
the
unit
of
analysis
in the
third approach
is
studies. Because results
reported within
a
study
are not
independent,
Rosenthal
and
Rubin
(1986)
recommended that
the
study
be
used
as the
unit
of
analysis
by
combining
the findings,
with
the
correlation between
the findings
within
the
studies taken into account.
However,
in
this
meta-analysis,
only
seven
studies reported these correlations. Furthermore,
the
calculations
assume homogeneous correlations between measures that
did not
seem
likely
in the
present research. Therefore,
finding
a
mean
dfoi
each study
seemed
to be the
most reasonable
way
to
adjust
for the
interdependence
between
measures
in
this
set of
studies.
Because
the
distribution
of
effect
sizes
is not
likely
to
meet
assump-
tions required
for
conventional
statistical
tests, alternative tests recom-
mended
by
Hedges
and
Olkin
(1985)
were
used.
Specifically,
Hedges
and
Olkin provided formulas
for
calculating
the
variance
of the
effect
sizes
and
confidence
intervals. Additionally, they provided formulas
for
calculating
the
homogeneity
of the
effect
sizes.
The
homogeneity tests
allow
one to
search
for
outliers
and to
test whether
the
effect
sizes
are
uniform.
If
effect
sizes
are
heterogeneous, they
may be
able
to be
catego-
rized into smaller groups that
are
homogeneous. Furthermore, once
ho-
EA
.fi
I
Year
Figure
1.
Number
of
replications
per
year.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

414
SWIM,
BORGIDA,
MARUaMA,
AND
MYERS
mogeneous
groups
are
found,
these
groups
can be
compared
to see if
their
average
effect
sizes
differ.
On
the
basis
of the
homogeneity
test,
three
findings
were
determined
to be
outliers
and
were
removed
from
the
data.
These
three
findings
came
from
results
reported by
Rosen
and
Jerdee
(1974)
and
Rosen,
Jerdee,
and
Prestwich
(1975).
Although
the findings
were
large
(d
=
-.85, -.74,
and
.62),
they
were
not the
largest
effect
sizes.
The findings
were
outliers
not
only
because
the
effect
sizes
were
among
the
largest,
but
also
because
the
study
had a
large
sample
size
(N
=
1,476).
Because
of
the
large
number
of findings in
this
review,
including
or
excluding
these
three
findings
does
not
appreciably
change
the
mean
weighted
effect
size.
However,
it
does
increase
the
likelihood
that
the findings
will
be
more
homogeneous.
It
should
be
noted
that
other
studies
using
sim-
ilar
stimulus
material
were
included
in
this
review.
Some
of the findings
from
these
studies
were
also
fairly
large;
however,
they
were
not
outliers,
and
therefore
they
were
not
excluded
from
the
data.
1
In
addition
to
calculating
mean
weighted
ds
and
voting
scores
for
the
main
effects
and
simple
effects
reported
in the
articles,
correlations
between
the
mean
weighted
main
effect
d and
continuous
variables
were
examined.
Differences
in the
mean
weighted
main
effect
ds
were
exam-
ined
across
noncontinuous
variables,
as
well
as
across
continuous
vari-
ables
put
into
the
form
of
categorical
data,
to
test
for
nonlinear
relation-
ships.
Significance
of
the
correlations
is
calculated
according
to the
rec-
ommendations
outlined
in
Hedges
and
Olkin
(1985).
Results
First,
descriptions
are
given
of the
articles,
subjects,
and
pro-
cedures reviewed. Second,
the
mean weighted main
effect
ds
and
simple
effect
ds and the
voting
scores
are
reported.
Finally,
correlations
and
other
analyses with
the
mean main
effect
ds
are
reported.
Articles
Of
the 106
articles
reviewed
with
a
total
of 595 findings, 12
of the
articles
reported
two
experiments,
1
reported
three
ex-
periments,
and 1
reported
four
experiments.
This
yields
123
studies. Seventeen percent
of the
studies were published
in
"high-quality"
journals,
58% in
"medium-quality"
journals,
and
25% in
"low-quality"
journals. Both
the
median
and the
mode
of the
publication
year
are
1977.
The
distribution
resem-
bles
a
normal distribution. Gender
of the first
author could
not
be
determined
for two
studies.
Of the
remaining
121
studies,
55%
of first
authors were women.
The
percent
of
female
authors
was
indeterminate
for
four
studies.
In the
remaining studies,
38%
had
less than
50%
female
authors,
22% had 50%
female
authors,
and 40% had
more than
50%
female
authors. Thirty-
two
percent
had
only
male authors,
and 35% had
only
female
authors.
Subjects
These
123
studies
engaged
21,379
subjects with
the
number
of
subjects
per
study ranging
from 20 to
3,261.
When
we ex-
cluded three studies that
had
more than 1,000 subjects,
the
mean
number
of
subjects
was
173,
and the
median
was 96.
Nine
percent
of the
studies
had
only
female
subjects,
17%
had
only
male
subjects,
and 72% had
both
male
and
female
subjects.
Of
the 121
studies
in
which
the
age
of
the
subjects
was
reported
or
could
be
inferred,
4% of the
studies involved subjects
who
were
younger
than college
students,
74%
involved college-aged
sub-
jects,
and
21
%
involved subjects older than college students.
Procedures
Forty-five
percent
of the
studies
had
subjects evaluate only
a
male
or a
female
target person,
and the
remaining
55%
had
sub-
jects
rate
more than
one
target person.
In
addition
to the
stimu-
lus
material supposedly produced
by the
target person,
the
amount
of
information
given
to
subjects about
the
target person
ranged
from
only
a
name
(24%);
to a
one-sentence description
(5%);
to a
paragraph
of
information
(11
%);
to
more than
a
para-
graph,
a
resume,
or an
application,
showing
a
picture,
a
video,
or
the
actual person
or
hearing
an
audio
tape
(60%).
Twenty-
three percent
of the
studies
had
stimulus materials that could
be
characterized
as
written
work
or art
work,
30% as
behaviors,
11%
as
resumes
or
applications,
and
11%
as
biographies.
Six
studies
had
both
an
application
and an
essay,
and
seven studies
had
stimulus materials categorized
as
other
(6%).
Of
the 595 findings
across
the
studies,
81
%
were
ratings
of the
target person,
14%
were ratings
of the
stimulus material,
and
4%
reported
results
by
combining
the
dependent variables,
making
it
impossible
to
determine
how
many variables
were
ratings
of the
target person
or of the
stimulus materials.
Main
and
Simple
Effects
Main
effects.
Table
1
presents
the
results
for the
main
effect
of
ratings
of
male
versus
female
target persons. Four studies,
representing
20 findings, did not
analyze
for the
main
effect;
in
these cases,
no
assumptions about
the
effect
sizes could
be
made
to
accurately estimate
the d and
voting
scores
for the
main
effects
(in
contrast
to the
simple
effects
discussed
later). Thus,
the
results
for the
main
effect
are
based
on
119
studies
and 575
findings. The
number
of findings in
which
the
a
1
and the
voting
score
were
assigned
to
zero
were
360
(63%).
When
these values
were
excluded,
215
findings
remained
to be
analyzed.
Some
studies
had
only
male
or
female
subjects
or
only
mascu-
line,
feminine,
or
sex-neutral stimulus materials.
The
main
effects
from
these studies
would
be
similar
to the
simple
effects
from
studies that
had sex of
subject
or sex
role
of
stimulus
mate-
rial
as
independent variables. Hence,
the
main
effects
were
cal-
culated including
and
excluding these
studies.
The
results
were
essentially identical.
The
main
effects
presented
in
Table
1 in-
clude
these studies.
It may be
seen that
the
main
effects
were
all
negative
and all
very
small.
A
negative
value
indicates
a
lower
rating
of
female
than male target persons.
The
values
for the
homogeneity test
were
compared
to
chi-square
distributions with degrees
of
free-
1
Thefirsttwoeffectsizes(-.85
and-.75)comefrom
subjects'ratings
of
recommended
procedures
for an
employee
who was
offered
a job in
a
different
company.
The two
procedures
were
"Try
to
convince
opera-
tor to remain
with
organization"
and
"Don't
try to
influence
operator."
The
negative
effect
sizes
indicate
that
the
responses
were
more
favorable
for
the
male
target
person
than
the
female
target
person.
The
third
effect
size
(.62)
comes
from
subjects
rating
a
target
person's
request for
leave
of
absence
because
of
family
obligations
as
being
more
appropriate
for
female
target
people
than
for
male
target
people.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
415
Table
1
Main
Effect:
Ratings
of
Male
Target
Persons
Versus
Ratings
of
Female
Target
Persons
Summary
statistic
Effect
size
weighted
by
variance
N
M±95%CI
Homogeneity
test
Voting
score
%
female
%male
%
neither
Study
as
unit
of
analysis
119
-.07
± .03
303*
7
20
73
Finding
as
unit
of
analysis
575
-.05
± .01
1,394*
8
17
76
Excluding
findings
assumed
equal
to
zero
215
-.08
± .01
1,315*
20
45
35
Note.
Negative
effect
size
indicates
lower
evaluations
of
women.
CI =
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
Table
2
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction:
Sex
of
Subject
X Sex
of
Target
Person
Summary
statistic
Female
subjects
N
M±95%CI
Homogeneity
test
Male
subjects
N
M±9S%CI
Homogeneity
test
Study
as
unit
of
analysis
91
-.02
± .04
82
99
-.06
± .03
151*
Finding
as
unit
of
analysis
410
-.02
± .02
385
479
-.04
±.01
888*
Excluding
findings
assumed
equal
to
zero
145
-.06
± .03
370
188
-.07
± .02
853*
Note.
Negative
effect
size
indicates
lower
evaluations
of
women.
CI
=
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
dom
equal
to
n
- 1. The
significance
of
these
tests
indicated
that
the
main
effect
ds
were heterogeneous.
An
accepted
rule
of
thumb
is
that
a
a
1
of
.80
is
considered
large,
.50 is
moderate,
and .20 is
small
(Eagly,
1987).
As a
com-
parison,
average
ds for sex
differences
have
been
found
to
range
from
small
ds
such
as .09 to .32 for
influenceability
(Eagly
&
Carli,
1981)
to
large
ds
such
as
2.18
for sex
differences
in the
motor performance
of
throwing velocity (Thomas
&
French,
1985).
When
we
compared
the
different
ways
of
combining
the re-
sults across studies,
the
main
effect
rf was
smallest when
findings
were
the
unit
of
analysis
(-.05),
largest when
ds
assumed
to
equal zero
were
excluded
(-.08),
and
intermediate
when
calcu-
lated
with
study
as the
unit
of
analysis
(-.07).
The
voting scores,
the
tally
of
significant
and
nonsignificant
findings,
showed
that
when
study
or finding
with values
as-
sumed
to be
equal
to
zero
is the
unit
of
analysis, more than
7
in 10
results were nonsignificant.
When
the findings
were
sig-
nificant,
they
were
more
likely
to
indicate that
the man had
been rated more
favorably.
Sixty-three percent
of the findings
had
to be
assumed
to be
equal
to
zero. When these values
are
excluded,
45% of the
results indicated that male target persons
were
preferred
to
female
target
persons.
It
should
be
noted
again
that although this excluded studies
in
which
the
only
in-
formation
reported about
the finding was
that
it was
nonsig-
nificant,
it
included studies
in
which
the
only information
re-
ported about
the finding was
that
it was
significant.
Simple
effects.
Tables
2
through
15
present
the results for the
simple
effects.
The
number
of findings and
studies
per
simple
effect
decreased
from
Table
2 to
Table
15.
Smaller
ns
decrease
the
reliability
and
power
of the
analysis
(Strube
&
Miller,
1986).
As
for the
main
effect
ds, the
size
of the
effect
sizes
and
voting
scores
for the
simple
effects
tended
to be
smallest
when
finding
was
the
unit
of
analysis,
to be
intermediate
when
study
was the
unit
of
analysis,
and to be
largest
when
values assumed
to be
equal
to
zero
were
excluded.
The
voting scores
showed
that
the
majority
of findings and
studies
reported
nonsignificant
differ-
ences. When
we
excluded values assumed
to be
equal
to
zero,
the
majority
of
simple
effects
indicated that
men
were
rated
higher
than women. Unlike
the
main
effect
d,
some
of
the
simple
effects
were
homogeneous.
