Social Incentives for Gender Differences in the Propensity to Initiate Negotiations: Sometimes It Does Hurt to Ask

Abstract

Four experiments show that gender differences in the propensity to initiate negotiations may be explained by differential treatment of men and women when they attempt to negotiate. In Experiments 1 and 2, participants evaluated written accounts of candidates who did or did not initiate negotiations for higher compensation. Evaluators penalized female candidates more than male candidates for initiating negotiations. In Experiment 3, participants evaluated videotapes of candidates who accepted compensation offers or initiated negotiations. Male evaluators penalized female candidates more than male candidates for initiating negotiations; female evaluators penalized all candidates for initiating negotiations. Perceptions of niceness and demandingness explained resistance to female negotiators. In Experiment 4, participants adopted the candidate’s perspective and assessed whether to initiate negotiations in same scenario used in Experiment 3. With male evaluators, women were less inclined than men to negotiate, and nervousness explained this effect. There was no gender difference when evaluator was female.

Full text

Organizational Behavior and Human Decision Processes 103 (2007) 84–103
www.elsevier.com/locate/obhdp
0749-5978/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.obhdp.2006.09.001
Social incentives for gender diVerences in the propensity
to initiate negotiations: Sometimes it does hurt to ask 夽
Hannah Riley Bowles a,¤, Linda Babcock b, Lei Lai b
a John F. Kennedy School of Government, Harvard University, 79 JFK Street, Cambridge, MA 02138, USA
b H. John Heinz III School of Public Policy and Management, Carnegie Mellon University, USA
Received 24 June 2005
Available online 7 November 2006
Abstract
Four experiments show that gender diVerences in the propensity to initiate negotiations may be explained by diVerential treat-
ment of men and women when they attempt to negotiate. In Experiments 1 and 2, participants evaluated written accounts of candi-
dates who did or did not initiate negotiations for higher compensation. Evaluators penalized female candidates more than male
candidates for initiating negotiations. In Experiment 3, participants evaluated videotapes of candidates who accepted compensation
oVers or initiated negotiations. Male evaluators penalized female candidates more than male candidates for initiating negotiations;
female evaluators penalized all candidates for initiating negotiations. Perceptions of niceness and demandingness explained resis-
tance to female negotiators. In Experiment 4, participants adopted the candidate’s perspective and assessed whether to initiate nego-
tiations in same scenario used in Experiment 3. With male evaluators, women were less inclined than men to negotiate, and
nervousness explained this eVect. There was no gender diVerence when evaluator was female.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Ask; Compensation; Gender diVerences; Gender role; Negotiation; Prescriptive sex stereotypes; Social outcomes; Status
Introduction
“There is no form of human excellence before which we
bow with profounder deference than that which appears
in a delicate womanƒand there is no deformity in
human character from which we turn with deeper loath-
ing than from a woman forgetful of her nature, and
clamorous for the vocation and rights of men.” Bledsoe
(1856, p. 224)1
Research on corporate managers suggests that women
are less likely than men to use negotiation in upward inXu-
ence attempts (Lauterbach & Weiner, 1996). Other studies
of broader populations indicate that women are less likely
than men, in general, to initiate negotiations (Babcock,
Gelfand, Small, & Stayn, 2006; Babcock & Laschever,
2003). Women report greater anxiety than men about
negotiating and are less likely than men to perceive situa-
tions as negotiable (Babcock et al., 2006).
Conventional wisdom (e.g., “it pays to ask” and “the
squeaky wheel gets the grease”) suggests that, if women
want the same resources and opportunities as men, then
夽We gratefully acknowledge support from the National Science
Foundation SES-0213474 and the Center for Public Leadership at the
Kennedy School of Government, Harvard University. The authors
thank the following people for their helpful comments on this manu-
script: Elaine Backman, Diane Burton, Robin Ely, Frank Flynn, Adam
Galinsky, Michele Gelfand, Fiona Greig, Laura Kray, Jennifer Lerner,
Kathleen McGinn, Denise Rousseau, Maureen Scully, and William
Simpson.
*Corresponding author. Fax: +1 617 496 3337.
E-mail address: hannah_bowles@harvard.edu (H.R. Bowles).
1As quoted by Jackman (1994, p. 79), who obtained quote from
Myrdal (1994, p. 1074).

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 85
they should learn to seek out, rather than shy away from,
opportunities to negotiate. For instance, one study of the
job negotiations of graduating professional school stu-
dents found that only 7% of female students attempted
to negotiate their initial compensation oVers as com-
pared to 57% of men. Those who negotiated gained on
average 7.4% over their initial oVers (Babcock & Lasch-
ever, 2003). Even small diVerences in starting salaries can
lead to substantial compensation gaps over time
(Bowles, Babcock, & McGinn, 2005; Gerhart & Rynes,
1991). Women’s reluctance as compared to men to initi-
ate negotiations may be an important and under-
explored explanation for the asymmetric distribution of
resources, such as compensation, within organizations.
So, why would women let such opportunities pass?
Maybe women need more training and practice in nego-
tiation to help them get over their nervous feelings and
to learn how to act more like the men when opportuni-
ties to negotiate arise. But, what if women’s relative hesi-
tation about initiating negotiations has less to do with
their negotiating ability than with the way they are
treated when they attempt to negotiate? “Fix the
women” solutions to gender issues often fail to take into
consideration the gendered social context out of which
gender diVerences in behavior emerge (Deaux & Major,
1987; Ely & Meyerson, 2000; Wade, 2001; Watson,
1994b).
Society rewards and reinforces diVerent types of
behavior for men and women (Eagly, 1987), and it is not
always good advice for women to act more like men in
order to claim the same resources and privileges.
Research on feminine modesty, for instance, shows that
women tend to present themselves more modestly than
do men (Daubman, Heatherington, & Ahn, 1992; Gould
& Slone, 1982; Heatherington, Daubman, Bates, & Ahn,
1993), and that a modest self-presentation style tends to
undermine perceived competence, particularly as com-
pared to those who self-promote in a stereotypically
masculine way (Rudman, 1998). However, if women
attempt to overcome this “deWciency” by behaving in a
more masculine self-promoting manner, they are per-
ceived as technically skilled but lacking in social compe-
tence. This lack of social competence then detracts from
their perceived hireability (Rudman, 1998). Similarly,
research on gender and leadership has found that female
leaders who attempt to establish their authority in a tra-
ditionally masculine (e.g., authoritative or directive)
manner are evaluated more harshly than their male peers
(Eagly, Makhijani, & Klonsky, 1992). Perhaps in
response to this resistance, women have tended to
develop a more participative leadership style, which is
correspondent with prescriptive gender roles for women
(Eagly & Johnson, 1990) and more eVective for them
than traditionally male leadership styles (Eagly, Joh-
annesen-Schmidt, & van Engen, 2003; Eagly, Karau, &
Makhijani, 1995).
The current research explores the question of whether
gender diVerences in the propensity to initiate negotia-
tions may be explained by diVerential treatment of men
and women when they attempt to negotiate.2 We exam-
ine whether women encounter more social resistance
than do men when they attempt to negotiate for higher
compensation and whether the gender of the evaluator
moderates that resistance. We investigate further
whether women are less inclined than men to initiate
compensation negotiations under those circumstances in
which they are more likely than men to encounter social
resistance. In this way, we are able to illuminate how
diVerential treatment of male and female negotiators
may motivate gender diVerences in the propensity to ini-
tiate negotiations over resources, such as compensation.
We focus on compensation negotiations, speciWcally,
because of their important economic implications and
because they represent a domain in which gender diVer-
ences in negotiated outcomes are well-documented (Bar-
ron, 2003; Bowles et al., 2005; Brett & Stroh, 1997;
Gerhart & Rynes, 1991; Stevens, Bavetta, & Gist, 1993).
Recent developments in the study of gender in negotia-
tion have made clear that gender eVects in negotiation
are situational (Bowles et al., 2005; Kray, Galinsky, &
Thompson, 2002; Kray, Reb, Galinsky, & Thompson,
2004; Kray, Thompson, & Galinsky, 2001; Stuhlmacher
& Walters, 1999; Walters, Stuhlmacher, & Meyer, 1998).
By focusing narrowly on compensation negotiations, we
intend from the outset to limit the potential for general-
ization of our results in terms of negotiating contexts.
That is to say, we would not expect women to encounter
more social resistance than men across all types of
potential negotiating contexts, nor would we expect
women always to be more reluctant than men to negoti-
ate. However, by demonstrating that women are more
reluctant than men to negotiate in a context in which
they face a greater social cost from doing so, we aim to
illuminate the broader phenomenon that gender diVer-
ences in the propensity to initiate negotiations may be
motivated by social incentives as opposed to individual
diVerences.
Initiation of compensation negotiations as a status
violation
Prescriptive sex stereotypes stem from men’s higher
status as compared to women within society (Conway,
Pizzamiglio, & Mount, 1996; Eagly & SteVen, 1984;
HoVman & Hurst, 1990; Jackman, 1994; Meeker &
Weitzel-O’Neill, 1977; Ridgeway & Bourg, 2004). Socie-
ties with more gender equity tend to espouse less sexist
2We use the term “gender” throughout, because “sex” connotes sta-
ble individual diVerences and we are studying the inXuence of social
situations on men’s and women’s behavior (Deaux & LaFrance, 1998).

86 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
beliefs (Glick et al., 2000). Within the U.S., as the pro-
portion of women in the workplace has grown and the
gender segregation of occupations has declined, women
have come to identify more with masculine personality
traits (Spence & Buckner, 2000; Twenge, 1997) and soci-
ety has come to view women of today and tomorrow as
more masculine than women of the past (Cejka & Eagly,
1999; Diekman & Eagly, 2000). Nevertheless, women are
still expected to fulWll prescriptions of feminine niceness
(e.g., warmth, kindness, sensitivity to the needs of oth-
ers), which are emblematic of their subordinate status
and therefore not fully compatible with all aspects of the
masculine personality type (Jackman, 1994; Prentice &
Carranza, 2002; Ridgeway, 2001a; Rudman & Glick,
2001).
In a recent test of contemporary college students’
responses to two classic gender identity scales (Bem Sex
Role Inventory by Bem, 1974; Personality Attributes
Questionnaire by Spence & Helmreich, 1978), Spence
and Buckner (2000) found that women identiWed more
strongly than men with all of the feminine items, but that
men identiWed more strongly than women with only 41%
of the masculine items. The majority of masculine items
with which both men and women identiWed tended to
relate to being active, independent, and expressing one’s
own beliefs, whereas the items with which men identiWed
more strongly than women tended to relate to being
forceful, competitive and in charge. Rudman and Glick
(2001) have argued that Spence and Buckman’s Wndings
reXect two distinct dimensions of the masculine stereo-
type: competence and dominance. The majority of mas-
culine traits with which both men and women equally
identify are consistent with the “competence” dimension
of the masculine stereotype. The minority of items with
which men identify more strongly than women represent
the “dominance” dimension of the masculine stereotype.
While both competence and dominance are associated
with higher status group members (Berger, Webster,
Ridgeway, & Rosenholtz, 1986; Carli, LaFleur, & Loe-
ber, 1995), Rudman and Glick argue that the traits and
behaviors associated with masculine “competence” pres-
ent less of a contradiction with the prescriptive norms of
feminine behavior than do the traits and behaviors asso-
ciated with masculine “dominance.”
Warmth and competence is an attractive, non-threat-
ening combination (Carli et al., 1995; Fiske, Cuddy,
Glick, & Xu, 2002). While a women who projects her
competence in a purely stereotypically masculine man-
ner runs a higher risk of social resistance than a similarly
self-presented man (Carli, 1990; Carli et al., 1995; Eagly
et al., 1992; Rudman, 1998; Rudman & Glick, 1999,
2001), women can eVectively convey their competence
and be as inXuential as men, if they soften their stereo-
typically masculine competence with feminine niceness
(Carli et al., 1995; Meeker & Weitzel-O’Neill, 1977;
Ridgeway, 1982; Rudman & Glick, 2001). For instance,
research on social inXuence shows that women can have
as much social inXuence as men, by ensuring that they
appear friendly as well as task-oriented rather than
purely task-oriented (Carli et al., 1995) or by communi-
cating their concern for the collective rather than their
personal self-interest (Ridgeway, 1982). By employing a
complementary combination of masculine competence
and feminine niceness, women can make substantive
contributions within traditionally male domains without
challenging the hierarchical structure of gender relations
(Carli et al., 1995; Meeker & Weitzel-O’Neill, 1977;
Ridgeway, 1982).
However, whereas women may temper resistance to
their displays of masculine competence by combining
them with stereotypically feminine behaviors, the display
of masculine dominance directly contradicts the deferen-
tial and relational character of the feminine gender role
(Eagly, 1987). Displays of masculine dominance by
women pose a direct challenge to the gender status hier-
archy and therefore a greater social risk than displays of
masculine competence (Rudman & Glick, 2001).
The act of attempting to negotiate competitively for
greater personal resources, such as compensation, calls
for a type of dominative masculine behavior that pre-
sents two problems for women. First, it violates prescrip-
tions of feminine niceness and that violation is likely to
engender social resistance (Burgess & Borgida, 1999;
Prentice & Carranza, 2002; Wade, 2001). Second,
women’s attempts to initiate negotiations over compen-
sation, in particular, may be resisted for the substance of
the claim as well as for the behavior inherent in the
request. Men’s relatively greater economic resources as
compared to women’s are a source of status for them
within society (Ridgeway, 2001b; Weber, 1968), and the
complementarity of masculine and feminine gender roles
(i.e., men as providers and women as caregivers) justiWes
and reinforces the asymmetric distribution of resources,
such as compensation, favoring men (Jackman, 1994;
Jost & Kay, 2005). As lower status group members mak-
ing claims to the privileges of higher status group mem-
bers, women are likely to appear inappropriately
demanding if they attempt to negotiate for higher levels
of compensation.
Hypothesis 1. The social cost of initiating negotiations
for higher compensation will be greater for women than
for men.
Hypothesis 2a. Perceived lack of niceness will explain the
social resistance to women who attempt to negotiate for
higher compensation as compared to those who do not.
Hypothesis 2b. Perceived demandingness will explain the
social resistance to women who attempt to negotiate for
higher compensation as compared to those who do not.
Because women’s initiation of negotiations for
resources, such as compensation, represents the type of

