British Journal of Social Psychology (2014), 53, 217–234
© 2013 The British Psychological Society
If you’re going to be a leader, at least act like it!
Prejudice towards women who are tentative in
Renata Bongiorno1,2*, Paul G. Bain1and Barbara David2
1University of Queensland, Brisbane, Australia
2Australian National University, Canberra, Australia
Role congruity theory predicts prejudice towards women who meet the agentic
requirements oftheleader role.Inline with recentfindings indicating greater acceptance
of agentic behaviour from women, we find evidence for a more subtle form of prejudice
towards women who fail to display agency in leader roles. Using a classic methodology,
presented through written (Study 1, N = 167) or verbal (Study 2, N = 66) communi-
cations. Consistent with predictions, assertive women were as likeable and influential as
assertive men, while being tentative in leadership reduced the likeability and influence of
women, but not of men. Although approval of agentic behaviour from women in
if they fail to show agency is important for understanding how they continue to be at a
distinct disadvantage to men in leader roles.
Despite important advances, the social positions of women and men in contemporary
western societies remain markedly different and fundamentally unequal. Women
continue to carry the burden of responsibility for unpaid domestic labour, including
and low status such as nursing, teaching, and retail (Connell, 2002; Kreimer, 2004).
Women remain conspicuously underrepresented in positions of power and status, with
men still holding the overwhelming majority of senior executive positions and seats in
national parliaments (United Nations Development Programme, 2007).
The persistence of gender inequality in western societies is particularly striking
because endorsement of traditional gender roles has substantially declined (Spence &
Hahn, 1997; Twenge, 1997). However, as western societies have liberalized, barriers to
women’s progression have also been changing (Barreto, Ryan, & Schmitt, 2009; Swim,
Australia (e-mail: firstname.lastname@example.org).
by role congruity theory (Eagly & Karau, 2002), whereby women who meet the agentic
requirementsoftheleaderroleareevaluated lessfavourablythan equivalentmen.Below,
we describe evidence for this traditional form of prejudice, and argue that greater
encouragement of agentic behaviour from women has given rise to a more subtle form of
prejudice, whereby women who fail to meet the agentic requirements of the leader role
are instead singled out for disapproval.
Gender stereotypes and prejudice towards agentic women
Animportant aspectofgenderstereotypes isthegreater attributionofagenticqualities to
men, reflecting their advantaged or high-status position relative to women in society
(Eagly, 1987; Eagly & Karau, 2002; Heilman, 2001). Women are still seen as relatively
lacking in the male-stereotypic agentic attributes associated with, and required of, high-
status positions, such as strength, dominance, assertiveness, and decisiveness (Diekman
& Eagly, 2000; Latu et al., 2011; Powell, Butterfield, & Parent, 2002; Ryan, Haslam,
of prejudice toward female leaders, this stereotype of women’s lesser agency is both
descriptive and prescriptive of their behaviour, generating two distinct forms of
prejudice. The descriptive component (that women are less agentic) is demonstrated in
prejudiced perceptions of women’s potential for leadership (Biernat & Kobrynowicz,
standard, whereby women must clearly outperform men to be selected as leaders (Foddy
& Smithson, 1999).
The focus of the current studies, however, is on prejudice based on the prescriptive
component of this stereotype (that women should be less agentic). Here, role congruity
theory predicts that women who behave agentically once in a leader role remain at a
distinct disadvantage due to: ‘…less approval of agentic behavior enacted by a woman
compared with a man’ (Eagly & Karau, 2002, pp. 583–584). That is, prejudice based on
this prescriptive stereotype is demonstrated in less favourable evaluations of women’s
indicates status and power, it is incongruent with women’s subordinate status, and
violations of this prescriptive gender stereotype are typically punished through negative
social appraisals (for related theorizing, see Carli, 1999). This form of prejudice places
female leaders in a bind, as the agentic behaviour required of leaders (e.g., assertive,
dominant) is incompatible with the non-agentic behaviour required of women (e.g.,
As supporting evidence of prejudice based on this prescriptive gender stereotype,
many researchers (e.g., Bowles, Babcock, & Lai, 2007; Eagly & Karau, 2002; Heilman,
Wallen, Fuchs, & Tamkins, 2004; Rudman & Glick, 1999) refer to Carli’s (1990, Study 2)
experiment, which showed that male participants favoured non-agentic to agentic
behaviour from women. Carli manipulated perceptions of agency through assertive or
tentative speech, and found that male participants liked, and were more influenced by,
women who spoke tentatively – behaviour consistent with the stereotype of their lesser
agency and lower status. Indeed, women who spoke assertively achieved less influence
and were liked less by men. However, male speakers’ likeability and influence was not
affected by whether they used assertive or tentative speech, a finding attributed to men’s