Sex
of
subject.
Contrary
to
predictions
of
greater bias among
male subjects, little
difference
was
found
between male
and fe-
male
subjects' ratings
of
male
and
female
target persons
(see
Table
2).
However,
the findings for
male subjects were substan-
tially more heterogeneous than
for
female
subjects.
Sex
role
orientation
of
stimulus
material.
As was
mentioned
earlier, some studies only
had one
type
of
stimulus
material.
Such
studies kept
the sex
role orientation
of the
stimulus
mate-
rial constant across subjects
by
presenting
all
subjects with
only
masculine, only
feminine,
or
only sex-neutral stimulus mate-
rial.
The
simple
effects
for sex
role orientation
of the
stimulus
material
were calculated
both
excluding these studies
and in-
cluding
these studies.
The
results were essentially identical
whether
or not
these studies
were
included. Table
4
presents
the
results with these studies included.
Table
3
Kiting
Score
for
Simple
Effects
From
Interaction:
Sex of
Subject
X Sex
of
Target
Person
Summary
statistic
Female
subjects
N
%
female
%male
%
neither
Male
subjects
N
%
female
%male
%
neither
Study
as
unit
of
analysis
91
9
15
76
99
7
22
70
Finding
as
unit
of
analysis
410
9
13
78
479
8
17
75
Excluding
findings assumed
equal
to
zero
145
25
36
39
188
21
42
37
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

416
SWIM,
BORGIDA,
MARLTVAMA,
AND
MYERS
Table
4
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction:
Sex
Role
Orientation
of
Stimulus
Table
6
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction:
Status
of
Target
Person
X Sex
of
Target
Person
Material
X Sex
of
Target Person
Summary
statistic
Feminine
sex
role
N
M±95%CI
Homogeneity
test
Masculine
sex
role
N
Af±95%CI
Homogeneity
test
Sex-neutral
sex
role
N
M±9S%Cl
Homogeneity
test
Study
as
unit
of
analysis
44
-.01
± .05
75'
46
-.12
±.05
75*
28
-.13
±.06
64*
Finding
as
unit
of
analysts
202
-.03
± .03
403*
226
-.10
±.02
424*
106
-.15
±.03
301*
Excluding
findings
assumed
equal
to
zero
73
-.08
± .04
391*
81
-.25
± .04
299*
45
-.32
±.04
185*
Summary
statistic
High
com
ence
N
Af±95%CI
Homogeneity
test
Medium
competence
N
M±95%CI
Homogeneity
test
Low
competence
N
Af±95%CI
Homogeneity
test
Study
as
unit
of
analysis
30
-.08
± .07
56*
6
-.18
±.20
4
29
-.07
± .06
21
Finding
as
unit
of
analysis
95
-.05
± .04
149*
14
-.13±.13
7
94
-.04
± .04
74
Excluding
findings
assumed
equal
to
zero
32
-.21
±.08
130*
5
-.34
±.22
1
31
-.19
±.08
57
Note.
Negative
effect
size
indicates
lower
evaluations
of
women.
CI
:
confidence
interval,
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
Note.
Negative
effect
size indicates
lower
evaluations
of
women.
CI
:
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
All
the
methods
of
combining
results
showed, consistent with
predictions,
that
a
greater
difference
in
ratings
of
male
and fe-
male
target persons (with
men
being
rated
more
favorably)
oc-
curred
when
the
stimulus
materials
were masculine
in
compari-
son
with
when
the
stimulus materials were
feminine.
However,
the
largest
effect
size
occurred
when
the
stimulus material
was
sex
neutral. Furthermore,
it
should
be
kept
in
mind that
the
cb
were
still small.
The
distribution
of the
effect
sizes
was
heteroge-
neous
for the
feminine,
masculine,
and
sex-neutral
sex
roles.
Competence
of
target person.
Six
studies
(14 findings) in-
cluded
medium-competent
target persons
in
addition
to
high-
and
low-competent
target persons (see
Table
6).
Because
of the
small
number
of
studies
and findings
with
a
target person
of
medium
competence,
it was
more
meaningful
to
examine
the
differences
between evaluations
of
male
and
female
target per-
sons
who
were
high
and low in
competence.
The
ds
were essentially
identical
whether
the
target person
was
of
high
or of low
competence (see Table
6).
However,
the
<fc
were
heterogeneous
for
target persons
of
high competence
and
Table
5
Voting
Score
for
Simple
Effects
From
Interaction:
Sex
Role
Orientation
of
Stimulus
Material
X Sex
of
Target Person
Summary
statistic
Feminine
sex role
N
%
female
%male
%
neither
Masculine
sex
role
N
%
female
%male
%
neither
Sex-neutral
sex
role
N
%
female
%male
%
neither
Study
as
unit
of
analysis
44
22
16
62
46
7
36
57
28
1
23
75
Finding
as
unit
of
analysis
202
16
12
72
226
8
21
71
106
4
27
69
Excluding
findings
assumed
equal
to
zero
73
45
33
22
81
22
59
18
45
9
64
27
Table?
Kiting
Score
for
Simple
Effects
From Interaction:
Status
of
Target
Person
X Sex
of
Target Person
Summary
statistic
High
competence
N
%
female
%
male
%
neither
Medium
competence
If
%
female
%male
%
neither
Low
competence
N
%
female
%male
%
neither
Study
as
unit
of
analysis
30
8
17
75
6
53
47
0
29
22
2
76
Finding
as
unit
of
analysis
95
7
13
80
14
0
43
57
94
5
15
80
Excluding
findings
assumed
equal
to
zero
32
22
41
41
5
0
100
0
31
16
45
39
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
417
Tables
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From Interaction:
Status
of
Stimulus
Material
X Sex
of
Target Person
Summary
statistic
High
quality
N
Af±95%CI
Homogeneity test
Medium
quality
N
M±95%CI
Homogeneity
test
Low
quality
N
M±95%CI
Homogeneity
test
Study
as
unit
of
analysis
10
-.02
±.11
30*
2
-.24
±.34
5
10
-.05
±.11
6
Finding
as
unit
of
analysis
38
.01
± .05
55*
7
-.19
±.18
17*
37
-.06
± .05
34
Excluding
findings
assumed
equal
to
zero
10
-.16±.17
51*
3
-.87
± .37
0
14
-.22
±.10
22
Note.
Negative
effect
size indicates
lower
evaluations
of
women.
CI
=
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
Table
10
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction: Physical
Attractiveness
of
Target
Person
X Sex
of
Target Person
Summary statistic
High
physical attractiveness
N
M±9S%Cl
Homogeneity
test
Medium
physical attractiveness
N
M±9S%Cl
Homogeneity
test
Low
physical attractiveness
N
M±9S%C1
Homogeneity
test
Study
as
unit
of
analysis
12
-.15±.12
9
5
-.11±.17
2
12
-.05
±.12
7
Finding
as
unit
of
analysis
62
-.07
±
.05
38
19
-.09
±.09
19
62
-.02
±.05
40
Excluding
findings
assumed
equal
to
zero
11
-.32
±.12
16
4
-.27
±.16
12
12
-.08
±.11
38*
Note.
Negative
effect
size indicates
lower
evaluations
of
women.
CI =
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous.
homogeneous
for
target persons
of
low
competence.
And
again,
the
voting
scores
indicated
that
the
majority
of findings and
studies
found
no
significant
differences
and
that
the ds
were
small.
Only
two
studies (seven
findings)
included
medium-quality
stimulus
material (see Table
8).
Again,
it was
more
meaningful
to
examine just
the
high-
and
low-quality
conditions.
As
with
the
simple
effects
for
competence,
the
sizes
of the
ds
for
high
and low
quality were quite similar. Also,
the
ds
for the
high-
Table
9
Voting
Score
for
Simple
Effects
From Interaction:
Status
of
Stimulus
Material
x Sex
of
Target Person
Summary
statistic
High
quality
N
%
female
%male
%
neither
Medium
quality
N
%
female
%male
%
neither
Low
quality
N
%
female
%male
%
neither
Study
as
unit
of
analysis
10
10
12
78
2
0
50
50
10
0
12
88
Finding
as
unit
of
analysis
38
3
10
87
7
0
43
57
37
0
10
90
Excluding
findings
assumed
equal
to
zero
10
10
40
50
3
0
100
0
14
0
29
71
quality
stimulus
materials
were more heterogeneous than
the ds
for
the
low-quality
stimulus
materials.
Physical attractiveness
of
target
persons.
Only
five
studies
(19
findings) had a
medium
level
of
physical attractiveness. Hence,
comparisons were only made between target persons
who
were
high
and who
were
low in
physical attractiveness (see Tables
10
and
11).
All but one of the
results
indicated
that
the findings
were
homogeneous. Contrary
to the
prediction
that gender
Table
11
fating
Score
for
Simple
Effects
From
Interaction: Physical
Attractiveness
of
Target Person
X Sex
of
Target Person
Summary
statistic
High
physical attractiveness
N
%
female
%male
%
neither
Medium
physical attractiveness
N
%
female
%male
%
neither
Low
physical attractiveness
N
%
female
%
male
%
neither
Study
as
unit
of
analysis
12
1
38
61
5
3
20
77
12
10
25
65
Finding
as
unit
of
analysis
62
2
18
81
19
5
16
79
62
5
11
84
Excluding
findings
assumed
equal
to
zero
11
10
91
0
4
25
75
0
12
25
58
17
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

418
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
Table
12
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction:
Bern
Sex
Role
Inventory
Table
14
Effect
Size
Weighted
by
Variance
for
Simple
Effects
From
Interaction:
Race
of
Target
Person
X Sex
of
Target
Person
oj
Subjects
X Sex
oj
Target
Person
Summary
statistic
Feminine
subjects
N
Homogeneity
test
Androgenous
subjects
N
A/±95%CI
Homogeneity
test
Study
as
unit
of
analysis
1
-.16
±.64
0
2
-.17
±.33
0
Finding
as
unit
of
analysis
14
-.15±.17
12
24
-.16±.10
25
Excluding
findings
assumed
equal
to
zero
3
-.76
±
.38
0
8
-.35±.15
14
Summary
statistic
N
M±95%Cl
Homogeneity
test
Non-White
N
M±95%CI
Homogeneity
test
Nntp
NiHffltivp
pffrrt
<j
Study
as
unit
of
analysis
7
-.003
±.14
16*
7
-.11
±.12
4
Finding
as
unit
of
analysis
14
-.01
±.11
18
14
-.08
± .09
6
findings
assumed
equal
to
zero
4
-.03
+
.17
18
4
-.16
+
.13
4
Note.
Negative
effect
size
indicates
lower
evaluations
of
women.
CI
=
confidence
interval.
confidence
interval.
*
p <
.05,
that
is,
efiect
sizes
are
heterogeneous.
would
influence
evaluations
when
the
target persons
were
unat-
tractive
but not
when they were attractive, there
was a
tendency
for
the
effect
sires
to be
larger
for
more attractive target persons.
Again,
the
effects
were
small
and
similar
for
target persons
who
were
high
and who
were
low in
physical attractiveness.
Bern
Sex
Role
Inventory.
The
number
of
studies
and
findings
assessing
the
effect
of sex
role orientation
of the
subjects
was
quite small,
and
hence
the
reliability
of
these results
is
question-
able.
No
studies
had
masculine subjects,
and
only
one
study
(one
finding) had
undifferentiated
(i.e., neither
feminine,
mas-
culine,
or
androgenous)
subjects. Hence results were
only
re-
ported
for the
feminine
and
androgenous subjects. Contrary
to
expectations,
feminine
subjects were more
likely
to
give
more
favorable
ratings
to
female
target persons,
and
androgenous
subjects
were more
likely
to
give
more
favorable
ratings
to
male
target persons.
However,
these
effect
sizes
were
still quite small
and
came
from
a
limited number
of
studies.
So,
although they
were
homogeneous,
the
results
may not be
reliable.