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 87
dominative masculine behavior that challenges the gen-
der status hierarchy, the relative social risk for women
(as compared to men) may be greater when attempting
to negotiate with a man as opposed to a woman (Carli,
1990; Deaux & Major, 1987; Ridgeway & Berger, 1986).
Gender diVerences in status are more salient in mixed-
gender than in same-gender interactions, and greatest
when the man is in a higher status role than the woman
(Berger, Fisek, Norman, & Zelditch, 1977; Deaux &
Major, 1987). Therefore, within the context of a poten-
tial compensation negotiation, gender diVerences in sta-
tus are likely to be most inXuential when the candidate is
female and the evaluator is male.
Research on prescriptive sex stereotypes has pro-
duced mixed Wndings with regard to the inXuence of the
gender of the perceiver. Some studies Wnd that female
evaluators penalize women for gender-role violations as
much as male evaluators (Butler & Geis, 1990; Heilman
& Chen, 2005; Heilman, Wallen, Fuchs, & Tamkins,
2004) and sometimes even more (Rudman, 1998). This
work suggests that prescriptive sex stereotypes are gen-
erally held and reinforced by both men and women
(Heilman & Chen, 2005). Other studies Wnd that male
evaluators are more resistant than female evaluators to
women who adopt high-status behaviors in their in
social inXuence attempts (Carli, 1990; Carli et al., 1995)
or who assume counterstereotypic high-status roles
(Eagly et al., 1992; Rudman & Kilianski, 2000; Schein,
2001). These other studies and research on social domi-
nance orientation (e.g., see Sidanius, Pratto, & Bobo,
1994) suggest that men may be more resistant than
women to violations that challenge the gender-status
hierarchy.
From this broader literature, the current work draws
inspiration most directly from the theory and research
on gender, status, and social inXuence, which suggest
that women’s persuasiveness with male evaluators, in
particular, is contingent on their ability to signal their
subordinate status (e.g., through niceness, tentativeness,
and other orientation) as well as their competence (Carli,
1990; Carli et al., 1995; Meeker & Weitzel-O’Neill, 1977).
Following these results and the logic that gender-related
status diVerences are likely to be most salient in compen-
sation negotiations when women are asking for more
personal resources from men, we propose that the gen-
der of the evaluator may moderate the relative social
cost for women (as compared to men) of attempting to
negotiate for higher compensation.
Hypothesis 3. The relative social cost for women as com-
pared to men of initiating negotiations for higher com-
pensation will be greater with male than with female
evaluators.
This Wrst set of hypotheses reXects the evaluator’s
perspective, describing the proposed inXuence of the
gender of the target and the gender of the perceiver on
evaluations of the initiation of compensation negotia-
tions. If these hypotheses are correct, then job candidates
should take these social costs into account when making
decisions about whether to negotiate. The following sec-
ond set of hypotheses reXects the candidate’s perspec-
tive. We propose that gender diVerences in the
propensity to initiate compensation negotiations will be
greater under those circumstances in which men and
women tend to face diVerential treatment when they
attempt to negotiate.
Hypothesis 4. Women will be more reluctant than men
to initiate compensation negotiations.
Hypothesis 5. Women’s relative reluctance (as compared
to men) to initiate compensation negotiations will be
greater when the evaluator is a man as opposed to a
woman.
Consistent with the expectation of greater social costs
from initiating compensation negotiations, we hypothe-
size that nervous feelings and anticipated negative social
consequences (backlash) will mediate gender diVerences
in the propensity to initiate negotiations.
Hypothesis 6a. Nervousness about attempting to negoti-
ate will explain gender diVerences in the propensity to
initiate compensation negotiations.
Hypothesis 6b. Anticipated backlash will explain gender
diVerences in the propensity to initiate compensation
negotiations.
In sum, as illustrated vividly by the opening quote, if
women are perceived to be “clamoring” for the same
resources as men, they may lose the grace of their ideal-
ized feminine niceness and be rejected for demanding
that which is not due to them. We argue that gender
diVerences in the social costs of attempting to negotiate
for resources, such as compensation, may help to explain
gender diVerences in the propensity to initiate negotia-
tions. In the following experiments, we examine the judg-
ments that male and female evaluators make when men
and women attempt to negotiate and the role of the gen-
der of the evaluator in shaping the decision to negotiate.
Overview of experiments
In Experiment 1, we conducted a preliminary test of
the hypothesis that there would be a higher social cost
for women than for men from initiating negotiations.
We then conducted a complementary set of experiments
that tested for gender eVects on both the likelihood of
social resistance to negotiation attempts and on the pro-
pensity to initiate negotiations. In Experiments 2 and 3,
participants adopted the role of senior manager in a cor-
poration and evaluated an internal candidate based
upon a transcribed (Experiment 2) or a videotaped
(Experiment 3) job placement interview. Across the

88 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
negotiation conditions, the candidates either accepted
their compensation oVers without comment (no ask) or
initiated negotiations (ask). In Experiment 4, we reversed
the participants’ perspectives. Using the same interview
scenario, participants adopted the role of the job candi-
date and evaluated whether to initiate negotiations. By
coupling these two perspectives on the same situation,
we were able to test whether gender diVerences in the
propensity to initiate negotiations (Experiment 4) would
reXect diVerential treatment (Experiments 2 and 3), sug-
gesting a social motivation for gender diVerences in the
initiation of negotiations.
Experiment 1
In Experiment 1, we conducted a preliminary test of
Hypotheses 1 and 3 in a 2 (gender of candidate) £2 (ini-
tiate negotiations: no ask vs. ask)£2 (gender of evalua-
tor) between-subjects design. Participants evaluated a
job candidate based on a resume and interview notes.
The interview notes indicated whether the candidate was
male or female and whether (or not) the candidate had
attempted to negotiate for extra compensation and job
beneWts.
Methods
Participants
The participants were 119 North American university
students (66 men, 53 women) recruited from various
points on a university campus (e.g., Xyers, dining halls)
to participate in a Hiring Decision Study. The median
age was 20 years (MD20.27, MinD18, Max D27). Fifty-
four percent of the participants were White, 26% were
Asian, 13% were African American, 5% were Hispanic,
and 2% reported “Other.” Participants received $5 for
completing the survey.
Procedure
After obtaining the participants’ consent to partici-
pate in the study, the experimenter distributed paper
packets containing background information on the
hiring decision, the candidate’s resume, and a set of
notes from an interview with the candidate. The back-
ground information instructed participants to adopt the
role of a commercial bank manager hiring a college stu-
dent for a summer internship and to review the resume
and interview notes. The candidate’s resume was identi-
cal across conditions. We gave the candidate a gender-
neutral Wrst name and then referred to the candidate in
the interview notes as either a man or a woman and with
gender-appropriate pronouns. Participants in the No
Ask condition read interview notes indicating a positive
evaluation of the candidate’s personal qualities and
experience (e.g., “[he/she] could deWnitely hit the ground
running”). Participants in the Ask condition read the
same set of general interview notes, and one additional
note indicating that the candidate had asked for more
compensation and had inquired about other additional
job beneWts (i.e., gym membership, access to a notebook
computer and metro-pass). The experimenter distributed
the four versions of the interview notes randomly among
participants. Participants responded on a 1–7 scale
(1 Dnot at all, 7 Dextremely) to two questions about
how likely they thought it was they or someone else at
the bank would hire this person for a summer internship.
After submitting their assessment of the candidate’s
hireability, the participants Wlled out an exit survey that
contained demographic questions, manipulation checks,
and a debrieWng form.
Results
All of the participants correctly identiWed whether the
candidate was male or female. For all studies we used
analysis of variance (ANOVA) to test the eVects of gen-
der of candidate, the ask manipulation, and gender of
evaluator on the dependent measure. We combined the
two hireability measures into one mean composite
dependent measure (D.83). As shown in Table 1, there
was a signiWcant main eVect for ask (F[1, 111] D29.97,
p< .001) and a signiWcant interaction of Gender of
Candidate £Ask (F[1, 111] D4.80, pD.03). None of the
other eVects were signiWcant (ps >.31). As indicated by
the means in Table 2, while the ask manipulation had a
Table 1
Experiment 1: ANOVA of hireability by gender of candidate, ask condition and gender of evaluator (ND119)
¤p< .05.
¤¤¤ p< .001.
Source df MS F 2
Gender of candidate 1 1.19 1.02 .01
Ask 1 34.84 29.97¤¤¤ .07
Gender of evaluator 1 0.28 0.24 .002
Gender of candidate£Ask 1 5.58 4.80¤.04
Gender of candidate£Gender of evaluator 1 0.13 0.11 .001
Ask £Gender of evaluator 1 0.82 0.71 .01
Gender of evaluator£Ask £Gender of candidate 1 0.02 0.02 .0001
Error 111

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 89
signiWcant negative eVect on the hireability of both male
and female candidates, the negative eVect for women
was more than twice as large as that for men.3
Discussion
The results of Experiment 1 supported our prediction
that evaluators would penalize a female candidate more
than a male candidate for initiating negotiations
(Hypothesis 1). Contrary to Hypothesis 3, the degree of
resistance to female relative to male candidates was the
same for male and female evaluators. It is noteworthy
that evaluators were less inclined, in general, to hire the
candidate who initiated negotiations. This suggested that
our ask manipulation may have been too strong, partic-
ularly for a candidate who had yet to receive a job oVer.
In Experiment 2, we adjusted the scenario to make the
potential negotiator an internal job candidate who had
already received an oVer for a new position, so that
negotiating for higher compensation would be generally
acceptable practice (Pinkley & Northcraft, 2000;
Thompson, 2005). We also tested two alternative scripts
for the initiation of negotiations that varied in terms of
the strength of argumentation that the candidate used to
make the case for higher compensation.
Experiment 2
In Experiment 2, we tested Hypotheses 1–3 in a 2
(gender of candidate) £3 (initiate negotiations: no ask
vs. moderate ask vs. strong ask) £2 (gender of evaluator)
between-subjects design. Participants evaluated a candi-
date based on a transcript of a job placement interview.
We ran two versions of the ask manipulation in order to
explore whether the manner in which the candidate initi-
ated negotiations would moderate the predicted interac-
tion eVect of Gender of Candidate£Ask on the
willingness to work with the candidate.
Methods
Participants
The participants were 299 college-educated adults
(152 men, 147 women) recruited by a market research
Wrm to participate in an online survey. The median age
was 37 years (MD39.2, MinD21, Max D64). Ninety-
one percent of the participants self-identiWed as White,
3% as Asian, 2% as Hispanic, 1% as African American/
Black, and 3% as “Other.” The median work experience
was 17 years (MD18.30, Min D0, Max D50). Eighty
percent were currently employed. Sixty-nine percent of
the participants had management experience. Those with
management experience reported an average of 9.7 years
of management experience. In exchange for completing
the survey, the participants received points redeemable
for prizes from the market research Wrm.
Procedure
Participants accessed the survey remotely by respond-
ing to an email from the market research Wrm and clicking
on a link to the Evaluation Survey website. After indicat-
ing their consent to participate in the study, the website
directed the participants to a page of Background Infor-
mation with links to a transcript from a job placement
interview and an Evaluation Survey. The Background
Information instructed participants to imagine they were
senior managers in a corporation, and informed them that
their task was to evaluate an internal candidate for a man-
agement position in their department. The Background
Information continued on to explain that the candidate
was in the process of completing a management training
program before being assigned to a more senior manage-
ment position within the company and that the candidate
(like all those oVered management positions) had gradu-
ated from a top school and performed well in the training
program. The participant (as senior manager) wanted
people in the department who were good team players
and who worked well with other people.
After reviewing the Background Information, the
website randomly assigned participants to read one of
six interview transcripts (2 [candidate: male, female] £3
[ask: no, moderate, strong]). We gave the candidate a
gender neutral Wrst name, and manipulated the gender of
the candidate by referring to the candidate as Mr. or Ms.
and with gender-appropriate pronouns. The Wrst two
interview questions and responses were identical across
conditions. The questions related to the candidate’s
3We ran a second preliminary study using the same basic procedure
described in Experiment 1. Participants were 176 adults recruited from
the website of a university-based experimental research laboratory.
The interview notes in the Ask condition said that the candidate had
asked to be paid at the top of the salary range. We observed a signiW-
cant interaction between Gender of Candidate £Ask, such that evalu-
ators penalized female candidates more than the male candidates for
attempting to negotiate (F[1, 172] D4.62, p< .04). There were no sig-
niWcant eVects of gender of evaluator (ps>.71).
T
a
bl
e
2
Experiment 1: Means of hireability by gender of candidate and ask
condition
Note: We report standard deviations in parentheses below the means.
DiVerent superscripts between ask conditions indicate signiWcant mean
diVerences, a,b at the level of p<.05 and a,c at the level of p< .001.
Male candidate Female candidate
No ask Ask EVect of ask No ask Ask EVect of ask
MMd MMd
5.94a5.26b0.65 6.19a4.63c1.44
(0.97) (1.10) (0.86) (1.26)
nn nn
26 33 29 31

90 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
management training and work experience, and the
responses indicated that the candidate had enjoyed and
learned a lot in the training program and had some man-
agement experience running a school newspaper in col-
lege (431 words). In order to signal that the candidate
was a good team player and worked well with other peo-
ple, the response to the second question included state-
ments such as, “I built a really strong team of people
working together” and “I learnedƒ how to get people
motivated to do a good job.” We embedded the ask
manipulation in the candidate’s response to the third
question about whether they had received their salary
and beneWts oVer. Appendix A contains the wording of
the third question and the scripts for the no ask, moder-
ate ask and strong ask conditions.
After reviewing the transcript, the participants clicked
on a link to the Evaluation Survey. The Wrst step of the
Evaluation Survey contained a list of characteristics,
which we used to create our potential mediators (i.e.,
perceived niceness and demandingness). To create a
measure of feminine niceness, we used seven of the ten
items from Rudman and Glick (1999) Social Skills Index
(good listener, helpful, kind, likeable, sensitive to the
needs of others, supportive, and warm). We removed
three (friendly, popular, and sincere) that seemed more
relevant to social skills in general than prescriptions of
feminine niceness, and added four more stereotypically
feminine characteristics (agreeable, cooperative, modest,
and nurturing). We included another set of twelve words
associated with demandingness (arrogant, cocky,
demanding, dominating, obnoxious, overbearing, over-
conWdent, presumptuous, pushy, self-centered, ungrate-
ful, and unreasonable). Participants rated on a scale of
1–7 (1 Dnot at all, 7 Dperfectly) how much each of the
words characterized their impression of the candidate.
Finally, to create our dependent measure of willingness
to work with the candidate, participants rated how bene-
Wcial it would be for them to have this person working
for them, how much they would enjoy having this person
working for them, and how likely it was that they would
hire this person (1 Dnot at all, 7 Dextremely).
After the participants submitted their responses to the
Evaluation Survey, the website directed them to an exit
survey that contained manipulation checks, demo-
graphic questions, and debrieWng materials. The manip-
ulation checks tested whether the participants knew the
gender of the candidate and how the candidate had
responded to the third question (e.g., asked for high sal-
ary, asked for bonus). The debrieWng materials included
a question about what the participants thought the study
was about while they were participating in it.
Results
We removed 63 cases because the participants
failed to correctly identify the gender of the candidate
or whether the candidate asked for a higher salary.4
None of the participants reported that they suspected
that it was a gender-related study. The sample of data
analyzed included 236 participants (121 men, 115
women).
Dependent measure: willingness to work with the
candidate
We combined the three measures of the evaluators’
willingness to work with the candidate to create one
mean composite dependent measure (D.91). Prelimi-
nary analyses of the data showed that how the candi-
date attempted to negotiate for higher compensation
(i.e., moderate ask vs. strong ask) had no signiWcant
eVect on the inXuence of the gender of the candidate or
the gender of the evaluator on the willingness to work
with the candidate. In an ANOVA that restricted the
analysis to the moderate ask and strong ask conditions,
there was no signiWcant interaction between of Gender
of Candidate £Type of Ask (F[1, 1 50] D0.44, pD.51),
Gender of Evaluator £Type of Ask (F[1, 150] D2.10,
pD.15), or Gender of Candidate £Gender of
Evaluator £Type of Ask (F[1, 150] D0.21, pD.65). This
indicates that the eVects of gender of candidate and
gender of evaluator on the willingness to work with the
candidate did not diVer by whether the ask manipula-
tion was moderate or strong. Therefore, for the sake of
parsimony, we collapsed the moderate ask and the
strong ask conditions into one ask condition and pre-
sented our tests of the hypothesized interactions com-
paring the no ask condition to the combined ask
condition.
Table 3 displays the results of the ANOVA of the
willingness to work with the candidate by gender of can-
didate, ask condition and gender of evaluator. We
observed a signiWcant main eVect for ask
(F[1, 228] D16.38, p< .001) and a signiWcant interaction
of Gender of Candidate £Ask (F[1, 228] D6.74, p<.02).
None of the other eVects were signiWcant (ps > .15). As
can be seen from the means displayed in Table 4,
attempting to negotiate for higher compensation had no
signiWcant eVect on the evaluators’ willingness to work
with a male candidate, t(116) D0.92, pD.92. In contrast,
attempting to negotiate signiWcantly reduced the evalua-
tors’ willingness to work with a female candidate,
t(116) D4.72, p< . 001.
4Using these criteria, we excluded more than twice as many cases in
the ask as in the no ask condition. We suspect that this was due to the
phrasing of the manipulation check questions, which tested the partici-
pants’ memory of what the candidate had asked for during the inter-
view. If participants did not read the full script, then they were more
likely to get these questions wrong in the ask vs. no ask condition.
There was no other discernable pattern in the cases excluded. Exclu-
sion of these cases had no eVect on the results.

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 91
Potential mediators: niceness and demandingness
We conducted a principal components factor analy-
sis of the impression items with promax rotation. As
anticipated, this analysis revealed two factors, which
we combined into composite indicators of perceived
niceness (eigenvalue D11.41; D.92) and perceived
demandingness (eigenvalueD2.18; D.95). Table 5 dis-
plays means by condition for perceived niceness and
demandingness. Niceness and demandingness were sig-
niWcantly correlated rD¡.66, p< .001. The correlation
between niceness and willingness to work was rD.75
(p< .001) and between demandingness and willingness
to work was rD¡.57 (p< .001). In a regression model
of willingness to work on niceness and demandingness,
each potential mediator had a signiWcant, independent
eVect on the dependent variable (D.65, p< .001 for
niceness; D¡.15, p< .02 for demandingness; adjusted
R2D.56).
We used Sobel tests (Sobel, 1982) for all of our media-
tion analyses.5 As predicted, both niceness and demand-
ingness fully mediated the negative eVect of attempting
to negotiate on the evaluators’ willingness to work with
a female candidate, Sobel zD4.90, p< .001 for niceness
and Sobel zD5.26, p<.001 for demandingness. These
variables, however, did not mediate the interaction eVect
of Gender of Candidate£Ask on willingness to work,
Sobel zD1.47, pD.14 for niceness and Sobel zD1.12,
pD.26 for demandingness.
Discussion
The results of Experiment 2 supported our hypoth-
esis that women would incur a greater social cost from
attempting to negotiate for higher compensation than
would men (Hypothesis 1). Indeed, there was no sig-
niWcant decline in the evaluators’ willingness to work
with a male candidate who attempted to negotiate (vs.
not). Women, in contrast, faced a large penalty—the
negative eVect of the ask manipulation was more than
5.5 times greater for women than for men. Interest-
ingly, whether the candidate asked simply (moderate
ask) or assertively (strong ask) had no eVect on the rel-
atively larger social cost for women as compared to
men.
Mediation analyses further supported our hypotheses
that social resistance to a female candidate who initiated
negotiations would be explained by the evaluators’ per-
ceptions that she was not nice and overly demanding
(Hypotheses 2a and 2b). It is noteworthy, however, that
T
a
bl
e
3
Experiment 2: ANOVA of willingness to work with candidate by gender of candidate, ask condition and gender of evaluator (ND236)
aModerate and strong ask conditions combined into one ask condition as compared to no ask condition.
¤p< .05.
¤¤¤ p< .001.
Source df MS F 2
Gender of candidate 1 3.27 2.01 .01
Aska1 26.62 16.38¤¤¤ .07
Gender of evaluator 1 3.24 1.99 .01
Gender of candidate£Ask 1 10.95 6.74¤.03
Gender of candidate£Gender of evaluator 1 0.16 0.10 .0004
Ask £Gender of evaluator 1 1.53 0.94 .004
Gender of candidate£Ask £Gender of evaluator 1 1.68 1.03 .01
Error 228
T
a
bl
e
4
Experiment 2: Means of willingness to work with candidate by gender
of candidate and ask condition
Note: We report standard deviations in parentheses beside the means.
DiVerent superscripts (a,b) between ask conditions indicate signiWcant
mean diVerences at the level of p< .001.
Male candidate Female candidate
No ask Ask EVect
of ask
No ask Ask EVect
of ask
MMdMMd
3.24a (1.39) 3.02a (1.24) .17 3.94a (1.16) 2.80b (1.22) .95
nn nn
45 73 40 78
Table 5
Experiment 2: Means of potential mediators, perceived niceness and
perceived demandingness, by gender of candidate and ask condition
Note: We report standard deviations in parentheses beside the means.
DiVerent superscripts (a,b) between ask conditions indicate signiWcant
mean diVerences at the level of p< .001.
Male candidate Female candidate
No ask Ask No ask Ask
MMMM
Niceness
4.14a (0.83) 3.54b (0.84) 4.55a (0.75) 3.62b (0.94)
Demandingness
3.20a (0.96) 4.60b (1.07) 2.83a (1.05) 4.59b (1.33)
nnnn
45 73 40 78
5For each mediation analysis, we also calculated the Aroian and the
Goodman tests and report if the results of the tests diVer across the
three methods (Aroian, 1947; Goodman, 1960).