Table
13
Voting
Score
for
Simple
Effects
From
Interaction:
Bern
Sex
Role
Inventory
of
Subjects
X Sex
of
Target
Person
Summary
statistic
Feminine
subjects
N
%
female
%male
%
neither
Androgenous
subjects
N
%
female
%male
%
neither
Study
as
unit
of
analysis
1
14
0
86
2
0
25
75
Finding
as
unit
of
analysis
14
14
0
86
24
0
21
80
Excluding
findings
assumed
equal
to
zero
3
67
0
33
8
0
62
38
Race
of
target
person.
The
number
of findings and
studies
was
also quite small
for the
simple
effects
associated
with
race
of
the
target person (see Tables
14 and
15). There appeared
to
be
a
slight tendency
for
comparisons between
non-Whites
to
yield
larger
effect
sizes than comparisons between Whites.
Yet
again,
these
effect
sizes
were
small,
and the findings
were
based
on
a
limited number
of
studies.
Correlations
Correlations
of the
main
effect
ds
with
continuous variables
are
presented
in
Table
16.
These correlations
are
presented
in
the
form
of
standardized regression
coefficients
(see Hedges
&
Olkin,
1985).
Positive
coefficients
indicate that gender bias
to-
ward
women
decreases
as the
other variables increase. None
of
the
coefficients
were
significant,
and
none
were
large.
Breakdown
of
Effect
Sizes
Main
effect
ds
were
calculated
for
different
categorical
levels
of
a
series
of
independent variables.
The
breakdown
of the
Table
15
Voting
Score
for
Simple
Effects
From
Interaction:
Race
of
Target
Person
X Sex
of
Target
Person
Summary
statistic
White
N
%
female
%male
%
neither
Non-white
If
%
female
%malc
%
neither
Study
as
unit
of
analysis
7
14
36
50
7
14
21
64
Finding
as
unit
of
analysis
14
7
21
71
14
7
14
79
Excluding
findings assumed
equal
to
zero
4
25
75
0
4
25
50
25
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
419
Table
16
Standardized
Regression
Coefficients
for
Relating
Variables
to
Weighted
Effect
Sizes
Excluding
Study
as
Finding
as findings
unit
of
unit
of
assumed
equal
analysis
analysis
to
zero
Variable
beta
N'
t*
beta
N
1
t*
beta
N"
%
female
authors
.00 115 0 .07 553 .16 .07 200 .16
Publication
year
.00
1190
.06575.15
.10215
.20
Number
of
independent
variables
associated
with
target
person
.00 119 0 .08 575 .18 .14 215 .24
Number
of
target
persons
rated
.00 119 0
-.06
575
.16
-.11
215
-.21
Note.
Negative
effect
sizes
indicate
lower
evaluations
of
women.
*
Sample
size.
b
/
test
weighted
ds
by
the
type
of
question wording
was
only
done
when
finding
was
the
unit
of
analysis.
Both
within-group
and
be-
tween-groups
homogeneity
tests
were
calculated.
This
allows
one
to
examine whether
the
effect
sizes within each category
can
be
considered identical
and
whether
the
effect
sizes
across
categories
were
different.
These results
are
presented
in
Tables
17,18,
and
19.
All
of the
effect
sizes within
the
different
categories remained
small,
and
most
were
heterogeneous.
Even
though
the
effect
sizes
were
heterogeneous,
a
few
of the
between-groups
homoge-
neity
tests
were
significant,
indicating that there
were
differ-
ences between
a few of the
categories.
However,
none
of the
effect
sizes
were
large.
The
categories
for age of
subject
differed,
but the
effect
sizes
within
the
categories
were
heterogeneous.
The
pattern
of
results
differed
depending
on
whether
finding or
study
was the
unit
of
analysis. When
finding was the
unit
of
analysis, with
or
without
zero
values,
a
greater
difference
in
ratings
was
found
for
younger
subjects;
when
study
was the
unit
of
analysis,
greater
differences
were
found
for
older subjects.
Like
the age of the
subjects,
the
categories
for
percentage
of
female
authors
differed,
but the
effect
sizes were heterogeneous
within
categories.
The
trend
was for
greater
differences
in
rat-
ings
when there
were
more than
50%
female
authors. Similarly,
greater
differences
in
ratings
were
found
when
the first
author
was
a
woman.
However,
though
the
categories
differed
when
finding
was
the
unit
of
analysis,
the
effect
sizes
were
again heter-
ogeneous within categories
for
all
combinatorial methods.
The
categories
for
year
of
publication
were
only heteroge-
neous when values assumed
to
equal zero
were
excluded
and
tended
to be
heterogeneous within
a
category.
The
trend
was to
find
fewer
differences
in
ratings
in the
middle years than
in the
earlier
or
later years.
The
categories
for
journal quality
were
heterogeneous,
but the
effect
sizes within categories
were
also
heterogeneous.
The
trend
was to find
greater
differences
in
rat-
ings
when
the
journals
were
of
either high
or low
quality.
The
categories
differ
for
the
amount
of
information
and a
few
of
the
categories
were
homogeneous.
The
trend
was to find
larger
differences
when
less information
was
given.
The
number
of
independent variables
was
also
used
as a
measure
of
amount
of
information.
The
same trend
of
greater
differences
with less
information
was
found
here.
However,
the
categories
are
hetero-
geneous,
and as
mentioned previously,
the
correlation
between
number
of
independent variables
and
effect
size
is
small.
The
number
of
target
persons rated
had
little
effect
on
ratings.
The
categories
for
type
of
stimulus material
differed,
though
most
of the
effect
sizes
within
categories
were
heterogeneous.
There tended
to be
greater
differences
when applications
were
used
as the
stimulus material.
The
question wording
had
little
effect
on the
effect
sizes.
Discussion
To
assess
the
robustness
of findings
from
Goldberg's
(1968)
original
study,
we
reviewed
studies that
were
paradigmatically
similar.
The
task situation
was
typically
one in
which subjects
were
given
some evidence about
a
hypothetical target person's
abilities,
for
example,
in the
form
of an
essay purportedly writ-
ten by the
target person
or a
target person's
job
application.
The
subject
was
asked
to
evaluate
the
target person,
the
product that
the
target person
had
generated,
or
both.
Some subjects
were
only
told
that
the
target person
was a
man; others
were
told that
the
target person
was a
woman.
In
other studies, subjects evalu-
ated more than
one
target person, some
of
whom
were
men and
some
of
whom
were
women.
One
hundred
and six
published
articles
using
North
Ameri-
can
subjects
and
reporting
on
123
studies
that
used this experi-
mental paradigm
were
selected
and
reviewed. About three-
fourths
of
these studies
were
published between
1974
and
1981.
There
were
approximately equal numbers
of
male
and
female
authors. Most
of
the
studies used college students
as
well
as
both
male
and
female
subjects. About
half
of the
studies
had
subjects
evaluate
only
male
or
only
female
target persons,
and the
other
half
had
subjects evaluate both male
and
female
target persons.
The
stimulus materials
were
approximately equally likely
to be
written
work
or art
work,
behavior,
resumes, applications
for
jobs,
or
short biographies
of
the
person.
Differences
between
the
average
ratings
of men and
women
were
analyzed across studies
by
tallying voting scores
and by
combining
effect
sizes.
The
unit
of
analysis
was
either
the
study
or the finding. The
study
is the
best
unit
of
analysis
because
within
each study
the
dependent variables
are
likely
to be
inter-
dependent.
An
upper limit
on
size
of the
difference
was
ob-
tained
by
using
finding as the
unit
of
analysis
and
excluding
those
findings in
which
no
difference
between
men and
women
had to be
assumed because
of
limited reporting
in the
studies.
We
found
that
the
size
of the
difference
in
ratings between
female
and
male target persons
was
extremely small
(-.07,
ac-
counting
for
less than
1
%
of the
variance, with study
as the
unit
of
analysis,
and
-.OS,
accounting
for
less than
1% of the
vari-
ance,
with
finding as the
unit
of
analysis). Excluding those
find-
ings
for
which
we had
insufficient
information
(i.e.,
values
were
assumed
to be
equal
to
zero),
the
effect
size
only
reached
a
value
of-.08,
clearly less than
1% of the
variance. This
difference
is
even
smaller than those typically
reported
for
gender
differences
in
social behavior
(e.g.,
Eagly
&
Carli,
1981;
Hyde
&
Linn,
1986;
see
Deaux,
1985).
Furthermore, when merely examining
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

420
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
Table
17
Breakdown
of
Main
Effect
Size
Weighted
by
Variance
(Study
as
Unit
of
Analysis)
Variable
Age
of
subject
Younger
than
college
age
College
age
Older
than
college
age
%
female
authors
<50%
50%
>50%
Sex
of first
author
Female
Male
Publication
year
1968-1973
1974-1978
1979-1985
Journal
quality
High
Medium
Low
Amount
of
information
about
target
person
Only
name
One
sentence
One
paragraph
More
than
one
paragraph
Number
of
independent
variables
associated
with
target
person
One
More
than
one
Number
of
target
persons
rated
One
More
than
one
Type
of
stimulus
material
Written
work
Behavior
Application
Biography
Application
and
essay
Other
Af±95%CI
-.01
±.15
-.07
±.03
-.11 +.05
-.08
±.04
-.01
±.05
-.12
±.04
-.09
±.04
-.06
±.04
-.10
±.08
-.10
±.04
-.04
±.04
-.14
±.06
-.04
±.03
-.11
±.05
-.12
±.06
-.05
±.13
-.001
±.06
-.08
±.03
-.13
±.04
-.04 +.03
-.07 +.04
-.08 +.03
-.04
±.05
-.07
±.05
-.10
±.06
+.03
±.08
-.04 +.09
-.16 +.11
N
5
85
23
42
26
47
65
52
14
60
45
21
68
30
27
6
13
73
54
65
54
65
28
34
29
13
6
7
Homogeneity
within
group
5
254*
42*
82*
73*
132*
147*
154*
12
198*
88*
38*
128*
129*
38
5
16
207*
148*
145*
152*
151*
28
155*
19
26*
10*
20*
Homogeneity
between
groups
7**
16**
2
5
8"
37**
10
0
45**
Note.
Negative
effect
sizes
indicate
lower
evaluations
of
women.
CI
=
confidence
interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous
within
category.
**
p <
.05,
that
is,
average
effect
sizes
are
significantly
different
between
categories.
the
number
of
studies that
found
significant
and
nonsignificant
results,
the
majority
of
studies
and findings
were
nonsignificant.
We
also
found,
however,
that
the
distribution
of the
effect
sizes
was
heterogeneous; that
is,
although
the
mean
effect
size
was
small, there
was a lot of
variability among
the
effect
sizes.
Although
the
potential moderators
of
effect
size that
we
consid-
ered
do not
fully
account
for
this
variability, other
moderators
not
analyzed
in
this
review
may be
able
to
partition
the
effect
sizes
into
homogeneous categories.
Variables
that
had
been manipulated
in the
studies
reviewed,
differences
in
characteristics associated with
the
methodologies
used
in
these studies,
and
differences
associated with
the
articles
themselves
were
examined
in an
attempt
to
identify
variables
that moderated
the
size
of
the
effects.
Even
when
one
takes into
consideration these variables,
however,
the
effect
sizes remain
extremely
small.
For
instance,
even
when
the
stimulus material
tapped
a
traditionally masculine
sex
role orientation
as
opposed
to a
traditionally
female
sex
role orientation,
the
effect
size
was
still
only
around
-.10.
Similar conclusions
were
found
for
other
variables such
as sex of
subject
and
competence
of the
target
person.
However,
like
the
effect
sizes
for the
main
effect,
the
distribu-
tion
of
weighted
effect
sizes
is
heterogeneous
for
male subjects,
sex-neutral
and
masculine stimulus materials,
highly
compe-
tent target people,
and
high-quality
stimulus material.