92 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
neither niceness nor demandingness mediated the inter-
action eVect between the gender of the candidate and the
ask manipulation. This suggests that, while perceived
niceness and demandingness help to explain the evalua-
tors’ resistance to a woman who initiated compensation
negotiations, they do not help to explain why there was a
relative lack of resistance to the same behavior by a man.
Hypothesis 3 was not supported. There was no diVer-
ence in the extent to which male and female evaluators
penalized female relative to male candidates for initiat-
ing negotiations.
Experiment 3
In Experiment 3, we conducted another test of
Hypotheses 1–3 in a 2 (gender of candidate) £2 (initiate
negotiations: no ask vs. ask)£2 (gender of evaluator)
between-subjects design. The scenario was identical to
the one used in Experiment 2, except that the partici-
pants evaluated candidates based on their behavior in a
videotaped interview. The videotaped candidates used
the no ask and strong ask scripts that participants read
in Experiment 2.
Our motivations for conducting the gender of candi-
date and ask manipulations with video rather than writ-
ten scripts were multifold. Video broadens the
bandwidth of communication relative to written tran-
scripts by conveying additional nonverbal cues and,
thereby, increases the social presence of the actor (Short,
Williams, & Christie, 1976; Walther & Parks, 2002).
Accordingly, we believed that the video would heighten
the realism of the participants’ evaluation task and
enhance the external validity of the results. While it is
unusual to watch a videotape of a person’s interview
performance, it is even more out of the ordinary to
review a transcript of their responses to interview ques-
tions. Interviewers generally beneWt from hearing and
seeing a candidate speak when making assessments
about how much they would like to work with the per-
son. Finally, Hypothesis 3 (regarding eVects of the gen-
der of the evaluator) was motivated in part by research
on gender and social inXuence, which had employed vid-
eotapes (Carli et al., 1995). We anticipated that a richer
communication medium might aVect how the partici-
pants responded to the job candidates’ behavior (Wal-
ther & Parks, 2002).
Methods
Participants
The participants were 285 adults (107 men, 178
women). Participants signed up to complete a web-based
survey called the Evaluation Study at one of two diVer-
ent websites that recruit experimental participants. Each
of the websites lists studies that adults over 18 years of
age may sign up to participate in for pay. The Evalua-
tion Study was advertised as follows: “Participants
watch a short video of an interview and then answer a
series of questions about the impression created by the
interviewee.” Participants signed up to complete the
study, and then the experimental laboratory sent them a
link to the online survey that was valid for a limited time
period of no longer than three days. There were no sig-
niWcant diVerences in the results obtained from the two
sites, so we pooled the data. The median age was 29
years (MD31.12, Min D18, Max D62). Sixty-nine per-
cent of the participants self-identiWed as White/Cauca-
sian, 24% as Asian, 3% as Hispanic, 2% as African
American/Black and 2% as Other (or no response).
Thirty percent of the participants were currently stu-
dents. Forty-four percent had no or some college, 32%
were college graduates, and 24% had more than a college
degree. The median work experience was seven years
(MD10.08, Min D0, Max D40). Forty-Wve percent of
the participants had management experience. Those with
management experience reported an average of 6.32
years of management experience. Participants received
$8 for completing the online survey.
Procedure
The procedure was identical to that used in Experi-
ment 2 except that the participants evaluated the candi-
date based on a videotape of their responses to the
interview questions (as opposed to a transcript). The
website randomly assigned participants to watch one of
eight videos, so that each participant saw a head-and-
shoulders shot of one of four candidates (two male, two
female) who either accepted their compensation oVer
without comment or attempted to negotiate for a higher
salary and bonus (no ask vs. ask). Participants in the ask
condition saw the same video as the participants in the
no ask condition, plus the segment in which the candi-
date attempted to negotiate for higher compensation.
After watching the video, participants clicked on a link
to the Evaluation Survey, which contained the same
items reported in Experiment 2. The survey also included
three questions about how competent they perceived the
candidate to be (i.e., how likely it was that the candidate
had the communication skills and analytical ability to be
an eVective manager and how conWdent the evaluator
was that the candidate would be an eVective manager).
We had conceptualized the initiation of compensation
negotiations as a form of dominative masculine behavior
that is distinct from demonstrations of masculine com-
petence, so we thought we should test whether (contrary
to our expectations) women’s perceived competence
would mediate the eVect of the ask manipulation.
Manipulation of gender of candidate. We recruited two
male and two female actors and recorded them enacting
both the ask and no ask scripts in order to avoid

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 93
confounds between actor and condition. We recruited
undergraduate actors so that all of the candidates would
appear to be in the same age cohort and age appropriate
for the scenario. We selected actors whom we perceived
to be of average attractiveness and asked them to wear
professional dress. During the rehearsal for the taping,
we coached them to enact the script as similarly as possi-
ble to one another (e.g., by providing instruction on tone
and pace of voice, etc.). The actors spoke for 2.3 min on
average in the no ask condition and for 3.5 min on aver-
age in the ask condition. The male (as compared to
female) actors took four seconds more on average to
complete the no ask script and ten seconds more on
average to complete the ask script.
Results of video rating
While the videos enriched and enhanced the realism
of the ask manipulation, they provided less control than
a pen-and-paper experiment. Although we coached the
actors to enact the script as similarly as possible, we
could not control for natural variation in the actors’
appearance and nonverbal behavior. For instance,
women tend to smile more often than men (Dovidio,
Brown, Heltman, & Ellyson, 1988; Hall & Halberstadt,
1986; LaFrance, Hecht, & Paluck, 2003). If the female
actors’ behavior diVered more between the no ask and
ask conditions than did the male actors’ behavior (e.g.,
they smiled relatively less), then that would suggest an
alternative explanation for any Wndings of interaction
eVects between the gender of the candidate and the ask
manipulation.
One-hundred and ninety-six participants (93 men, 103
women) rated the actors (Mdn age D21). The online
video-rating survey randomly assigned participants to
watch one of the eight experimental videos with the
sound turned oV, so that the substance of the script
would not bias their perceptions of the actor. Partici-
pants rated the candidates’ appearance in terms of their
age (1 D20–24, 2 D25–29, 3 D30–34, 4 D35–39), socio-
economic status (1 Dlow class, 2 Dmiddle class,
3Dupper class) and physical attractiveness (1–7 scale:
1Dnot at all, 7Dextremely). They rated on a 1–7 scale
(1 Dnot at all, 7 Dperfectly) how well the following
adjectives characterized the actors’ facial expressions:
angry, happy, friendly, frowning, smiling, and scowling.
We used the participants’ ratings of the actors’ facial
expressions to create composite indicators of friendly
expressions (happy, friendly, smiling; D.88) and angry
expressions (angry, frowning, scowling; D.80). We used
ANOVA for the 1–7 scale variables and ordered Logit
regression for the other categorical variables to test
whether the ask manipulation, the gender of the candi-
date, or the gender of the evaluator inXuenced raters’
perceptions of the appearance and nonverbal behavior
of the candidates.
There were no signiWcant two-way or three-way inter-
action eVects among the gender of candidate, gender of
evaluator, and the ask conditions on the raters’ percep-
tions of the appearance or nonverbal behavior of the
candidates (ps > .13). There were two main eVects on the
raters’ perceptions of the candidates’ appearance and
non-verbal behavior. Raters perceived the female candi-
dates to be more physically attractive than the male can-
didates (F[1, 188] D27.08, p< .001, female MD3.97,
SD D1.12, male MD3.18, SD D1.12) and to make more
friendly expressions (F[1,188] D13.86, p< .001, female
MD4.22, SD D1.27, male MD3.47, SD D1.27).
Although the female candidates came across as more
attractive and friendly than the male candidates, these
impressions, importantly, did not diVer across the ask
conditions.
Results of candidate evaluation
We now turn to the analysis of the participants’ will-
ingness to work with the candidates. All of the partici-
pants correctly identiWed the gender of the candidate,
and none of the participants reported that they sus-
pected that it was a gender-related study. We removed
38 cases because the participants failed to correctly iden-
tify how the candidate had responded to the third ques-
tion.6 The data analyzed contained 247 cases (95 men,
152 women).
Dependent measure: willingness to work with the
candidate
As in Experiment 2, we combined the three measures of
the evaluators’ willingness to work with the candidate to
create one composite dependent measure (D.91). Table 6
displays the results of the ANOVA of willingness to work
with the candidates by gender of candidate, ask condition,
and gender of evaluator. We observed a signiWcant main
eVect for ask (F[1,239] D34.71, p< .001) and a signiWcant
three-way interaction of Gender of Candidate£Ask £
Gender of Evaluator (F[1,239] D5.81, p<.02). No other
eVects were signiWcant (ps>.11).
7
For the purpose of interpreting the three-way interac-
tion, we conducted separate ANOVAs for female candi-
dates and male candidates. The ANOVA of the
willingness to work with female candidates by ask
6The website required participants to launch the video before pro-
ceeding to the Evaluation Survey, but it did not prevent participants
from closing down the video midstream. We suspect that these partici-
pants did not watch the entire video. The excluded cases appeared to
be randomly distributed across conditions, and had no eVect on the re-
sults.
7For the ANOVA presented in Table 6, we tested whether there
were actor-speciWc eVects—that is, whether evaluations of the female
actors diVered from one another and whether the evaluations of the
male actors diVered from one another. We found no statistically sig-
niWcant diVerences between the actors of the same gender (all ps > .14).

94 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
condition and gender of evaluator revealed a signiWcant
main eVect for the ask manipulation (F[1, 120] D20.69,
p< .001, 2D.15) and no signiWcant eVects by gender of
evaluator (Fs <1.35, ps>.24, 2< .02). As indicated by
the means in Table 7, both male and female evaluators
were less inclined to work with female candidates who
initiated negotiations as compared to those who did not,
t(74) D2.61, p< .02 for female evaluators and
t(46) D3.94, p< .001 for male evaluators. ANOVA of the
willingness to work with male candidates showed a sig-
niWcant main eVect for the ask manipulation
(F[1, 119] D14.21, p< .001, 2D.11) and a signiWcant
interaction eVect of Ask£Gender of Evaluator
(F[1, 119] D5.27, pD.02, 2D.04). There was no signiW-
cant main eVect of gender of evaluator, (F[1,119] D0.36,
pD.55, 2< .01). As shown by the means in Table 7,
attempting to negotiate for higher compensation had no
signiWcant eVect on male evaluators’ willingness to work
with male candidates, t(45) D0.90, pD.37. However, the
ask manipulation had a signiWcantly negative eVect on
female evaluators’ willingness to work with male candi-
dates, t(74) D5.06, p< .001.
Another way of interpreting the three-way interaction
is to split the ANOVA by gender of evaluator. For the
ANOVA for female evaluators, there was a signiWcant
main eVect for the ask manipulation (F[1, 148] D27.82,
p< .001, 2D.16) and no signiWcant main or interaction
eVects by gender of candidate (Fs<2.24, ps>.13,
2< .02). This indicates that male and female candidates
incurred the same social cost for initiating negotiations
with female evaluators. With male evaluators, there was
a signiWcant main eVect for the ask manipulation
(F[1, 91] D11.38, pD.001, 2D.11), which was qualiWed
by a signiWcant interaction eVect of Gender of
Candidate £Ask (F[1, 91] D4.27, pD.04, 2D.05), and
no signiWcant main eVect of gender of candidate
(F[1, 91] D0.71, pD.40, 2< .01). As shown by the means
in Table 7, with male evaluators, female candidates
incurred a signiWcant social cost for initiating negotia-
tions, whereas male candidates did not.
Potential mediators: niceness, demandingness, and
competence
We replicated the principal components factor analy-
sis reported in Experiment 2. We combined the Wrst fac-
tor (eigenvalue D12.32) into a composite indicator of
perceived niceness (D.94) and the second factor
(eigenvalue D2.93) into a composite indicator of per-
ceived demandingness (D.96). We combined the three
measures of how competent the evaluators perceived the
candidates to be into one composite indicator of per-
ceived competence (D.87).
Table 8 displays means by condition for perceived
niceness, demandingness and competence. Niceness and
demandingness were signiWcantly correlated rD¡.62,
p< .001. Evaluators perceived nicer candidates to be
more competent (rD.61, p< .001) and more demanding
candidates to be less competent (rD¡.37, p< .001).
T
a
bl
e
6
Experiment 3: ANOVA of willingness to work with candidate by gender of candidate, ask condition and gender of evaluator (ND247)
¤p< .05.
¤¤¤ p< .001.
Source df MS F 2
Gender of candidate 1 4.17 2.51 .01
Ask 1 57.78 34.71¤¤¤ .13
Gender of evaluator 1 0.45 0.27 .001
Gender of candidate£Ask 1 0.97 0.58 .002
Gender of candidate£Gender of evaluator 1 0.14 0.09 .0003
Ask £Gender of evaluator 1 0.85 0.51 .002
Gender of candidate£Ask £Gender of evaluator 1 9.67 5.81¤.02
Error 239
T
a
bl
e
7
Experiment 3: Means of willingness to work with candidate by gender of candidate, ask condition, and gender of evaluator
Note: We report standard deviations in parentheses beside the means. DiVerent superscripts between ask conditions indicate signiWcant mean diVer-
ences, a,b at the level of p< .001 and a,c at the level of pD.01.
Gender of evaluator Male candidate Female candidate
No ask Ask EVect of ask No ask Ask EVect of ask
MMd MMd
Male 4.10a (1.07) 3.75a (1.53) 0.27 4.85a (1.27) 3.43b (1.22) 1.14
Female 4.49a (1.03) 3.09b (1.38) 1.16 4.53a (1.48) 3.68c (1.33) 0.60
nn nn
Male 28 19 25 23
Female 41 35 38 38