Thus,
it
appears that
further
research
will
be
necessary
to
clarify
the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER STEREOTYPES
421
TablelS
Breakdown
of
Main
Effect
Size
Weighted
by
Variance
(Finding
as
Unit
of
Analysis)
Variable
Age
of
subject
Younger
than
college
age
College
age
Older than college
age
%
female
authors
<50%
50%
>50%
Sex
of first
author
Female
Male
Publication year
1968-1973
1974-1978
1979-1985
Journal quality
High
Medium
Low
Amount
of
information
about
target person
Only name
One
sentence
One
paragraph
More than
one
paragraph
Number
of
independent
variables
associated
with
target person
One
More than
one
Number
of
target
persons
rated
One
More than
one
Type
of
stimulus material
Written
work
Behavior
Application
Biography
Application
and
essay
Other
Question wording
Rate
of
target person
Rate
of
stimulus material
Rate
of
both
M±
95% CI
-.06
±.08
-.06 +.01
-.03
±.02
-.04
±.01
-.01
±.03
-.09
±.02
-.06
±.02
-.04 ±.0t
-.07
±.05
-.004
±.01
-.06
±.02
-.12
±.03
-.03
±.01
-.07
±.03
-.13
±.04
-.02
±.09
-.01
±.04
-.05
±.01
-.14
±.02
-.01
±.01
-.03
±.02
-.05 +.01
-.03
±.03
.00
±.02
-.13
±.02
-.06
±.05
-.18
±.06
-.11
±.05
-.05
±.01
-.04
±.03
-.01
±.07
N
13
413
143
230
115
208
310
250
40
351
184
121
341
113
74
14
47
440
233
342
269
306
86
216
198
41
16
18
472
81
22
Homogeneity Homogeneity
within
group,
between groups,
<ff-k-p
df=p-l
15
10**
880*
489*
620*
62**
180*
532*
665*
20**
709*
31
5
954*
406*
309*
33**
829*
223*
152*
28"
10
78*
1126*
664*
111**
619*
433*
4
957*
82
121**
518*
379*
142*
40*
112*
1252*
2
124*
16
Note.
Negative
effect
sizes indicate lower evaluations
of
women.
CI =
confidence interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous within category.
**
p <
.05,
that
is,
average
effect
sizes
are
significantly
different
between
categories.
conditions under which these variables will
be
associated
with
large
versus
small
effect
sizes.
Correlations with
and
breakdowns
of the
effect
size
were
used
to
test
for the
influence
of
methodological
differences
and
char-
acteristics
of
the
articles
themselves.
All
of
the
correlations
were
small
and
nonsignificant,
and
none
of the
breakdowns
of the
mean
effect
size yielded large
effect
sizes.
There
was
some
indication,
however,
that women will
be
rated less
favorably
than
men
when
less information
is
pre-
sented.
The finding
that
the
amount
of
information provided
may
influence
effect
size
is
consistent with
Tosi
and
Einbender's
(1985)
quantitative
review
of a
smaller
set of
these studies. Tosi
and
Einbender
used voting scores
to
assess
the
effect
of the
amount
of
information provided
on sex
bias,
and
they
used
a
continuous variable that represented
the
number
of
indepen-
dent variables manipulated
and a
subjective
assessment
of all
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

422
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
Table
19
Breakdown
of
Main
Effect
Size
Weighted
by its
Variance
(Finding
as
Unit
of
A
nalysis
Excluding
Findings
Assumed
Equal
to
Zero)
Variable
Age
of
subject
Younger
than
college
age
College
age
Older than college
age
Percent
female
authors
<50%
50%
>50%
Sex
of first
author
Female
Male
Publication
year
1968-1973
1974-1978
1979-1985
Journal quality
High
Medium
Low
Amount
of
information
about target person
Only
name
One
sentence
One
paragraph
More than
one
paragraph
Number
of
independent variables
associated
with target person
One
More than
one
Number
of
target persons
rated
One
More than
one
Type
of
stimulus material
Written
work
Behavior
Application
Biography
Application
and
essay
Other
Question wording
Rate
of
target person
Rate
of
stimulus material
Rate
of
both
M±95%
CI
-.14±.12
.00
±.02
-.04
±.02
-.06
± .02
-.02
± .05
-.26
± .04
-.20
±.03
-.05
± .02
-.28
±.10
-.07
± .02
-.15
±.03
-.26
± .04
-.05
± .01
-.23
± .05
-.38
± .06
-.04±.l2
-.02
± .07
-.08
± .01
-.23
± .02
-.02
± .02
-.11
±.04
-.08
± .02
-.14
±.06
.00
± .02
-.25
± .03
-.11
±.06
-.23±.l6
-.17
±.06
-.08
± .01
-.13
±.06
-.02
±.14
N
6
131
76
105
23
72
95
108
11
131
73
54
133
28
25
8
14
168
91
124
79
136
20
88
69
19
9
10
184
26
5
Homogeneity
within group
12
720*
479*
602*
180*
386*
527*
698*
9
922*
352*
216*
816*
166*
64
10
78*
1011*
512*
610*
402*
910*
47*
509*
253*
136*
30*
102*
1186*
110*
16*
Homogeneity
between groups
104**
147**
90**
32**
117**
152**
193*
3
238**
3
Note.
Negative
effect
sizes
indicate
lower evaluations
of
women.
CI
=
confidence interval.
*
p <
.05,
that
is,
effect
sizes
are
heterogeneous within category.
**
p <
.05,
that
is,
average
effect
sizes
are
significantly
different
between
categories.
the
information provided. They reported stronger support
for
the
hypothesized relationship than
what
we
found
here. This
finding is
also consistent
with
research
by
Locksley
et
al.
(1980),
who
found
that
when
no
information
was
presented about
a
target
person, subjects made stereotypical judgments.
However,
when
behavior relevant
to the
trait being judged
was
presented,
subjects
did not
appear
to use
their stereotypes (also
see
Rasin-
ski,
Crocker,
&
Hastie,
1985).
Our
review
included
only
studies
in
which
the
target
had
ostensibly produced
a
behavior
or
prod-
uct.
If
we
had
included studies that provided subjects
with
only
the
name
of a
target person, perhaps
larger
effect
sizes
would
have
been obtained.
There
was
also some indication
of
greater
bias
when
the
stim-
ulus
material
was a
resume
or
application. There
may be
some-
thing
unique
to
studies
in a job
context;
the
three
findings
that
were
excluded because they
were
outliers came
from
a job
con-
text.
The
significance
of
this
finding
should
not be
overesti-
mated,
however,
although there
are
indications
of
small
differ-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
423
ences
in the
magnitude
of
effect
sizes between categories,
it
must
be
stressed
that
the
effect
sizes
are
uniformly
very
small
within
categories.
Given
the
lack
of
subsequent
support
for
Goldberg's original
conclusion
that
women
will
be
evaluated less
favorably
than
men
even
for
identical work,
why
have
Goldberg's study
and
subsequent replications been
so
frequently
cited
as
evidence
for
this
conclusion?
2
Several explanations
may be
advanced,
any
or all of
which
are
plausible. First, unlike quantitative
reviews,
qualitative
reviews
are
limited
in
their ability
to
summarize
a
large
number
of
studies, thereby restricting their accuracy.
For
instance, readers
may
have
been
overly
influenced
by the
sig-
nificance
of the findings and
simply unaware that
the
size
of
the
effects
across studies
was not
large. Second, researchers
may
have
lacked knowledge about other replications because
of in-
complete literature
reviews
or
because
of the
diverse array
of
journals that contain such studies. Third, literature
reviews
of-
ten
selectively emphasize
significant
results
and
trends
that,
if
due to
chance,
would
cause
inflated
perceptions
of the
strength
of
the
evidence supporting
the
tendency
to
discriminate
against
women.
In
fact,
nearly
two
thirds
of the findings for the
main
effect
had to be
assumed
to be
equal
to
zero
because their non-
significant
findings
were
not
accompanied with
sufficient
statis-
tical information.
Fourth,
it
could
be
argued that
in the
early
years
of the
wom-
en's movement, there were
no
clear experimental demonstra-
tions
of
sex
discrimination
in
social
psychology.
Goldberg's
study
filled
this
gap by
providing straightforward evidence.
As
Deaux
(1985)
suggested,
at the time
when
Goldberg's study
was
conducted, there
was
"satisfaction with simple (yet dramatic)
demonstrations
of
differential
evaluation
and
reliance
on
per-
functory
reference
to
gender
stereotypes"
(p.
66).
A fifth and
related explanation
is
that textbook writers generally
prefer
to
use
compelling illustrations such
as
Goldberg's experiment
to
buttress their arguments. They
may
have
felt
that
this experi-
ment
in
particular would
be
interesting
to
students because
it
was
simple
to
understand
and had
obvious implications. Also
some
authors
may
have
been unaware
of or may
have
felt
that
it
was too
complicated
to
address
the
numerous replications
that
followed
Goldberg's original study.
What
implications does
the
lack
of
empirical support
for
Goldberg's conclusions
have
for
research
on
gender
stereotyp-
ing?
One
obvious
but
erroneous implication might
be
that
peo-
ple
do not
hold gender stereotypes
or use
such stereotypes when
thinking
about others.
However,
recent research continues
to
demonstrate that people
are
willing
to
express gender stereo-
types (e.g., Deaux
&
Lewis,
1983; Martin, 1986)
and
that peo-
ple
still
follow
traditional
sex
roles (e.g.,
Lueptow,
1980).
One
could also argue that
the
paradigm
we
reviewed
is not
ideal
for
revealing gender stereotypes. First, subjects
in
psychol-
ogy
experiments
are
likely
to be
trying
to
present themselves
as
unbiased individuals. Second,
the
widespread citing
and de-
scribing
of
Goldberg's
findings in
introductory
psychology
texts,
in
conjunction with society's attention
to sex
discrimina-
tion,
may
well
sensitize most subjects
to the
particular para-
digm
used.
In
effect,
it is
possible that
the
approaches
we re-
viewed
are
neither blatant enough
to
command
sttreotypic
evaluations
nor
subtle enough
to
expose
any
real biases that
subjects
may
harbor. Such biases
may
emerge
most
clearly
only
when
the
array
of
available information
is
such that there
are
other plausible explanations
for
subjects' judgments
and
behav-
iors (and subjects even
may be
unaware
that
they
are
using
ste-
reotypic
information
in
their judgments). Note that
the
some-
what
stronger
effects
for the job
resume
paradigm
in
this
review
could
be
interpreted
as
supporting such
a
position.
On the
other
hand, because
the
studies
we
reviewed
do not
vary greatly
in the
amount
of
information available (e.g., virtually none matched
the
complexity
of
real-life
hiring
or job
evaluation decisions),
the
previously mentioned explanation awaits
a
more conclusive
empirical
test.
A
more substantive interpretation
of our
meta-analytic
find-
ings
is
that
people's
evaluations
of men and
women
are a
com-
plex
function
of a set of
factors
that
influence
the
process
of
gender
stereotyping.
In the
studies reviewed
in
this
meta-analy-
sis,
factors
that might
have
influenced
subjects'
use of
global
gender
stereotypes (but
were
not
systematically examined
in
these studies) included
(a) the
content
of
gender stereotypes
(e.g.,
the
specific
components
and
subtypes that
have
been
acti-
vated
in a
situation [Deaux
&
Lewis, 1984; Deaux, Winton,
Crowley,
&
Lewis,
1985]);
(b)
the
information
presented
about
the
target person
or the
stimulus material (e.g.,
the
perceived
diagnosticity
of the
stimulus material
for
the
evaluation
[Ginos-
sar
&
Trope,
1980]);
(c) the
interaction between
the
stereotype
and
the
information presented (e.g., whether
the
stimulus mate-
rial
even
activates
a
gender
stereotype
[Rothbart
and
John,
1985]);
(d) the
goals
or
motivations
of the
subjects (e.g.,
the
sub-
jects'
level
of
involvement
in the
evaluation [Ginossar
&
Trope,
1987;
Neuberg
&
Fiske,
1987]);
and (e) the
task demands (e.g.,
the
complexity
of the
task
[Bodenhausen
&
Lichtenstein,
1987]),
or
whether
the
task activates
a
more clinical/narrative
judgmental orientation
or a
more scientific/paradigmatic ori-
entation
[Zukier,
1985;Zukier&Pepitone,
1984]).
How
such
factors
might account
for the
small
effect
sizes
found
in the
present
meta-analysis
can be
illustrated
by
consid-
ering recent gender stereotyping research
on
levels
of
categori-
zation,
along with research
on the
integration
of
prior gender
beliefs
and
specific
case information.