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 95
Evaluators were signiWcantly more inclined to work with
candidates they perceived to be nice (rD.79, p< .001)
and competent (rD.74, p<.001), and signiWcantly less
inclined to work with candidates they perceived to be
demanding (rD¡.63, p< .001). In a regression model of
willingness to work on niceness, demandingness, and
competence, each potential mediator had a signiWcant,
independent eVect on the dependent variable (D.39,
p< .001 for niceness; D¡.23, p<.001 for demanding-
ness; D.41, p< .001 for competence; adjusted R2D.76).
Consistent with Hypotheses 2a and 2b, both perceived
niceness and perceived demandingness fully mediated the
eVect of the ask manipulation on the evaluators’ willing-
ness to work with female candidates, Sobel zD5.03,
p< .001 for niceness and Sobel zD5.46, p< .001 for
demandingness. The ask manipulation had no eVect on the
perceived competence of female candidates (p> .20), so it
was not a potential mediator. Post hoc analyses showed
that perceived niceness, demandingness and competence
mediated the interaction eVect of Ask£Gender of Evalua-
tor on the willingness to work with male candidates (Sobel
zD2.94, p<.01 for niceness, Sobel zD2.01, pD.04 for
demandingness, Sobel zD2.03, pD.04 for competence),
explaining why female evaluators penalized male candi-
dates for asking but male evaluators did not. None of the
potential mediators explained the three-way interaction of
Gender of Target £Ask £Gender of Evaluator on the will-
ingness to work with the candidates (Sobel test ps>.12).
Discussion
The results of Experiment 3 supported Hypothesis 3:
women paid a higher social cost for initiating compensa-
tion negotiations than men, but only with male evalua-
tors. Attempting to negotiate for higher compensation
had no eVect on men’s willingness to work with men, but
it had a signiWcantly negative eVect on men’s willingness
to work with women. Women penalized men and women
equally for attempting to negotiate.
The mediation analyses supported Hypotheses 2a and
2b. Even though the evaluators perceived the women
who initiated compensation negotiations to be just as
competent as those who did not, they were disinclined to
work with them because they appeared less nice and
more demanding. The mediation analyses did not
explain why men evaluated male and female candidates
diVerently, while women did not.
The eVects of gender of evaluator in Experiment 3 fol-
low a similar pattern to results observed in other video-
based research on gender and social inXuence, in which
female evaluators held men’s and women’s inXuence
attempts to a more similar standard than did male evalu-
ators (Carli et al., 1995). However, these results contrast
with those of Experiments 1 and 2, in which both male
and female evaluators penalized women more than men
for initiating negotiations. It is possible that the media
richness of the video manipulation caused the women to
perceive the male candidates’ negotiation attempts
diVerently; when we presented the same behavior in a
more socially distant, text format, the behavior did not
seem so unattractive.
While there was some inconsistency in women’s eval-
uations of men between the video-based and text-based
experiments, men evaluated women more negatively
than men for initiating negotiations across all three stud-
ies. This suggests that, while men may be as hesitant as
women to initiate compensation negotiations with a
female evaluator, women should clearly be more hesitant
than men about initiating compensation negotiations
with a male evaluator. In Experiment 4, we instructed
participants to adopt the perspective of the candidate in
the same scenario used in Experiments 2 and 3, in order
to test whether gender diVerences in the propensity to
initiate negotiate would reXect gender diVerences in the
social risks of doing so.
Experiment 4
In Experiment 4, we tested our second set of Hypoth-
eses 4–6 in a 2 (gender of participant)£2 (gender of
evaluator) between-subjects experimental design. Adopt-
ing the perspective of the candidate in the job placement
interview scenario, participants reviewed two potential
strategies for how to respond to a question about their
salary and beneWts oVer. The two strategies were identi-
cal to the no ask and ask scripts in Experiment 3. In
order to test for eVects by gender of evaluator, we
manipulated whether the participants (as candidate)
wanted to work for a man or a woman.
Table 8
Experiment 3: Means of potential mediators, perceived niceness, per-
ceived demandingness, and competence, by gender of candidate, ask
condition, and gender of evaluator
Note: DiVerent superscripts between ask conditions indicate signiWcant
mean diVerences, a,b at the level of p<.05 and a,c at the level of p< .001.
Gender of evaluator Male candidate Female candidate
No askAsk No askAsk
MSDMSDMSDMSD
Niceness
Male 4.18a0.81 3.77a1.08 4.63a1.01 3.92b1.13
Female 4.66a0.90 3.18c1.00 4.82a0.96 3.67c0.91
Demandingness
Male 3.57a0.99 5.13c1.22 3.02a1.06 4.85c0.81
Female 2.86a1.24 5.35c1.17 2.82a1.22 4.88c1.10
Competence
Male 4.17a1.17 4.44a1.25 4.83a0.93 4.54a1.17
Female 4.54a1.22 3.82b1.38 4.45a1.44 4.18a1.15
nnnn
Male 28 19 25 23
Female 41 35 38 38

96 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
Method
Participants
The participants were 367 adults (184 men, 183
women) recruited from the website of a university-based
experimental research laboratory to participate in an
Interview Preparation Study. The study was advertised
as follows: “Participants will review a job interview sce-
nario and answer questions about how they would pre-
pare for the interview. After evaluating the interview
information, participants will answer two brief surveys
about themselves.” The median age of the participants
was 21 years (MD23.92, MinD18, Max D64). Sixty per-
cent of the participants self-identiWed as White/Cauca-
sian, 18% as Asian, 10% as African American/Black, 8%
as Hispanic, and 4% as Other. Seventy-eight percent of
the participants were full-time university students. The
median work experience was four years (MD5.72,
Min D0, MaxD42). Participants received $15 for partic-
ipating in the study.
Procedure
After obtaining the participants’ consent to partici-
pate in the study, the experimenter distributed paper
packets containing background information on the
interview preparation scenario. The background mate-
rial instructed participants to imagine that they had just
completed a management training program in prepara-
tion for a more senior management position within their
company. According to the background materials, the
participants were about to participate in a placement
interview to determine the department to which they
would be assigned. Participation in the study involved
preparing for that interview.
Ask vs. no ask plan. The background information pre-
sented participants with three questions that the inter-
viewer was likely to ask, and provided prepared
answers to the Wrst two questions. (The answers to the
Wrst two questions were identical to the responses given
by the candidates in Experiments 2 and 3.) The task for
participants was to decide between two alternative
answers (i.e., Plan A or Plan B) to the third question,
which related to their salary and beneWts oVer from the
company. Plan A was the no ask script used in Experi-
ments 2 and 3, and Plan B was the strong ask script
used in Experiment 2 and the ask script used in Experi-
ment 3.
Gender of evaluator. Consistent with Experiments 2 and
3, the background information informed the partici-
pants that the senior manager they really wanted to
work for was known to hire people who are good team
players and who work well with other people. In order to
manipulate the gender of the evaluator, we gave the
manager a gender neutral name, and then referred to the
manager either as a man or as a woman and with corre-
sponding male or female pronouns.
Measures. After reviewing the background materials, the
participants completed an online Interview Preparation
Survey in which they assessed each of the plans. Partici-
pants rated each plan individually in terms of how they
would feel about using the plan and whether they antici-
pated negative social consequences from using the plan.
We used these ratings to create our potential mediators:
nervousness and anticipated backlash. Participants rated
on a scale of 1–7 (1Ddisagree completely, 7 Dagree
completely) how embarrassed, comfortable, nervous,
relaxed, and anxious they would feel using each of the
plans. Participants answered two questions on a scale of
1–7 (1 Ddisagree completely, 7 Dagree completely)
about whether they thought a man (or woman) like
[name of manager] would not want to hire or work with
them if they used either plan.8 Finally, participants rated
which plan they would be more likely to use relative to
the other (1 DI would be much more likely to use [the no
ask plan], 7 DI would be much more likely to use [the
ask plan]). This was our dependent measure of the pro-
pensity to initiate negotiations.
Because our hypotheses derived from the basic prop-
osition that gender diVerences in the propensity to initi-
ate negotiations are motivated by diVerences in the
social feedback that men and women receive when they
attempt to negotiate, we thought we should explore the
alternative hypothesis that gender diVerences in the pro-
pensity to initiate negotiations are a function of the
negotiator’s gender identity (i.e., the extent to which the
negotiator identiWes with masculine and feminine per-
sonality traits regardless of whether the negotiator is a
man or a woman). After submitting their responses to
the Interview Preparation Survey, the participants com-
pleted a second online Personality ProWle Survey that
included the 30 items from the Bem Sex-Role Inventory
(BSRI) short form (Bem, 1981).
A Wnal online exit survey contained manipulation
checks, demographic questions and debrieWng materials.
The manipulation checks tested whether the participants
knew the diVerence between Plans A and B and the gen-
der of the evaluator. DebrieWng materials included a
question about what they thought the study was about
while they were participating in it.
Results
We removed 18 cases because the participants failed
to provide correct answers on the manipulation checks.
We removed 8 cases because participants suspected the
8There were no gender eVects on nervousness or anticipated back-
lash for the no ask plan (ps > .12). To streamline the presentation of
results, we do not discuss these variables further.