Predictions about
the
evaluation
of a
specific
target
person
may
be
based
on
more
specific
categories than global gender
stereotypes about
men and
women.
For
instance, Deaux
et
al.
(1985)
found
that subjects' descriptions
of
subtypes
of
women
and men
such
as
housewife
or
businesswoman,
and
blue-collar
working
manor
businessman
were
as
rich
as the
broader catego-
ries
of
woman
and
man.
The
activation
of
these more
specific
constructs
may not be
assessed
in
studies
in
which
the
depen-
dent measures were designed
to
detect only
the
effects
of
more
general
stereotypes.
In
addition,
social roles
or
other social cate-
gories such
as
race
or
occupation
may
also
have
been activated
in the
judgment setting
and may
have
been more
influential
than gender categories (e.g.,
Taylor,
1980).
In the
studies
re-
viewed,
subtypes
may
have
matched
the
target information
more closely than broader stereotypes about
men and
women.
2
Even
Roger
Brown
(1987),
in his
insightful
social
psychology
text,
concluded
his
chapter
on
stereotyping
by
commenting
that
"the
general
result,
varying
with
the
date
of the
study,
the
nature
of the
job,
is
that
(no
surprise)
discrimination
in
favor
of men
exists
and
(some
surprise)
exists
whether
the
evaluators
are
themselves
male
or
female"
(p.
601).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

424
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
Furthermore, there
may not
have
been evaluative
differences
between
the
subtypes.
For
example, subjects
may
have
subtypes
about high-status
men and
high-status women,
but the
content
of
these subtypes
may not
differ
(Deaux
&
Lewis,
1984).
3
Even
though male
and
female
subtypes
with
different
im-
plications
may be
activated, subjects' judgments
may be
influ-
enced
more
by the
case
information
presented than
by the
prior
beliefs
associated with
the
subtypes.
Locksley
et
al.
(1980)
dem-
onstrated that subjects disregarded global gender stereotypes
when
individuating information
was
presented.
Locksley
et al.
found
that case information
nondiagnostic
of
gender
stereo-
types yielded greater
use
of
gender
stereotypes
than
did
diagnos-
tic
case
information
(see also Ginossar
&
Trope, 1980, 1987;
Locksley,
Hepburn,
&
Ortiz,
1982).
Consistent
with
such
find-
ings,
this review
found
that sex-neutral
sex
role stimulus
mate-
rial
reflected
greater gender bias than either masculine
or
femi-
nine
sex
role stimulus material.
In
addition, this
review
gener-
ally
found
that there
was a
slight tendency
for
studies
in
which
subjects
were
provided
with
less information
to
yield
larger
effect
sizes.
However,
the
effect
sizes remained relatively small
even
when
sex-neutral stimulus material
was
used
and
when
only
minimal amounts
of
information
about
the
target
were
provided.
It
should
be
noted that this
review
does
not
represent
a
strong test
of the
Locksley
et al.
(1980)
findings,
despite
the
amount
of
information about
the
target person that subjects
were
given,
because
all
subjects
received
a
description
of
some
evidence
of
the
person's
abilities. Thus,
the
present research
still
leaves
open
the
possibility that
the
presence
of
individuating
information
accounts
for
people's disregard
of
their
stereo-
types.
In
summary, little evidence
was
found
for the
simple predic-
tion
that subjects
differentially
evaluate
men and
women
in the
context
of the
classic paradigm that Goldberg introduced
in
1968.
This
contradicts previous qualitative
reviews
of
the
litera-
ture that
have
in
part
perpetuated misconceptions about
the
data base
on
gender-biased evaluations. Gender-biased evalua-
tions
indeed
occur,
but as
research
in
social cognition and, more
specifically,
as
recent research
in
gender stereotypes suggest,
the
complexity
of the
conditions under
which
such evaluations
oc-
cur
and the flexibility of
social
perceivers'
thinking must
be
taken
into consideration.
3
The
frequency
of
occurrence
of
different
subtypes
of men and
women
may
perhaps explain
why
discrimination
is
more likely
to
occur
outside
an
experimental setting.
For
instance, within
a
given
situation,
a man is
more
likely
to
have
a
higher status position than
a
woman.
Thus,
comparisons between
men and
women
are
likely
to be
con-
founded
by
status
level
(cf.
Eagly,
1987;
Eagly
&
Steffen,
1984).
References
Arvey,
R. D.
(1979).
Unfair
discrimination
in the
employment inter-
view:
Legal
and
psychological aspects.
Psychological
Bulletin,
86,
736-765.
Bangert-Drowns,
R.
(1986).
Review
of
developments
in
meta-analytic
method.
Psychological
Bulletin,
99,
388-399.
Bar-Tal,
D.,
&
Saxe,
L.
(1976).
Physical
attractiveness
and its
relation-
ship
to
sex-role stereotyping.
Sex
Roles,
2,
123-148.
Basow,
S. A.
(1986).
Gender
stereotypes:
Traditions
and
alternatives.
Monterey,
CA:
Brooks/Cole.
Berscheid,
E.
(1985).
Interpersonal attraction.
In G.
Lindzey
and E. Ar-
onson
(Eds.),
Handbook
of
social
psychology
(Vol.
2,3rd
ed.,pp.
413-
484).
HiLlsdale,
NJ:
Erlbaum.
Bodenhausen,
G.
V.,
&
Lichtenstein,
M.
(1987). Social stereotypes
and
information-processing
strategies:
The
impact
of
task complexity.
Journal
of
Personality
and
Social
Psychology.
52,
871-880.
Bodenhausen,
G. V, &
Wyer,
R.
S.,
Jr.
(1985).
Effects
of
stereotypes
on
decision-making
and
information-processing strategies.
Journalof
Personality
and
Social
Psychology,
48,267-282.
Brown,
R.
(1987).
Social
psychology
(2nd
ed.).
New
York:
Free Press.
Cash,
T.
F,
&
Trimer,
C. A.
(1984).
Sexism
and
beautyism
in
women's
evaluations
of
peer performance.
Sex
Roles,
10,
87-98.
Cooper,
H.
M.,
&
Rosenthal,
R.
(1980). Statistical
versus
traditional
procedures
for
summarizing research
findings.
Psychological
Bulle-
tin,
87,442-449.
Deaux,
K.
(1985).
Sex and
gender.
Annual
Review
of
Psychology,
36,
49-81.
Deaux,
K.,
&
Lewis,
L. L.
(1983).
Assessment
of
gender
stereotypes:
Methodology
and
components.
Psychological
Documents,
13, 25.
(Ms.
No.
2583)
Deaux,
K.,
&
Lewis,
L. L.
(1984).
The
structure
of
gender
stereotypes:
Interrelationships among components
and
gender labels.
Journal
of
Personality
and
Social
Psychology,
46,
991-1004.
Deaux,
K.,
Winton,
W.,
Crowley,
M.,
&
Lewis,
L. L.
(1985).
Level
of
categorization
and
context
of
gender
stereotypes.
Social
Cognition,
3,
145-167.
Dion,
K.
L.,
Berscheid,
E.,
&
Walster,
E.
(1972). What
is
beautiful
is
good.
Journal
of
Personality
and
Social
Psychology,
24,285-290.
Eagly,
A. H.
(1978).
Sex
differences
in
influenceability.
Psychological
Bulletin,
85,
$6-\\6.
Eagly,
A. H.
(1987).
Sex
differences
in
social
behavior:
A
social-role
interpretation.
Hillsdale,
NJ:
Erlbaum.
Eagly,
A.
H.,
&
Carli,
L. L.
(1981).
Sex of researchers and
sex-typed
communications
as
determinants
of sex
differences
in
influenceabil-
ity:
A
meta-analysis
of
social
influence
studies.
Psychological
Bulle-
tin,
90,
1-20.
Eagly,
A.
H.,
&
Crowley,
M.
(1986).
Gender
and
helping
behavior
A
meta-analytic
review
of the
social psychological literature.
Psycholog-
ical
Bulletin,
100,
283-308.
Eagly,
A.
H.,
&
Steffen,
V. J.
(1984).
Gender stereotypes stem
from
the
distribution
of
women
and men
into
social roles.
Journal
of
Personal-
ity
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.
Psyctmlog-
KalBulletin,
100,
309-330.
Eysenck,
H. J.
(1978).
An
exercise
in
mega-silliness.
American
Psychol-
ogist,
33,
517.
Putoran,
G. C, &
Wyer,
R. S.
(1986).
The
effects
of
traits
and
gender
stereotypes
on
occupational suitability judgments
and the
recall
of
judgment-relevant
information.
Journal
of
Experimental
Social
Psy-
chology,
22.475-503.
Ginossar,
Z.,
&
Trope,
Y.
(1980).
The
effects
of
base
rates and
individu-
ating
information
on
judgments about another person.
Journal
of
Ex-
perimental
Social
Psychology,
16,228-242.
Ginossar,
Z.,
&
Trope,
Y.
(1987). Problem solving
in
judgment under
uncertainty.
Journal
of
Personality
and
Social
Psychology,
52,
464-
474.
Glass,
G. V.
(1976).
Primary, secondary,
and
meta-analysis
of research.
Educational
Researcher,
5,
3-8.
Glass,
G.
V.,
McGaw,
B.,
&
Smith,
M. L.
(1981).
Mela-analysis
in
social
research.
Beverly
Hills,
CA:
Sage.
Gleitman,
H.
(1981).
Psychology.
New
York:
Norton.
Goldberg,
P.
(1968).
Are
women
prejudiced against
women?
Transac-
tion,
5,
28-30.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
425
Gornick,
N., &
Moran,
B.
K.
(1971).
Woman
in
sexist
society:
Studies
of
power
andpowerlessness.
New
York:
Basic Books.
Green,
B. E, &
Hall,
J. A.
(1984).
Quantitative methods
for
literature
reviews.
Annual Review
of
Psychology,
35,
37-53.
Greenwald,
A. G.
(1975). Consequences
of
prejudices against
the
null
hypothesis.
Psychological
Bulletin,
82,1-20.
Hedges,
L.
V.,
&
Olkin,
I.
(1985).
Statistical
methods
for
meta-analysis.
Orlando,
FL:
Academic Press.
Hewlett,
S. A.
(1986).
A
lesser
life:
The
myth
of
women's
liberation
in
America.
New
York:
William Morrow.
Hunter,
J.
E.,
Schmidt,
F.
L.,
&
Jackson,
G. B.
(1982).
Meta-analysis:
Cumulating
research
findings
across
studies.
Beverly
Hills,
CA:
Sage.
Hyde,
J.
S.,
&
Linn,
M. C.
(1986).
The
psychology
of
gender:
Advances
through
meta-analysis.
Baltimore,
MD:
Johns Hopkins
University
Press.
Johnson,
R.
(1980).
Elementary
statistics.
North
Scituate,
MA:
Dux-
bury
Press.
Jones,
E. E.
(1985).
Major
developments
in
social
psychology
during
the
past
five
decades.
In G.
Lindzey
and E.
Aronson
(Eds.),
Handbook
of
social
psychology
(Vol.
2, 3rd
ed.,
pp.
47-107).
Hillsdale,
NJ:
Erl-
baum.
Lane,
D.
M.,
&
Dunlap,
W. P.
(1978).
Estimating
effect
size:
Bias
result-
ing
from
the
significance
criterion
in
editorial decisions.
British
Jour-
nal
of
'Mathematical
and
Statistical
Psychology,
31,
107-112.
Lips,
H.
M.,
&
Colwill,
N. L.
(1978).
The
psychology
of
sex
differences.
Englewood
Cliffs,
NJ:
Prentice-Hall.
Locksley,
A.,
Borgida,
E.,
Brekke,
N.,
&
Hepburn,
C.
(1980).
Sex
stereo-
types
and
social judgment.
Journal
ofPersonality
and
Social
Psychol-
ogy,
19,
821-831.
Locksley,
A.,
Hepburn,
C.,
&
Ortiz,
V.
(1982). Social stereotypes
and
judgments
of
individuals.
Journal
of
Experimental
Social
Psychology,
18,
23-42.
Lueptow,
L. B.
(1980).
Social
change
and
sex-role change
in
adolescent
orientations toward
life,
work,
and
achievement:
1964-1975.
Social
Psychology
Quarterly,
43,48-59.
Martin,
C. L.
(1986).
A
ratio measure
of sex
stereotyping.
Journal
of
Personality
and
Social
Psychology,
52,489-499.
Neuberg,
S.
L.,
&
Fiske,
S. T.
(1987).
Motivational
influences
on
impres-
sion formation: Outcome dependency,
accuracy-driven
attention,
and
individuating
processes.
Journal
ofPersonality
and
Social
Psy-
chology,
52,431^*50.
Nieva,
V.
F.,
&
Gutek,
B. A.
(1980).
Sex
effects
on
evaluation.
Academy
of
Management
Review,
5,
267-276.
Orwin,
R.
G.,
&
Cordray,
D. S.
(1985).
Effects
of
deficient
reporting
on
meta-analysis:
A
conceptual
framework
and
reanalysis.
Psychological
Bulletin,
97.
133-147.
Paludi,
M.
A.,
&
Bauer,
W. D.
(1983). Goldberg revisited: What's
in an
author's name.
SexRoles,
9(3),
387-396.
Pheterson,
G.
I.,
Kiesler,
S.
B.,
&
Goldberg,
P. A.
(1971).
Evaluation
of
the
performance
of
women
as
a
function
of
their
sex,
achievement
and
personal history.
Journal
ofPersonality
and
Social
Psychology,
19,
114-118.
Rasinski,
K.
A.,
Crocker,
J.,
&
Hastie,
R.
(1985).
Another
look
at sex
stereotypes
and
social
judgments:
An
analysis
of
the
social
perceiver's
use of
subjective
probabilities.
Journal
ofPersonality
and
Social
Psy-
chology,
49,
317-326.
Rosen,
B.,
&
Jerdee,
T. H.
(1974).
Sex
stereotyping
in the
executive
suite.
Harvard
Business
Review,
52(March-April),
45-58.
Rosen,
B.,
Jerdee,
T.
H.,
&
Prestwich,
T. L.
(1975).
Dual-career
marital
adjustment:
Potential
effects
of
discriminatory managerial attitudes.
Journal
of
Marriage
and
the
Family,
37,
565-573.
Rosenthal,
R.,
&
Rubin,
D. B.
(1986).
Meta-analytic
procedures
for
combining studies
with
multiple
effect
sizes.
Psychological
Bulletin,
99,
400-406.
Rothbart,
M.,
&
John,
O.
P.
(1985).
Social categorization
and
behav-
ioral episodes:
A
cognitive analysis
of
effects
of
intergroup
contact.
Journal
of
Social
Issues,
41,
81-104.
Ruble,
D.
N.,
&
Ruble,
T. L.
(1982).
Sex
stereotypes.
In A. G.
Miller
(Ed.),
In the eye
of
the
beholder:
Contemporary
issues
in
stereotyping
(pp.
188-251).
New
York:
Praeger.
Slavin,
R. E.
(1986).
Best evidence synthesis:
An
alternative
to
meta-
analysis
and
traditional
reviews.
Educational
Researcher,
15,
5-11.
Strube,
M.
J.,
&
Miller,
R. H.
(1986). Comparison
of
power
rates
for
combined probability procedures:
A
simulation
study.
Psychological
Bulletin,
99,
407-415.
Taylor,
S. E.
(1980).
A
categorization approach
to
stereotyping.
In
D. L.
Hamilton
(Ed.),
Cognitive
processes
in
stereotyping
and
inter-
group
behavior(pp.
83-114).
Hillsdale,
NJ:
Erlbaum.
Thomas,
J.
R.,
&
French,
K. E.
(1985).
Gender
differences
across
age
in
motor performance:
A
meta-analysis.
Psychological
Bulletin,
98,
260-282.
Tosi,
H.
L.,
&
Einbender,
S. W.
(1985).
The
effects
of the
type
and
amount
of
information
in sex
discrimination
research:
A
meta-analy-
sis.
Academy
of
Management
Journal,
28,
712-723.
Unger,
R. K.
(1976).
Male
is
greater than
female:
The
socialization
of
status inequality.
Counseling
Psychologist,
6,2-9.
Wallston,
B.,
&
O'Leary,
V.
(1981).
Sex
makes
a
difference:
Differential
perceptions
of
women
and
men.
Review
ofPersonality
and
Social
Psychology,
2,9-41.
Whitley,
B.
E.,
&
Frieze,
I. H.
(1986).
Measuring causal attributions
for
success
and
failure:
A
meta-analysis
of the
effects
of
question-wording
style.
Basic
and
Applied
Social
Psychology,
7,
35-51.
Zukier,
H.
(1985).
The
paradigmatic
and
narrative modes
in
goal-
guided
inference.
In R. M.
Sorrentino
& E. T.
Higgins
(Eds.),
Hand-
book
of
motivation
and
cognition
(pp. 465-502).
Hillsdale,
NJ:
Erl-
baum.
Zukier,
H.,
&
Pepitone,
A.
(1984).
Social roles
and
strategies
in
predic-
tion:
Some determinants
of
the use of
base-rate
information.
Journal
of
Personality
and
Social
Psychology,
47,349-360.
(Appendixes
follow
on
next
page)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

426
SWIM,
BORGIDA,
MARUYAMA,
AND
MYERS
Appendix
A
Articles Reviewed
Abramson,
P.,
Goldberg,
P.
A.,
Greenberg,
J. H., &
Abramson,
L. M.
(1977).
The
talking platypus phenomenon:
Competency
ratings
as a
function
of sex and
professional status.
Psychology
of
Women
Quar-
terly,
2,
114-124.
Anderson,
R., &
Nida,
S.
(1978).
Effect
of
physical attractiveness
on
opposite-
and
same-sex
evaluations.
Journal
of
Personality
46,401-
413.
Bartol,
K.
M.,
&
Butterfield,
A. D.
(1976).
Sex
effects
in
evaluating
leaders.
Journal
of
Applied
Psychology,
61,446-454.
Baruch,
G. K.
(1972).
Maternal
influences
upon college women's atti-
tudes toward
women
and
work.
Developmental
Psychology,
6,
32-37.
Basow,
S.
A.,
&
Distenfeld,
M. S.
(1985). Teacher expressiveness: More
important
for
male teachers than
female
teachers?
Journal
of
Educa-
tional
Psychology,
77,
45-52.
Basow,
S.
A.,
&
Howe,
K.
G.
(1979).
Sex
bias
and
career evaluations
by
college
women.
Perceptual
and
Motor
Skills,
49,
705-706.
Beattie,
M. Y., &
Diehl,
L. A.
(1979).
Effects
of
social
conditions
on the
expression
of
sex-role stereotypes.
Psychology
of
Women
Quarterly,
4(2),
241-255.
Bigoness,
W.
J.
(1976).
Effect
of
applicant's
sex,
race,
and
performance
on
employers' performance ratings: Some additional
findings.
Jour-
nal
of
Applied
Psychology,
61,
80-84.
Brawley,
L.
R.,
Landers,
D.
M.,
Miller,
L.,
&
Reams,
K. F.
(1979).
Sex
bias
in
evaluating motor performance.
Journal
of
Sport
Psychology,
1,
15-24.
Brief,
A. P., &
Wallace,
M.
J.,
Jr.
(1976).
The
impact
of
employee
sex
and
performance
on the
allocation
of
organizational rewards.
Journal
of
Psychology,
92,25-34.
Brown,
V.,
&
Geis,
F. L.
(1984). Turning lead into gold: Evaluations
of
men
and
women leaders
and the
alchemy
of
social
consensus.
Journal
of
Personality
and
Social
Psychology,
46,
811-824.
Cann,
A.,
Siegfried,
W.
D.,
&
Pearce,
L.
(1981).
Forced attention
to
specific
applicant qualifications: Impact
on
physical attractiveness
and sex of
applicant
biases.
Personnel
Psychology,
34,
65-75.
Cash,
T.
F.,
Gillen,
B., &
Burns,
S.
(1977). Sexism
and
"beautyism"
in
personnel
consultant decision making.
Journal
of
Applied
Psychology.
62,301-310.
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.
Cash,
T.
F.,
&
Trimer,
C. A.
(1984). Sexism
and
beautyism
in
women's
evaluations
of
peer performance.
SexRoles,
10(1/2),
87-98.
Chobot,
D.
S.,
Goldberg,
P.
A.,
Abramson,
L.
M.,
&
Abramson,
P. R.
(1974).
Prejudice against women:
A
replication
and
extension. Psy-
chological
Reports,
35,478.
Clifford,
M.
M.,
&
Walster,
E.
(1972).
The
effect
of sex on
college
admis-
sion,
work
evaluation,
and job
interviews.
Journal
of
Experimental
Education,
41(2),
1-5.
Cline,
M.
E.,
Holmes,
D.
S.,
&
Werner,
J. C.
(1977). Evaluations
of the
work
of men and
women
as a
function
of
the
sex of the
judge
and
type
of
work.
Journal
of
Applied
Social
Psychology,
7(1),
89-93.
Cohen,
S.
L.,
&
Bunker,
K. A.
(1975). Subtle
effects
of sex
stereotypes
on
recruiters' hiring decisions.
Journal
of
Applied
Psychology,
60(5),
566-572.
Cohen,
S.
L.,
Bunker,
K.
A.,
Burton,
A.
L.,
&
McManus,
P. D.
(1978).
Reactions
of
male subordinates
to the
sex-role
congruency
of
imme-
diate
supervision. SexRoles,
4(2),
297-311.
Cohen,
S.
L.,
&
Leavengood,
S.
(1978).
The
utility
of the
WAMS:
Shouldn't
it
relate
to
discriminatory behavior.
Academy
of
Manage-
ment
Journal,
21,
742-748.
Deaux,
K.,
&
Taynor,
J.
(1973). Evaluations
of
male
and
female
ability:
Bias
works
two
ways.
Psychological
Reports,
32,
261-262.
Dell,
D.
M.,
&
Schmidt,
L. D.
(1976).
Behavioral cues
to
counselor
expertness.
Journal
of
Counseling
Psychology,
23,
197-201.
Dipboye,
R.
L.,
Arvey,
R.
D.,
&
Terpstra,
D. E.
(1977).
Sex
and
physical
attractiveness
of
raters
and
applicants
as
determinants
of
resume
eval-
uations.
Journal
of
Applied
Psychology,
62,288-294.
Dipboye,
R.
L.,
Fromkin,
H.
L.,
&
Wiback,
K.
(1975). Relative impor-
tance
of
applicant attractiveness,
and
scholastic standing
in
evalua-
tion
of
job
applicant resumes.
Journal
of
Applied
Psychology,
60, 39-
43.
Dipboye,
R.
L.,
&
Wiley,
J. W.
(1977).
Reactions
of
college recruiters
to
interviewee
sex and
self-presentation
style.
Journal
of
Vocational
Behavior,
10,1-12.
Dipboye,
R.
L.,
&
Wiley,
J. W.
(1978).
Reactions
of
male
raters to
inter-
viewee
self-presentation style
and
sex:
Extensions
of
previous
re-
search.
Journal
of
Vocational
Behavior,
13.
192-203.
Ekstrand,
L.
E.,
&
Eckert,
W. A.
(1981).
The
impact
of
candidates
sex
on
voter choice.
Western
Political
Quarterly,
34,
78-87.
Etaugh,
C.,
&
Kasley,
H. C.
(1981).
Evaluating competence:
Effects
of
sex,
marital
status,
and
parental status.
Psychology
of
Women
Quar-
terly,
6(2),
196-203.
Etaugh,
C.,
&
Riley,
S.
(1983). Evaluating competence
of
women
and
men:
Effects
of
marital
and
parental status
and
occupational
sex-typ-
ing.
SexRoles,
9(9),
943-952.
Etaugh,
C.,
&
Rose,
S.
(1975).
Adolescents'
sex
bias
in the
evaluation
of
performance.
Developmental
Psychology,
11,
663-664.
Etaugh,
C.,
&
Sanders,
S.
(1974). Evaluation
of
performance
as a
func-
tion
of
status
and
sex
variables.
The
Journal
of
Social
Psychology,
94,
237-241.