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 97
study was about gender in hiring or negotiation.9 The
data analyzed contained 341 cases (171 men, 170
women).
Dependent variable: use of ask vs. no ask plan
The overall mean rating of the likelihood of using the
ask vs. the no ask plan was signiWcantly lower than the
indiVerence point between the two options (4), indicating
that participants tended to favor the no ask over the ask
plan, MD2.24, t(340) D18.79, p< .001. As shown in
Table 9, ANOVA revealed a signiWcant interaction of
Gender of Participant £Gender of Evaluator on the
likelihood of using the ask vs. no ask plan
(F[1, 337] D4.13, pD.04), but no main eVects for gender
of participant or gender of evaluator (ps> .28). As shown
in Table 10, when the evaluator was male, women were
signiWcantly less inclined than men to use the ask vs. no
ask plan, t(159) D2.06, pD.04. Consistent with the
results of Experiment 3, when the evaluator was female,
men were as likely as women to use the ask vs. no ask
plan, t(178) D¡0.72, pD.47.
Potential mediators: nervousness, anticipated backlash,
and gender identity
Following Hypothesis 6, we tested whether nervous-
ness and anticipated backlash would explain the gender
diVerence in the propensity to initiate negotiations with
a male evaluator.
Nervousness. We used the emotions variables to create
a mean composite indicator of nervousness about using
the ask plan (D.87). Feeling nervous about using the
ask plan was negatively correlated with the likelihood of
using the ask vs. no ask plan (rD¡.57, p< .001). As
shown in Table 11, women reported signiWcantly more
nervousness than men about using the ask plan when
facing a male evaluator, t(159)D2.78, p< .01, dD.43.
Nervousness about using the ask plan fully mediated the
eVect of gender of participant on the likelihood of using
the ask vs. no ask plan with a male evaluator, Sobel
zD2.61, p< .01. However, nervousness did not mediate
the interaction of Gender of Candidate £Gender of
Evaluator on the likelihood of using the ask vs. no ask
plan (Sobel zD1.07, pD.28). This means that, while ner-
vousness explained why there was a gender diVerence in
the propensity to initiate negotiations with a male evalu-
ator, it did not explain why the gender diVerence was
greater with a male than with a female evaluator.
Anticipated backlash. We combined the two items about
whether using the ask plan would make the evaluator
not want to hire or work with them into a mean compos-
ite indicator of anticipated backlash (D.86). Antici-
pated backlash from using the ask plan was negatively
correlated with the likelihood of using the ask vs. no ask
plan (rD¡.55, p< .001). The more backlash the partici-
pants anticipated, the more nervous they felt about using
the ask plan, rD.51, p< .001. However, as shown in
Table 11, there was no signiWcant diVerence by gender of
participant in anticipated backlash from using the ask
plan with a male evaluator, t(159) D1.04, pD.30, dD.16.
Anticipated backlash did not mediate the eVect of gen-
der of participant on the likelihood of using the ask plan
with a male evaluator (Sobel zD1.03, pD.30), nor did it
mediate the interaction between gender of participant
and gender of candidate on the likelihood of using the
ask plan (Sobel zD1.77, pD.08).
Gender identity. Finally, we examined whether identiWca-
tion with masculine and feminine traits would mediate
or moderate gender eVects on the propensity to initiate
negotiations. There was no gender diVerence in the par-
ticipants’ identiWcation with masculine personality traits,
female MD4.97, SD D.81, male MD5.05, SD D.84,
t(339) D0.94, pD.35, dD.10. Female (vs. male) partici-
pants identiWed more strongly with feminine personality
traits, female MD5.55, SD D.79, male MD5.25,
SD D.88, t(339) D3.33, p<.001, dD.36. The participants’
masculinity and femininity scores were positively corre-
lated, rD.11, pD.04. IdentiWcation with masculine traits
was negatively correlated with nervousness (rD¡.15,
p< .01), uncorrelated with anticipated backlash
(rD¡.03, pD.55), and positively correlated with use of
the ask vs. no ask plan (rD.14, pD.01). There was no
signiWcant correlation between the participants’ feminin-
ity scores and nervousness, anticipated backlash or use
of the ask vs. no ask plan (rs< .04, ps > .54).
Given the potential for participants to identify
strongly with one, both or neither of the scales (i.e., be
masculine or feminine identiWed, androgynous or
undiVerentiated) (Bem, 1974), we used regression analy-
sis to test whether the interaction of the masculinity and
femininity scores was correlated with nervousness, antic-
ipated backlash or use of the ask vs. no ask plan; we
observed no signiWcant eVects (ps > .12). Because there
was no gender diVerence in the participants’ masculinity
scores and no other correlations with use of the ask vs.
no ask plan, gender identity was not a candidate for
9The excluded cases appeared to be randomly distributed across
conditions. Consistent with the premise that they added noise to the
analyses, exclusion of these cases increased the statistical signiWcance
of the Wndings, but had no other eVect on the overall pattern of results.
T
a
bl
e
9
Experiment 4: ANOVA of likelihood of using ask vs. no ask plan
(ND341)
¤p< .05.
Source df MS F 2
Gender of participant 1 3.46 1.17 .003
Gender of evaluator 1 0.06 0.02 .000
Gender of participant£Gender of evaluator 1 12.23 4.13¤.01
Error 337

98 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
mediation. Gender identity also had no signiWcant mod-
erating eVect on the propensity to initiate negotiations;
we observed no signiWcant interaction eVects between
the BSRI scores and gender of participant (ps>.09) or
gender of evaluator (ps > .20) on the likelihood of using
the ask vs. no ask plan.
Discussion
The female (as compared to male) participants in
Experiment 4 were more reluctant to attempt to negoti-
ate for higher compensation, but only when the evalua-
tor was male. When the evaluator was female, women
were as inclined as men to attempt to negotiate for
higher compensation. Consistent with Hypothesis 5 and
mirroring the results of Experiment 3, the gender of the
evaluator was a signiWcant moderator of the eVect of
gender of participant on the propensity to initiate com-
pensation negotiations.
Mediation analysis showed that women (as compared
to men) were signiWcantly more reticent to initiate nego-
tiations with a male evaluator because the prospect of
doing so made them more signiWcantly more nervous
(Hypothesis 6a). Contrary to our predictions, antici-
pated backlash did not mediate gender diVerences in the
propensity to initiate negotiations with a male evaluator
(Hypothesis 6b). Neither nervousness nor anticipated
backlash explained why the gender diVerence in the pro-
pensity to negotiate was greater with a male than with a
female evaluator. The results of the mediation analyses
suggest that women’s greater hesitation (as compared to
men) about attempting to negotiate for higher compen-
sation may be informed more by emotional intuition
than a conscious cost-beneWt calculus based upon the
anticipated social consequences of initiating negotia-
tions. Future research should take advantage of develop-
ments in the study of emotions in decision making
(Damasio, 1994) and explore the relative inXuence of
emotions and conscious reasoning in the reinforcement
of status-based behavioral norms.
Finally, we tested for possible mediating and moder-
ating eVects of gender identity on the propensity to initi-
ate negotiations. The higher the participants’
masculinity scores, the more likely they were to choose
the negotiation option. However, gender identiWcation
did not explain gender diVerences in the propensity to
initiate negotiations with a male evaluator. These results
further support the proposition that gender eVects on
the propensity to initiate negotiations are motivated by
diVerential treatment of men and women as opposed to
personality diVerences.
General discussion
We posed the question at the beginning of this article of
whether women’s greater reluctance (as compared to men)
to initiate negotiations over resources, such as higher com-
T
a
bl
e
10
Experiment 4: Mean gender diVerences in the likelihood of using ask vs. no ask plan by gender of evaluator
Note: We report standard deviations in parentheses beside the means. DiVerent superscripts (a,b) indicate signiWcant gender diVerence within condi-
tion at the level of p< .05.
Male evaluator Female evaluator
Male participant Female participant Gender diVerence Male participant Female participant Gender diVerence
MM d MM d
2.52a (1.91) 1.94b (1.66) .32 2.17a (1.64) 2.34a (1.64) .10
nn nn
81 80 90 90
T
a
bl
e
11
Experiment 4: Mean gender diVerences in nervousness about using ask plan and anticipated backlash from using ask plan by gender of evaluator
Note: We report standard deviations in parentheses beside the means. DiVerent superscripts (a,b) indicate signiWcant gender diVerence within condi-
tion at the level of p< .05.
Male evaluator Female evaluator
Male participant Female participant Male participant Female participant
MM MM
Nervousness about using ask plan
5.36a(1.27) 5.90b (1.22) 5.58a (1.23) 5.83a (1.18)
Anticipated backlash from using ask plan
4.56a (1.67) 4.83a (1.62) 5.06a (1.37) 4.72a (1.68)
nn nn
81 80 90 90

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 99
pensation, could be explained by the diVerential treatment
of male and female negotiators. The results of these experi-
ments suggest that the answer to this question is yes. In the
Wrst three experiments, male evaluators penalized women
more than men for attempting to negotiate for higher com-
pensation. In Experiment 4, women were more reticent
than men about attempting to negotiate for higher com-
pensation with a male evaluator, and nervousness about
attempting to negotiate explained this gender diVerence.
The results of the mediation analyses in Experiments 2 and
3 were consistent with the proposition that women
encounter resistance when they attempt to negotiate for
higher compensation because such behavior is a status vio-
lation. Men were signiWcantly more inclined to work with
nicer and less demanding women who accepted their com-
pensation oVers without comment than they were with
those who attempted to negotiate for higher compensa-
tion, even though they perceived women who spoke up to
be just as competent as women who demurred.
The behavior of female evaluators diVered across
Experiments 1–3. When evaluating written descriptions or
transcripts of the candidates’ interview responses (Experi-
ments 1 and 2), female evaluators penalized women more
than men for attempting to negotiate for higher compen-
sation. After watching and listening to the candidates
respond to the interview questions on video (Experiment
3), female evaluators were disinclined to work with both
men and women who initiated negotiations. Interestingly,
the behavior of the participants in Experiment 4 mirrored
most closely the results of the video-based experiment;
there was no diVerence in men’s and women’s propensity
to initiate negotiations with a female evaluator.
Previous video-based research on gender and social
inXuence produced a similar pattern of results to those
reported in Experiment 3, with men demanding a higher
degree of likeability from female than from male targets
to be persuaded by them and women perceiving likeabil-
ity to be important to the persuasiveness of both men
and women (Carli et al., 1995). More theory and
research are needed to explore more deeply the role of
the gender of the evaluator in the reinforcement of pre-
scriptive sex stereotypes and to disentangle the inconsis-
tencies in the eVect of the gender of the evaluator
observed across these studies and in the broader litera-
ture on prescriptive sex stereotypes (Heilman & Chen,
2005). The current set of studies suggests that future
research on the eVect of the gender of the evaluator
should explore potential methodological as well as con-
textual moderators (Carli, 1990; Deaux & LaFrance,
1998; Deaux & Major, 1987; Kanter, 1977).
Negotiation and the distribution of organizational
resources
This research has important implications for the distri-
bution of resources and opportunities within organiza-
tions. If men have more freedom to negotiate for
themselves than do women, particularly with senior men,
then that could help to explain phenomena, such as the
gender wage gap and glass ceiling. In most organizations,
those who control organizational resources and opportu-
nities for advancement tend to be men. If women are jus-
tiWably less inclined than men to initiate negotiations
with men, then they may have fewer opportunities to
increase their compensation and promotion potential.
It is not clear from the results of these experiments
that men consciously resist women’s attempts to negoti-
ate. The results of Experiment 4 indicated that women
(as compared to men) were more reticent about negoti-
ating with a male evaluator because the idea of doing so
made them more nervous and not because they antici-
pated more backlash. It warrants further investigation
whether men’s resistance to women who initiate com-
pensation negotiations is also motivated more by a feel-
ing of aversion or discomfort than by a conscious
decision that such behavior by women should be dis-
couraged. Research on the challenges to women of
breaking the glass ceiling, for instance, indicates that
many male CEOs think that women should take more
initiative to signal their interest in critical developmental
experiences (Ragins, Townsend, & Mattis, 1998). Future
research should explore whether raising awareness about
the systemic reasons for gender diVerences in the initia-
tion of negotiations might help to mitigate the social
risks for women. Moreover, both male and female man-
agers should keep in mind that negotiation is a funda-
mental form of social interaction within organizations,
and a potentially important mechanism for the retention
and attraction of talented labor (Rousseau, 2005). More
research is needed to understand better the organiza-
tional implications of inhibiting the initiation of negotia-
tions over issues such as compensation.
The decision to negotiate
Whether our participants’ behavior was optimal, in
terms of weighing the actual social and economic costs
and beneWts of initiating compensation negotiations,
remains an open question. The beneWts of initiating
negotiations in this type of context would obviously
include the expected compensation gains, and the costs
would include the risks of undermining potentially
important working relations and missing out on desir-
able work opportunities. If the expected economic gains
were large enough to outweigh the social costs, then the
rational course of action would be to initiate negotia-
tions, in spite of the social costs. If the social costs and
their long-term career implications outweighed the bene-
Wts of higher compensation, then reticence would be the
more prudent choice. We cannot claim, based on our
research, that either men or women are initiating negoti-
ations too much or too little. We show with this research