Etaugh,
C, &
Stern,
J.
(1984).
Person perception:
Effects
of
sex, marital
status,
and
sex-typed
occupation.
SexRoles,
11,
413-424.
Fidell,
L. S.
(1970). Empirical
verification
of sex
discrimination
in
hir-
ing
practices
in
psychology.
American
Psychologist,
25,
1094-1098.
Francesco,
A.
M.,
&
Hakel,
M. D.
(1981).
Gender
and sex
as
determi-
nants
of
hireability
of
applicants
for
gender-typed
jobs.
Psychology
of
Women
Quarterly.
3,
747-757.
Frank,
F.
D.,
&
Drucker,
J.
(1977).
The
influence
of
evaluatee's
sex on
evaluations
of a
response
on a
managerial selection instrument.
Sex
Roles,
3(1),
59-64.
Gerdes,
E. P., &
Kelman,
J. H.
(1981).
Sex
discrimination:
Effects
of
sex-role
incongruence,
evaluator
sex,
and
stereotypes.
Basic
and
Ap-
plied
Social
Psychology,
2(3),
219-226.
Gilbert,
L.
A.,
Lee,
R. N., &
Chiddix,
S.
(1981).
Influence
of
presenter's
gender
on
students'
evaluations
of
presenters discussing
sex
fairness
in
counseling:
An
analogue
study.
Journal
of
Counseling
Psychology,
28,
258-264.
Goldberg,
P.
(1968).
Are
women prejudiced against
women?
Transac-
tion,
5,
28-30.
Gordan,
T.,
&
Draper
T. W.
(1982).
Sex
bias against male
workers
in
day
care.
Child
Care
Quarterly,
11(3),
215-217.
Gross,
M.
M.,
&
Geffner,
R. A.
(1980).
Are the times
changing?
An
analysis
of
sex-role prejudice.
Sex
Roles,
6(5),
713-722.
Gruber,
K.
J.,
&
Gaebelein,
J.
(1979).
Sex
differences
in
listening com-
prehension.
SexRoles,
.5(3),
299-310.
Gutek,
B.
A.,
&
Stevens,
D. A.
(1979).
Effects
of sex of
subjects,
sex of
stimulus
cue,
and
androgyny
level
on
evaluations
in
work
situations
which
evoke
sex
role stereotypes.
Journal
of
Vocational
Behavior,
14,
23-32.
Haccoun,
D.
M.,
Haccoun,
R.
R.,
&
Sallay,
G.
(1978).
Sex
differences
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER
STEREOTYPES
427
in the
appropriateness
of
supervisory
styles:
A
nonmanagement
view.
Journal
of
Applied
Psychology,
63,
124-127.
Haefner,
J. E.
(1977).
Race, age, sex,
and
competence
as
factors
in em-
ployer
selection
of the
disadvantaged.
Journal
of
Applied
Psychology,
62,
199-202.
Hagen,
R.
L.,
&
Kahn,
A.
(1975).
Discrimination
against competent
women.
Journal
of
Applied
Social
Psychology,
5(4), 362-376.
Hall,
F. S., &
Hall,
D. T.
(1976).
Effects
of
job
incumbents' race
and sex
on
evaluations
of
managerial performance.
Academy
of
Management
Journal,
19(3),
476^*81.
Hamner,
W.
C,
Kim,
J.
S.,
Baird,
L.,
&
Bigoness,
W. J.
(1974).
Race
and
sex as
determinants
of
ratings
by
potential employers
in a
simu-
lated
work-sampling task.
Journal
of
Applied
Psychology,
59,
705-
711.
Harris,
M. B.
(197
5). Sex
role stereotypes
and
teacher evaluations.
Jour-
nal
of
Educational
Psychology,
67,751-756.
Harris,
M. B.
(1976).
The
effects
of
sex, sex-stereotyped descriptions,
and
institution
on
evaluations
of
teachers.
SexRoles,
2,
15-21.
Heilman,
M.
E.,
&
Guzzo,
R. O.
(1978).
The
perceived cause
of
work
success
as a
mediator
of
sex
discrimination
in
organizations.
Organi-
zational
Behavior
and
Human
Performance,
21,
346-357.
Heilman,
M.
E.,
&
Saruwatari,
L. R.
(1979).
When beauty
is
beastly:
The
effects
of
appearance
and sex on
evaluations
of
job
applicants
for
managerial
and
nonmanagerial
jobs.
Organizational
Behavior
and
Human
Performance,
23,
360-372.
Heneman.
H. E.
(1977).
Impact
of
test information
and
applicant
sex
on
applicant evaluations
in a
selection simulation.
Journal
of
Applied
Psychology,
62,
524-526.
Heppner,
P. P.,
&Pew,
S.
(1977).
Effects
of
diplomas, awards,
and
coun-
selor
sex on
perceived
expertness.
Journal
of
Counseling
Psychology,
24,
147-149.
Hodgins,
D.
C.,
&
Kalin,
R.
(1985). Reducing
sex
bias
in
judgments
of
occupational suitability
by the
provision
of
sex-typed personality
information.
Canadian
Journal
of
Behavioral
Science,
17,346-358.
Isaacs,
M. B.
(1981).
Sex
role stereotyping
and the
evaluation
of the
performance
of
women: Changing trends.
Psychology
of
Women
Quarterly,
6(2),
187-195.
Kaschak,
E.
(1978).
Sex
bias
in
student evaluations
of
college
professors.
Psychology
of
Women
Quarterly,
2,235-243.
Kaschak,
E.
(1981).
Another look
at sex
bias
in
students' evaluations
of
professors:
Do
winners
get the
recognition that they
have
been
given?
Psychology
of
Women
Quarterly,
5,
767-772.
Kryger,
B.
R.,
&
Shikiar,
R.
(1978).
Sexual
discrimination
in the use of
letters
of
recommendation:
A
case
of
reverse
discrimination.
Journal
of
Applied
Psychology,
63,
309-314.
Labovitz,
S.
(1974). Some evidence
of
Canadian ethnic, racial,
and
sex-
ual
antagonism. Review
of
Canadian
Sociology
and
Anthropology,
11(3),
247-254.
Lao,
R. C,
Upchurch,
W.
H.,
Corwin,
B.
J.,
&
Grossnickle,
W. F.
(1975).
Biased attitudes toward
females
as
indicated
by
ratings
of in-
telligence
and
likeability.
Psychological
Reports,
3
7,
1315-1320.
Larwood,
L.,
Rand,
P., &
Hovanessian,
A.
D.
(1979).
Sex
differences
in
response
to
simulated employee discipline cases.
Personnel
Psychol-
ogy,
32,
539-550.
Lee,
D.
M.,
&
Alvares,
K. M.
(1977).
Effects
of sex on
descriptions
and
evaluations
of
supervisory behavior
in a
simulated industrial setting.
Journal
of
Applied
Psychology,
62,
405-410.
Lee,
D.
Y,
Hallberg,
E. T,
Jones,
L.,
&
Haase,
R. F.
(1980).
Effects
of
counselor
gender
on
perceived credibility.
Journal
of
Counseling
Psychology,
27,1'1-75.
Lenny,
E.,
Mitchell,
L.,
&
Browing,
C.
(1983).
The
effect
of
clear evalua-
tions criteria
on sex
bias
in
judgments
of
performance.
Psychology
of
Women
Quarterly,
7,
313-328.
Levenson,
H.,
Burford,
B.,
Bonno,
B.,
&
Davis,
L.
(1975).
Are
women
still
prejudiced against women?
A
replication
and
extension
of
Gold-
berg's study.
The
Journal
of
Psychology,
89,67-71.
Lewin,
A.
Y.,
&
Duchan,
L.
(1971).
Women
in
academia.
Science,
173,
892-895.
Linsenmeier,
J. A. W, &
Wortman,
C. B.
(1979).
Attitudes toward
work-
ers
and
toward their
work:
More evidence that
sex
makes
a
difference.
Journal
of
Applied
Social
Psychology,
4,
326-334.
Linville,
P.
W.,
&
Jones,
E. E.
(1980). Polarized appraisals
of
out-group
members.
Journal
of
Personality
and
Social
Psychology,
38,689-703.
London,
M.,
&
Stumpf,
S. A.
(1983).
Effects
of
candidate characteristics
on
management promotion decisions:
An
experimental study. Per-
sonnel
Psychology,
36,
241-259.
Mai-Dalton,
R.
R.,
Feldman-Summers,
S.,
&
Mitchell,
T. R.
(1979).
Effect
of
employee gender
and
behavioral style
on the
evaluations
of
male
and
female
banking executives.
Journal
of
Applied
Psychology,
64,221-226.
Miller,
G.
R.,
&
McReyndds,
M.
(1973). Male chauvinism
and
source
competence:
A
research
note.
Speech
Monographs,
40,
154-155.
Mischel,
H. N.
(1974).
Sex
bias
in the
evaluation
of
professional
achievements.
Journal
of
Educational
Psychology,
66,
157-166.
Morrow,
W.
R.,
Lowenberg,
G.,
Larson,
S.,
Redfeam,
M.,
&
Schoone,
J.
(1983).
Evaluations
of
business memos:
Effects
of
writer
sex and
organizational position, memo quality,
and
rater sex.
Personnel
Psy-
chology,
36,
73-85.
Muchinsky,
P.
M.,
&
Harris,
S. L.
(1977).
The
effect
of
applicant
sex
and
scholastic standing
on the
evaluation
of
job
applicant resumes
in
sex-typed
occupations.
Journal
of
Vocational
Psychology,
11,
95-108.
Noel,
R.
C.,
&
Allen,
M. J.
(1976).
Sex and
ethnic bias
in the
evaluation
of
student editorials.
Journal
of
Psychology,
94,53-58.
Norton,
S.
D.,
Gustafson,
D. P., &
Foster,
C. E.
(1977). Assessment
for
management
potential: Scale design
and
development,
training
effects
and
rater/ratee
sex
effects.
Academy
of
Management
Journal,
20(1),
117-131.
Paludi,
M.
A.,
&
Bauer,
W. D.
(1983). Goldberg revisited: What's
in an
author's name. SexRoles, 9(3), 387-396.
Paludi,
M.
A.,
&
Strayer,
L. A.
(1985).
What's
in an
author's name?
Differential
evaluations
of
performance
as a
function
of
author's
name.
Sex
Roles,
12(3/4),
353-361.
Panek,
P.
E.,
Deitchman,
R.,
Burkholder,
J.
H.,
Speroff,
T, &
Haude,
R.
H.
(1976).
Evaluation
of
feminine
professional competence
as a
function
of
level
of
accomplishment.
Psychological
Reports,
38,875-
880.
Peck,
T.
(1978). When
women
evaluate women, nothing succeeds
like
success:
The
differential
effects
of
status upon evaluations
of
male
and
female
professional ability.
Sex
Roles,
4(2),
205-213.
Pheterson,
G.
I.,
Kiesler,
S.
B.,
&
Goldberg,
P. A.
(1971).
Evaluation
of
the
performance
of
women
as a
function
of
their
sex,
achievement,
and
personal history.
Journal
of
Personality
and
Social
Psychology,
19,
114-118.
Piacente,
B. S.
(1974).
Women
as
experimenters.
American
Psycholo-
gist,
29,
527-529.
AI
Piacente,
B.
S.,
Penner,
L.
A.,
Hawkins,
H.
L.,
&
Cohen,
S. L.
(1974).
Evaluation
of the
performance
of
experimenters
as
a
function
of
their
sex and
competence.
Journal
of
Applied
Social
Psychology,
4(4),
321-
329.
Renwick,
P.
A.,
&
Tosi,
H.
(1978).
The
effects
of
sex,
marital status,
and
educational background
on
selection decisions.
Academy
of
Manage-
ment
Journal,
21(1),
93-103.
Rhue,
J.
W.,
Lynn,
S.
R.,
&
Garske,
J. A.
(1984).
The
effects
of
compe-
A1
Although Piacente (1974)
and
Piacente,
Penner, Hawkins,
and
Co-
hen
(1974)
reported
the
same study, Piacente
et
al.
(1974)
is
more com-
plete; hence this reference
was
used
as the
primary source
for
the
data.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

428
SWIM,
BORG1DA,
MARUYAMA,
AND
MYERS
tent behavior
on
interpersonal
attraction
and
task leadership.