100 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
that women’s disinclination relative to men to initiate
negotiations over resources, such as compensation, may
be traced to the higher social costs that they face when
doing so.
It deserves highlighting that, on average, both men
and women in Experiment 4 preferred the no-negotia-
tion to the negotiation plan for responding to a question
about their salary and beneWts oVer. This may be attrib-
utable in part to the artiWciality of choosing between two
pre-determined scripts rather than choosing one’s own
words to negotiate. However, research on the propensity
to initiate negotiations reveals relatively low overall
rates of initiation. For instance, studies of the salary
negotiations of graduating professional school students
suggest that less than a third of students initiate compen-
sation negotiations (32% in study by Babcock & Lasch-
ever, 2003; 21% in study by Gerhart & Rynes, 1991). In
recent laboratory research, only 12% of participants ini-
tiated negotiations when they knew they might earn
more if they asked (Babcock, Laschever, Gelfand, &
Small, 2003). These low baselines of negotiation raise
important questions about the decision to negotiate that
have yet to be addressed in the literature. Future
research should investigate the motivations for this hesi-
tancy, particularly in situations in which the economic
costs to not negotiating are substantial.
Limitations
One limitation of our experimental design was the
artiWciality of the negotiation and no-negotiation
scripts. If the candidates had been able to choose their
own words, it is possible that men and women would
have presented themselves diVerently (Barron, 2003).
We weighed this limitation, however, against the bene-
Wts of enabling us to test the eVects of gender of partic-
ipant, gender of evaluator and gender identiWcation on
perceptions of a speciWc set of behavioral choices and
to test in Experiments 2 and 3 how evaluators would
perceive men and women enacting the precise behav-
iors that the participants assessed in Experiment 4.
Future research should explore whether men and
women initiate negotiations in diVerent ways and
whether variation in accounts (Scott & Lyman, 1968)
or self-presentation style (Carli, 1990; Carli et al., 1995)
would moderate the evaluation of their negotiation
attempts.
Contributions
Our Wndings reinforce the importance for negotiation
scholars and practitioners of considering the social as
well as economic outcomes of negotiation (Curhan,
Elfenbein, & Xu, 2006; Morris, Larrick, & Su, 1999).
This work adds to this understudied area of research in
negotiation by providing another demonstration that
the social costs of engaging in certain negotiating behav-
iors may not outweigh the economic beneWts (Morris
et al., 1999). When focusing primarily on economic out-
comes of negotiation, we fail to appreciate fully the costs
and beneWts of negotiation processes and their products.
The current research also contributes to the growing
body of literature on gender in negotiation in at least
four respects. First, whereas previous research has
focused on internal motivations for gender diVerences in
the propensity to initiate negotiations (Babcock et al.,
2006), this set of studies demonstrates clearly that men
and women face diVerent social incentives when deciding
whether to initiate negotiations over issues such as com-
pensation. This advancement is important because it
should shift the discussion of prescriptive implications
away from Wxing the women to addressing the social
conditions that motivate these gender diVerences (Wat-
son, 1994b). Second, by demonstrating that there are
contextual explanations for gender diVerences in negoti-
ation behavior, we contribute empirical research to a
long line of theoretical work that has criticized the sex-
diVerence approach to the study of gender in negotiation
(Gray, 1994; Kolb & Putnam, 1997; Kolb & Williams,
2000; Wade, 2001; Watson, 1994a, 1994b).
Third, this research adds to the recent body of work
on sex stereotypes in negotiation by illuminating the
inXuence of prescriptive, as opposed to descriptive, sex
stereotypes (Burgess & Borgida, 1999)—or, in terms of
social role theory, stereotypes about gender roles as
opposed to sex-typed skills (Eagly, 1987). Whereas previ-
ous research has demonstrated various ways in which
gender-based performance expectations shape negotia-
tion outcomes (Kray et al., 2001, 2002, 2004), this
research shows that gender-based norms of appropriate
behavior may inXuence whether individuals decide to
negotiate and the social outcomes of their negotiations.
Future research should continue this exploration of how
gendered expectations of appropriate negotiating behav-
ior may inXuence negotiation performance (Wade,
2001).
Fourth, this work contributes to the documentation
of situational moderators of gender eVects in negotia-
tion. Contrary to the proposition that women are always
more reluctant than men to negotiate, we found that
women were only more reluctant than men to attempt to
negotiate in the situation in which the relative social risk
was greatest (i.e., with a male evaluator). Future research
should draw motivation from developments in the study
of situational moderators of gender eVects in negotiation
(Bowles et al., 2005; Kray & Thompson, 2005; Walters
et al., 1998) and continue to explore the boundaries of
gender diVerences in the propensity to initiate negotia-
tions. For instance, the activation of explicit sex stereo-
types favoring men or of implicit sex stereotypes
favoring women might prompt women to initiate negoti-
ations more often than men (Kray et al., 2001, 2002).

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 101
To the extent that men and women diVer in relational
orientation (Cross & Madson, 1997; Gabriel & Gardner,
1999), manipulation of the current or potential future
relationship between negotiating parties might moderate
gender eVects in the propensity to initiate negotiations
(Gelfand, Major, Raver, Nishii, & O’Brien, 2006). Gen-
der diVerences in the propensity to initiate negotiations
might also be inXuenced by the prospective negotiator’s
representation role (i.e., for self vs. other). Previous
research suggests that women are more motivated in
compensation negotiations when they are representing
someone else as opposed to themselves (Bowles et al.,
2005).
Finally, these Wndings have important implications
for the teaching and practice of negotiation, because
they show that one-size-Wts-all prescriptions may not
turn out to be “best practice” for both male and female
negotiators. This research suggests that gender diVer-
ences in the initiation of negotiations cannot be resolved
simply by encouraging women to speak up more.
Addressing this issue requires an understanding of the
situational circumstances that motivate gender diVer-
ences in the propensity to initiate negotiations and a set
of prescriptions that alter the behavior of evaluators as
well as negotiators.
Appendix A. Question 3
By now you must have received the salary and bene-
Wts oVer from the company. Is there anything else you
would like us to keep in mind as we consider your man-
agement placement?
Responses by condition
No ask
Yes, I received the salary and beneWts package. The
beneWts information was very clear. Geographically, I
am totally unconstrained. I am happy to work anywhere,
as long as I have got interesting stuV to do.
Moderate ask
Yes, I received the salary and beneWts package. The
beneWts information was very clear. Geographically, I
am totally unconstrained. I am happy to work anywhere,
as long as I have got interesting stuV to do. What was
not clear to me, however, was whether that salary repre-
sented the top of the pay range. I understand that there
is a range in terms of how much junior managers are
paid in their Wrst placement. I would like to be paid at
the top of that range. I would also like to be eligible for
an end-of-year performance bonus.
Strong ask
Yes, I received the salary and beneWts package. The
beneWts information was very clear. Geographically, I
am totally unconstrained. I am happy to work anywhere,
as long as I have got interesting stuV to do. What was
not clear to me, however, was whether that salary repre-
sented the top of the pay range. I understand that there
is a range in terms of how much junior managers are
paid in their Wrst placement. I think I should be paid at
the top of that range. This is really important to me; I
think I deserve it. I also would like to be eligible for an
end-of-year bonus. I know performance bonuses are not
standard for junior managers, but I would certainly be
more motivated if I could look forward to a perfor-
mance bonus at the end of the year. I am thinking of
something in the 25–50% of salary range. Not doubling
my salary or anything. And, listen, I do not care if it’s in
cash or stocks—and I promise you I’ll earn it. So, those
are the two things that I am asking with regard to my
compensation: one, paying me at the top of the junior
manager salary range and, two, providing me with an
end of year, 20–50% of salary performance bonus.
References
Aroian, L. A. (1947). The probability function of the product of two
normally distributed variables. Annals of Mathematical Statistics,
18, 265–271.
Babcock, L., Gelfand, M., Small, D., & Stayn, H. (2006). Gender diVer-
ences in the propensity to initiate negotiations. In D. D. Crèmer, M.
Zeelenberg, & J. K. Murnighan (Eds.), Social psychology and eco-
nomics (pp. 239–259). Mahwah, NJ: Lawrence Erlbaum.
Babcock, L., & Laschever, S. (2003). Women don’t ask. Princeton, NJ:
Princeton University Press.
Babcock, L., Laschever, S., Gelfand, M., & Small, D. (2003). Nice girls
don’t ask. Harvard Business Review, 81, 14–15.
Barron, L. (2003). Ask and you shall receive? Gender diVerences in beliefs
about requests for a higher salary. Human Relations, 56, 635–663.
Bem, S. L. (1974). The measurement of psychological androgyny. Jour-
nal of Consulting and Clinical Psychology, 42, 155–162.
Bem, S. L. (1981). Bem sex-role inventory. San Francisco, CA: Consult-
ing Psychology Press.
Berger, J., Fisek, M. H., Norman, R. Z., & Zelditch, M. E. (1977). Status
characteristics and social interaction: An expectation-states
approach. New York: Elsevier.
Berger, J., Webster, M., Jr., Ridgeway, C. L., & Rosenholtz, S. J. (1986).
Status cues, expectations and behaviors. In E. Lawler (Ed.), Advances
in group processes (Vol. 3, pp. 1–22). Greenwich, CT: JAI Press.
Bledsoe, A. T. (1856). An essay on liberty and slavery. Philadelphia: J.B.
Lippincott.
Bowles, H. R., Babcock, L., & McGinn, K. L. (2005). Constraints and
triggers: Situational mechanics of gender in negotiation. Journal of
Personality and Social Psychology, 89, 951–965.
Brett, J. M., & Stroh, L. K. (1997). Jumping ship: Who beneWts from an
external labor market career strategy? Journal of Applied Psychol-
ogy, 82, 331–341.
Burgess, D., & Borgida, E. (1999). Who women are, who women should
be: Descriptive and prescriptive gender stereotyping in sex discrimi-
nation. Psychology, Public Policy, and Law, 5, 665–692.