Sex
Roles,
10,
925-937.
Rose,
G. L., &
Andiappan,
P.
(1978).
Sex
effects
on
managerial hiring
decisions.
Academy
of
Management
Journal,
21(1),
104-112.
Rose,
G.
L.,
&
Stone,
T. H.
(1978).
Why
good
job
performance
may
(not)
be
rewarded:
Sex
factors
and
career development.
Journal
of
Vocational
Behavior,
12,197-207.
Rosen,
B., &
Jerdee,
T. H.
(1973).
The
influence
of
sex-role
stereotypes
on
evaluations
of
male
and
female
supervisory behavior.
Journal
of
Applied
Psychology,
57,
44-48.
Rosen,
B.,
&
Jerdee,
T. H.
(1974a).
Effects
of
applicant's
sex and
diffi-
culty
of job on
evaluations
of
candidates
for
managerial
positions.
Journal
of
Applied
Psychology,
59,511-512.
Rosen,
B.,
&
Jerdee,
T. H.
(I974b).
Influence
of sex
role stereotypes
on
personnel decisions.
Journal
of
Applied
Psychology,
59,
9-14.
Rosen,
B.,
&
Jerdee,
T. H.
(1974c).
Sex
stereotyping
in the
executive
suite.
Harvard
Business
Review,
52,45-58.
A2
Rosen,
B.,
Jerdee,
T.
H.,
&
Prestwich,
T. L.
(1975).
Dual-career
marital
adjustment:
Potential
effects
of
discriminatory managerial attitudes.
Journal
of
Marriage
and the
Family,
37,
565-572.
Sanders,
G. S., &
Schmidt,
T.
(1980).
Behavioral discrimination against
women.
Personality
and
Social
Psychology
Bulletin,
6(3),
484-488.
Schmitt,
N., &
Lappin,
M.
(1980).
Race
and sex as
determinants
of
the
mean
and
variance
of
performance
ratings.
Journal
of
Applied
Psychology,
65,428-435.
Sharp,
C,
&
Post,
R.
(1980). Evaluation
of
male
and
female
applicants
for
sex-congruent
and
sex-incongruent
jobs.
Sex
Roles,
6,391
-401.
Sigelman,
L.,
&
Sigelman,
C.
K.
(1982). Sexism, racism,
and
ageism
in
voting
behavior
An
experimental analysis.
Social
Psychology
Quar-
terly.
45(4),
263-269.
Soto,
D.
H.,
&
Cole,
C.
(1975). Prejudice against
women:
A new
per-
spective.
SexRoles,
1(4),
385-393.
Starer,
R.,
&
Denmark,
F.
(1974). Discrimination against aspiring
women.
International
Journal
of
Group
Tensions,
4(1),
65-70.
Stumpf,
S.
A.
,&
London,
M.(1981).
Capturing
rater
policies
in
evaluat-
ing
candidates
for
promotion.
Academy
of
Management
Journal,
24,
752-766.
Tanner,
L. R.
(1977).
Sex
bias
in
children's response
to
literature. Lan-
guage
Arts,
54(1),
43-50.
Taylor,
S. E., &
Falcone,
H. T.
(1982). Cognitive bases
of
stereotyping:
The
relationship between categorization
and
prejudice.
Personality
and
Social
Psychology
Bulletin,
S,
426-432.
Terborg,
J.
R.,
&
llgen,
D. R.
(1975).
A
theoretical
approach
to sex
discrimination
in
traditionally masculine occupations.
Organiza-
tional
Behavior
and
Human
Performance,
13,
352-376.
Toder,
N. L.
(1980).
The
effect
of the
sexual composition
of a
group
on
discrimination against women
and
sex-role attitudes.
Psychology
of
Women
Quarterly,
5(2),
92-310.
Valasek,
D.,
Avolio,
B.
J.,
&
Forbringer,
L. R.
(1979).
Effects
of
sex-
stereotyping
in
evaluating males
in a
traditional
female
role.
Psycho-
logical
Reports,
44,
1196-1198.
Walsh,
R.
P.,
&
Connor,
C. L.
(1979).
Old men and
young
women:
How
objectively
are
their skills assessed?
Journal
ofGerontology,
34,
561-
568.
Wiley,
M.
G.,
&
Eskilson,
A.
(1982).
The
interaction
of sex and
power
base
on
perceptions
of
managerial
effectiveness.
Academy
of
Manage-
ment
Journal,
25,
671-677.
Wiley,
M.
G.,
&
Eskilson,
A.
(1985).
Speech style,
gender
stereotypes,
and
corporate success: What
if
women
talk more
like
men?
Sex
Roles,
12(9/10),
993-1007.
"
2
Some
of the findings
from
Rosen
and
Jerdee
(1974b)
are
also
re-
ported
in
Rosen, Jerdee,
and
Prestwich (1975). Hence,
when
there
were
overlaps,
the
data
were
taken
from
Rosen
et
al.
(1975).
It
should
be
noted that studies
were
included
in the
present
meta-analysis
only
if
they
were
close approximations
to the
original Goldberg paradigm.
Appendix
B
General Assumptions
1.
For all
results,
negative values represent
a
more
favorable
rating
given
to
men,
and
positive values represent
a
more
favorable
rating
given
to
women.
If the
study
reported that
a
main
effect
or
simple
effect
was
significant,
but the
authors
did not
report
the
direction
of the findings
or report the
means,
then
the findings
were
recorded as
nonsignificant.
For
instance, some studies reported
the
means
for
each cell without
re-
porting
the
sample
size
per
cell. Without
the
sample
sizes
the
main
effect
means could
not be
calculated
nor
could
the
simple
effects
be
calculated
for
one
variable summed
across
a
second variable. Thus,
it
was
impossi-
ble to
tell
if the finding was
more
favorable
for men or for
women.
2. If the
dependent variable
did not
clearly
represent
a
rating
of
fa-
vorability.
it was not
included
in the
analyses.
For
instance, ratings
of
masculinity
or
femininity
were
not
included.
3. If a
study
only
reported
effects
for
a
male
(female)
target person
in
one
condition (e.g.,
high
competence) versus
a
male
(female)
target per-
son
in a
second condition (e.g.,
low
competence),
it was
assumed that
the
other comparisons between male
and
female
target persons
were
not
significant
(e.g., that
between
a
highly
competent
man and a
highly
competent woman
and
that between
a
less competent
man and a
less
competent woman
were
assumed
to be
nonsignificant.)
4.
If
both
multivariates
and
univariates
were
reported,
the
univari-
ates
were
used
in
calculations
to
make
the
results across
the
studies
comparable. Only
four
studies reported
multivariate
lvalues.
5.
If
nonsignificant
interactions
were
reported but
follow-up
tests
of
the
simple
effects
resulted
in
significant
findings,
nonsignificant
results
were
recorded
for the
simple
effects.
6.
Results
from
nonparametric
tests
were
assigned nominal levels
of
significance.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER STEREOTYPES
Appendix
C
Calculations
429
Voting
scores
and
effect
sizes
were
assigned
to the findings in
each
study.
Additionally,
z
scores
were
calculated
and
used
in the
calculation
of
effect
sizes.
The
voting score
is a
tally
of the
number
of
significant
and
nonsignificant results.
The
effect
size measures
the
magnitude
of an
effect
independent
of the
significance
level
and the
sample size.
Voting
score.
If a finding was
significant
and
favored
men,
it
was
as-
signed
a
value
of
-
1
.
If
it
favored
women,
it was
assigned
a
value
of
+1. If the finding was
nonsignificant
or
if no
result
was
reported,
it was
assigned
a
value
of
0.
When
study
is the
unit
of
analysis,
the
voting score
represents
the
mean
of
the
percentage
of findings
within each study that
found
significantly
more
favorable
ratings
of
female
target persons
and
male
target persons
and no
significant
differences.
Z
scores.
If
only
probability levels were reported,
z
scores
for
these
levels
were
assigned.
Z
scores
were
assigned
a
value
of
1.96
if the
study
only
reported that
it was
significant.
If
the
study
reported that
the
level
of
significance
was set at
certain levels
(e.g.,
p
£
.05 or p
£
.
10), then
the z
scores
for
significant
findings
were
reported at
comparable values
(e.g.,
1.96
or
1.645,
respectively).
If the
study reported that
the
effect
was
nonsignificant
or did not report any
effect,
it was
assigned
a z
score
of
zero.
If
chi-squares
were
used, nominal levels
of
significance
were
used
un-
less proportions
were
reported.
In the
latter case
the
following
formulas
were
used (Johnson, 1980):
(a) For
tests
in
which subjects must make
a
choice between
a man and a
woman,
where
p\ =
proportion
of men or
proportion
of
women
chosen,
ni
=
number
of men or
number
of
women
chosen,
and
/i*
=
q*
= .5.
(b)
For
tests
in
which
the
subject chooses
a man or
chooses
a
woman
but
does
not
make
a
choice between
a man and a
woman,
z
=
(P,
-
P3)/[<P*«*(l/n.
+
l/nz)]"
2
,
where
p,
—
proportion
of men
chosen,
n,
=
number
of men
chosen,
P!
=
proportion
of
women
chosen,
n,
=
number
of
women chosen,
P*
=
(Pi
+
PJ)/(«.
+
«z),
and q* = 1 - p*.
Effect
size
calculations.
When means
and
standard deviations
were
reported, these
were
used
to
calculate main
and
simple
effect
sizes
(Hedges
&
Olkin,
1985; Hunter,
Schmidt,
&
Jackson, 1982).
Effect
size
=
(AT,
-
X
2
)/S
f
,
where
A'
i
=
mean
for rating of
female
target
person,
A'
2
=
mean
for
rating
of
male target person,
and
5
P
-
pooled
standard deviation.
The
standard deviations
were
pooled
in the
same manner
for
between-
groups
and
within-group
comparisons.
This
causes
the
effect
size calcu-
lated
in
this
manner
for
within-group comparisons
to be
larger because
variances tend
to be
smaller
for
this
design.
When
only
z
scores,
F
tests,
or t
tests
were
available,
these were used
to
calculate
the
effects
(Hunter
et
al.
1982). When
gender
of the
target
person
was
analyzed
as
a
between-subjects
variable,
Effect
size
=
A/«(l/n,
+
\/n
2
)
>r2
,
where
M
= z
score,
t
test,
or the
square
root
of
the
f
test;
n,
=
sample
size
for
those rating male target persons;
and
n
2
=
sample size
for
those
rating
female
target persons.
If
not
mentioned,
it was
assumed that there
were
an
equal number
of
subjects
per
condition. When gender
of the
target person
was
analyzed
as a
within-subject variable,
Effect
size
=
M«(l/n)"
2
,
where
M
=
z
score,
(test,
or the
square root
of the F
test;
and n =
number
of
subjects rating each pair
of
male
and
female
target persons.
If
both
means
and
test
statistics
were
reported,
effect
sizes
were
calcu-
lated
from
the
means
and
standard deviations.
If
exact values
of the z
score,
t
tests,
or F
tests
were
not reported, the
assigned
z
score
was
used
in the
calculations.
If F
tests
were
reported as
being
less
than
1,
they
were
assumed
to
equal
0.
Test
statistic values reported
were
assumed
to
be
accurate, although
in a few
cases this
did not
seem
likely.
For
in-
stance,
a
few
studies used
between-groups
tests
when
they should
have
used
within-group
tests.
Using
between-groups
tests
rather than within-
group
tests
causes
the
error terms
to be
smaller
and the
resulting
test
statistics
to be
larger.
This overestimate also causes
the
effect
sizes
to be
overestimated.
If
main
effects
were
reported as
significant
and the
interactions
as
nonsignificant,
then
the
effect
size assigned
to the
main
effect
was
also
assigned
to the
simple
effects.
Weighting
of
effect
sizes
by
variances,
confidence intervals
of the
effect
sizes, tests
of the
beta weights
in
regressions,
and
homogeneity
tests
were
calculated
by
methods recommended
by
Hedges
and
Olkin
(1985).
Received
November
9,1987
Revision
received
September
12,1988
Accepted
September
13,1988
•
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.