102 H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103
Butler, D., & Geis, F. L. (1990). Nonverbal aVect responses to male and
female leaders: Implications for leadership evaluations. Journal of
Personality and Social Psychology, 58, 48–59.
Carli, L. L. (1990). Gender, language, and inXuence. Journal of Person-
ality and Social Psychology, 59, 941–951.
Carli, L. L., LaFleur, S. J., & Loeber, C. C. (1995). Nonverbal behavior,
gender, and inXuence. Journal of Personality and Social Psychology,
68, 1030–1041.
Cejka, M. A., & Eagly, A. H. (1999). Gender-stereotypic images of
occupations correspond to the sex segregation of employment. Per-
sonality and Social Psychology Bulletin, 25, 413–423.
Conway, M., Pizzamiglio, M. T., & Mount, L. (1996). Status, commu-
nality, and agency: Implications for stereotypes of gender and other
groups. Journal of Personality and Social Psychology, 71, 25–38.
Cross, S. E., & Madson, L. (1997). Models of the self: Self-construals
and gender. Psychological Bulletin, 122, 5–37.
Curhan, J. R., Elfenbein, H. A., & Xu, A. (2006). What do people care
about when they negotiate? Mapping the domain of subjective
value in negotiation. Journal of Personality and Social Psychology,
91, 493–512.
Damasio, A. R. (1994). Descartes’ error: Emotion, reason, and the
human brain. New York: G.P. Putnam.
Daubman, K. A., Heatherington, L., & Ahn, A. (1992). Gender and the
self-presentation of academic achievement. Sex Roles, 27(3–4),
187–204.
Deaux, K., & LaFrance, M. (1998). Gender. In D. T. Gilbert, S. Fiske,
& G. Lindsey (Eds.), The handbook of social psychology (4th ed., pp.
788–827). New York: Wiley.
Deaux, K., & Major, B. (1987). Putting gender into context: An interac-
tive model of gender-related behavior. Psychological Review, 94,
369–389.
Diekman, A. B., & Eagly, A. H. (2000). Stereotypes as dynamic con-
structs: Women and men of the past, present, and future. Personal-
ity and Social Psychology Bulletin, 26, 1171.
Dovidio, J. F., Brown, C. E., Heltman, K., & Ellyson, S. L. (1988).
Power displays between women and men in discussions of gender-
linked tasks: A multichannel study. Journal of Personality and
Social Psychology, 55, 580–587.
Eagly, A. H. (1987). Sex diVerences in social behavior: A social-role
interpretation. Hillsdale, NJ: Erlbaum.
Eagly, A. H., & Johnson, B. T. (1990). Gender and leadership style: A
meta-analysis. Psychological Bulletin, 108, 233–256.
Eagly, A. H., Johannesen-Schmidt, M. C., & van Engen, M. L. (2003).
Transformational, transactional and laissez-faire leadership styles:
A meta-analysis comparing men and women. Psychological Bulle-
tin, 129, 569–591.
Eagly, A. H., Karau, S. J., & Makhijani, M. G. (1995). Gender and the
eVectiveness of leaders: A meta-analysis. Psychological Bulletin,
117, 125–145.
Eagly, A. H., Makhijani, M. G., & Klonsky, B. G. (1992). Gender and
the evaluation of leaders: A meta-analysis. Psychological Bulletin,
111, 3–22.
Eagly, A. H., & SteVen, V. J. (1984). Gender stereotypes stem from the
distribution of women and men into social roles. Journal of Person-
ality and Social Psychology, 46, 735–754.
Ely, R. J., & Meyerson, D. E. (2000). Theories of gender in organiza-
tions: A new approach to organizational analysis and change.
Research in Organizational Behavior, 22, 103–151.
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of
(often mixed) stereotype content: Competence and warmth respec-
tively follow from perceived status and competition. Journal of Per-
sonality and Social Psychology, 82, 878–902.
Gabriel, S., & Gardner, W. L. (1999). Are there ’his’ and ’hers’ types
of interdependence? The implications of gender diVerences in
collective versus relational interdependence for aVect, behavior,
and cognition. Journal of Personality and Social Psychology, 77,
642–655.
Gelfand, M., Major, V. S., Raver, J., Nishii, L. H., & O’Brien, K. (2006).
Negotiating rationally: The dynamics of the relational self in nego-
tiations. Academy of Management Review, 31, 427–451.
Gerhart, B., & Rynes, S. (1991). Determinants and consequences of sal-
ary negotiations by male and female MBA graduates. Journal of
Applied Psychology, 76, 256–262.
Glick, P., Fiske, S. T., Mladinic, A., Saiz, J. L., Abrams, D., Masser, B.,
et al. (2000). Beyond prejudice as simple antipathy: Hostile and
benevolent sexism across cultures. Journal of Personality and Social
Psychology, 79, 763–775.
Goodman, L. A. (1960). On the exact variance of products. Journal of
the American Statistical Association, 55, 708–713.
Gould, R. J., & Slone, C. G. (1982). The “feminine modesty” eVect: A
self-presentational interpretation of sex diVerences in causal attri-
bution. Personality and Social Psychology Bulletin, 8, 477–485.
Gray, B. (1994). The gender-based foundations of negotiation theory.
In R. Lewicki, B. Sheppard, & M. Bazerman (Eds.), Research on
negotiations in organizations (Vol. 4, pp. 3–36).
Hall, J. A., & Halberstadt, A. G. (1986). Smiling and gazing. In J. S.
Hyde & M. C. Inn (Eds.), The psychology of gender: Advances
through meta-analysis (pp. 136–158). Baltimore, MD: John Hopkins
University Press.
Heatherington, L., Daubman, K. A., Bates, C., & Ahn, A. (1993). Two
investigations of “female modesty” in achievement situations. Sex
Roles, 29, 739–754.
Heilman, M. E., & Chen, J. J. (2005). Same behavior, diVerent conse-
quences: Reactions to men’s and women’s altruistic citizenship
behavior. Journal of Applied Psychology, 90, 431–441.
Heilman, M. E., Wallen, A. S., Fuchs, D., & Tamkins, M. M. (2004).
Penalties for success: Reactions to women who succeed at male
gender-typed tasks. Journal of Applied Psychology, 89, 416–427.
HoVman, C., & Hurst, N. (1990). Gender stereotypes: Perception or
rationalization? Journal of Personality and Social Psychology, 58,
197–208.
Jackman, M. R. (1994). The velvet glove: Paternalism and conXict in
gender, class, and race relations. Berkeley, CA: University of Cali-
fornia Press.
Jost, J. T., & Kay, A. C. (2005). Exposure to benevolent sexism and
complementary gender stereotypes: Consequences for speciWc and
diVuse forms of system justiWcation. Journal of Personality and
Social Psychology, 88, 498–509.
Kanter, R. M. (1977). Men and women of the corporation. New York:
Basic Books.
Kolb, D. M., & Putnam, L. L. (1997). Through the looking glass: negotia-
tion theory refracted through the lens of gender. In S. Gleason (Ed.),
Frontiers in dispute resolution in industrial relations and human resources
(pp. 231–257). East Lansing, MI: Michigan State University Press.
Kolb, D. M., & Williams, J. (2000). The shadow negotiation: How
women can master the hidden agendas that determine bargaining suc-
cess. New York: Simon & Schuster.
Kray, L. J., Galinsky, A., & Thompson, L. (2002). Reversing the gender
gap in negotiations: An exploration of stereotype regeneration. Orga-
nizational Behavior and Human Decision Processes, 87, 386–409.
Kray, L. J., Reb, J., Galinsky, A. D., & Thompson, L. (2004). Stereotype
reactance at the bargaining table: The eVect of stereotype activa-
tion and power on claiming and creating value. Personality and
Social Psychology Bulletin, 30, 399–411.
Kray, L. J., & Thompson, L. (2005). Gender stereotypes and negotia-
tion performance: A review of theory and research. In B. M. Staw &
R. Kramer (Eds.), Research in organizational behavior series (Vol.
26, pp. 103–182). Greenwich, CT: JAI Press.
Kray, L. J., Thompson, L., & Galinsky, A. (2001). Battle of the sexes:
Gender stereotype conWrmation and reactance in negotiations.
Journal of Personality and Social Psychology, 80, 942–958.
LaFrance, M., Hecht, M. A., & Paluck, E. L. (2003). The contingent
smile: A meta-analysis of sex diVerence in smiling. Psychological
Bulletin, 129, 305–334.

H.R. Bowles et al. / Organizational Behavior and Human Decision Processes 103 (2007) 84–103 103
Lauterbach, K. E., & Weiner, B. J. (1996). Dynamics of upward inXu-
ence. Leadership Quarterly, 7, 87–107.
Meeker, B. F., & Weitzel-O’Neill, P. A. (1977). Sex roles and interper-
sonal behavior in task-oriented groups. American Sociological
Review, 42, 91–105.
Morris, M. W., Larrick, R. P., & Su, S. K. (1999). Misperceiving negoti-
ation counterparts: When situationally determined bargaining
behaviors are attributed to personality traits. Journal of Personality
and Social Psychology, 77, 52–67.
Myrdal, G. (1994). An American dilemma: The Negro problem and mod-
ern democracy. New York: Harper & Brothers.
Pinkley, R. L., & Northcraft, G. B. (2000). Get paid what you’re worth:
The expert negotiator’s guide to salary and compensation. New
York: St. Martin’s Press.
Prentice, D. A., & Carranza, E. (2002). What women and men should
be, shouldn’t be, are allowed to be, and don’t have to be: The con-
tents of prescriptive gender stereotypes. Psychology of Women
Quarterly, 26, 269–281.
Ragins, B. R., Townsend, B., & Mattis, M. (1998). Gender gap in the
executive suite: CEOs and female executives report on breaking the
glass ceiling. The Academy of Management Executive, 12, 28–42.
Ridgeway, C. L. (1982). Status in groups: The importance of motiva-
tion. American Sociological Review, 47, 76–88.
Ridgeway, C. L. (2001a). Gender, status, and leadership. Journal of
Social Issues, 57, 637–655.
Ridgeway, C. L. (2001b). How do status beliefs develop? The role of
resources and interactional experiences. In J. Jost & B. Major
(Eds.), The psychology of legitimacy: Emerging perspectives on ideol-
ogy, justice and intergroup relations (pp. 357–377). Cambridge,
England: Cambridge University Press.
Ridgeway, C. L., & Berger, J. (1986). Expectations, legitimation, and
dominance behavior in task groups. American Sociological Review,
51, 603–617.
Ridgeway, C. L., & Bourg, C. (2004). Gender as status: An expectation
states theory approach. In A. H. Eagly, A. E. Beall, & R. J. Stern-
berg (Eds.), The psychology of gender (pp. 217–241). New York:
Guliford Press.
Rousseau, D. (2005). I-deals, idiosyncratic deals employees bargain for
themselves. Armonk, NY: M.E. Sharpe.
Rudman, L. A. (1998). Self-promotion as a risk factor for women: The
costs and beneWts of counterstereotypical impression management.
Journal of Personality and Social Psychology, 74, 629–645.
Rudman, L. A., & Glick, P. (1999). Feminized management and back-
lash toward agentic women: The hidden costs to women of a kin-
der, gentler image of middle managers. Journal of Personality and
Social Psychology, 77, 1004–1010.
Rudman, L. A., & Glick, P. (2001). Prescriptive gender stereotypes
and backlash toward agentic women. Journal of Social Issues,
57, 743–762.
Rudman, L. A., & Kilianski, S. E. (2000). Implicit and explicit attitudes
toward female authority. Personality and Social Psychology Bulle-
tin, 26, 1315–1328.
Schein, V. E. (2001). A global look at psychological barriers to
women’s progress in management. Journal of Social Issues, 57, 675–
688.
Scott, M. B., & Lyman, S. M. (1968). Accounts. American Sociological
Review, 33, 46–62.
Short, J., Williams, E., & Christie, B. (1976). The social psychology of
telecommunications. New York: John Wiley & Sons.
Sidanius, J., Pratto, F., & Bobo, L. (1994). Social dominance orienta-
tion and the political psychology of gender: A case of invariance?
Journal of Personality and Social Psychology, 67, 998–1011.
Sobel, M. E. (1982). Asymptotic intervals for indirect eVects in struc-
tural equations models. In S. Leinhart (Ed.), Sociological methodol-
ogy (pp. 290–312). San Francisco: Jossey-Bass.
Spence, J. T., & Buckner, C. E. (2000). Instrumental and expressive
traits, trait stereotypes, and sexist attitudes. Psychology of Women
Quarterly, 24, 44–62.
Spence, J. T., & Helmreich, R. L. (1978). Masculinity and femininity:
Their psychological dimensions, correlates, and antecedents. Austin:
University of Texas Press.
Stevens, C. K., Bavetta, A. G., & Gist, M. E. (1993). Gender diVerences
in the acquisition of salary negotiation skills: The role of goals, self-
eYcacy, and perceived control. Journal of Applied Psychology, 78,
723–735.
Stuhlmacher, A. F., & Walters, A. E. (1999). Gender diVerences in
negotiation outcome: A meta-analysis. Personnel Psychology, 52,
653–677.
Thompson, L. (2005). The mind and heart of the negotiator (3rd ed.).
Upper Saddle River, NJ: Prentice Hall.
Twenge, J. M. (1997). Changes in masculine and feminine traits over
time: A meta-analysis. Sex Roles, 36, 305–325.
Wade, M. E. (2001). Women and salary negotiation: The costs of self-
advocacy. Psychology of Women Quarterly, 25, 65–76.
Walters, A. E., Stuhlmacher, A. F., & Meyer, L. L. (1998). Gender and
negotiator competitiveness: A meta-analysis. Organizational Behav-
ior and Human Decision Processes, 76, 1–29.
Walther, J. B., & Parks, M. R. (2002). Cues Wltered out, cues Wltered in:
Computer-mediated communication and relationships. In M. L.
Knapp & J. A. Daly (Eds.), Handbook of interpersonal communica-
tion (3rd ed., pp. 529–563). Thousand Oaks, CA: Sage.
Watson, C. (1994a). Gender diVerences in negotiating behavior and
outcomes: Fact or artifact? In A. Taylor & J. Beinstein-Miller
(Eds.), ConXict and gender. Cresskill, NJ: Hampton Press, Inc.
Watson, C. (1994b). Gender versus power as a predictor of negotiation
behavior and outcomes. Negotiation Journal(April), 117–127.
Weber, M. (1968). Economy and society (E. FrischoV, Trans.). New
York: Bedminister Press.