218 Renata Bongiorno et al.
Althoughthe descriptive stereotypeof women’s lesser agencylargely persists,the extent
of the difference has reduced over time. Thus, compared to the past, women are now
more likely to describe themselves as possessing agentic attributes (Spence & Buckner,
and are projected to have even more agentic attributes in the future (Diekman & Eagly,
2000; Morton, Rabinovich, & Postmes, 2011).
We propose that the shift over time in the prescriptive gender stereotype has been
more dramatic, such that there is no longer endorsement of non-agentic behaviour as
prescriptive for women. Indeed, the original evidence for this prediction dates back to
research conducted in the 1970s and 1980s (for a meta-analysis, see Eagly, Makhijani, &
Klonsky, 1992). In these decades, women were rarely seen in positions of public
leadership (Adler, 1999), and attitudes towards women were considerably more
conservative (Carli & Eagly, 2001; Spence & Hahn, 1997; Twenge, 1997).
Research conducted more recently shows that prescriptions against women’s
agentic behaviour have eased. Based on role congruity theory, Diekman (2007)
predicted prejudice towards agentic (dominant) women, but found that agentic
women and men received similar interpersonal evaluations (for similar findings
showing no prejudice based on women’s agentic behaviour, see Heilman & Okimoto,
2007). Greater acceptance of agentic behaviour from women has led some
researchers to suggest that it is now only a particular subset of agentic behaviours
that remain a source of prejudice – specifically those that also lead women to violate
the prescriptive stereotype that women should be communal (e.g., aggressive,
ruthless, see; Rudman & Glick, 1999, 2001). The implication is that the prescriptive
stereotype that women should be less agentic than men is no longer endorsed, so
women are no longer punished for agentic behaviour unless this behaviour also
demonstrates that they are non-communal (e.g., unfriendly, insensitive).
A subtle form of prejudice towards non-agentic women
Greater acceptance of agentic behaviour from women is positive, showing that it has
becomeacceptable for women to behave with the strength and decisiveness expectedof
leaders, and helping to resolve the bind faced by agentic female leaders. However,
assuming this is the case, we propose that there may be a previously unidentified
drawback of greater endorsement of women’s agentic behaviour in leadership –
specifically that woman who fail to display agency in leadership may now be punished, a
consequence that will not apply to men acting in the same ways. That is, while agentic
behaviour from women leaders (e.g., assertive, dominant) is now met with social
approval, social penalties may instead be applied to women who are non-agentic (e.g.,
tentative, submissive) in such roles. That is, we make the opposite prediction from
research conducted at a time when women’s non-agentic behaviour in leader roles was
preferred (e.g., Carli, 1990; Wiley & Eskilson, 1985).
is already some support for this prediction. Although men showed prejudice towards
agentic women in early studies, women reacted more negatively towards non-agentic
little attention in subsequent research). More recently, men have been found to react
more negatively to non-agentic (self-effacing) women when their own outcomes depend
Prejudice towards tentative women219
on the woman’s performance (Rudman, 1998).1Moreover, Davies-Netzley (1998)
interviewed men and women in elite corporate positions and noted that some people
described significant social pressure on women to display agentic traits to fit in and be
taken seriously. For example, a male interviewee reported non-agency in women as
unsuitable: ‘If you’re going to be a frilly little sweetheart and have people open your
doors…that isn’t what it’s about’ (Davies-Netzley, 1998, p. 349, italics added).
Unlike traditional prejudice towards women’s agentic behaviour, prejudice towards
women’s non-agentic behaviour in leadership is subtle, and may prove more difficult to
overcome. This is because it is likely to appear fair and legitimate to evaluate women’s
non-agentic behaviour negatively, without considering that male leaders avoid the same
level of scrutiny (cf. Dovidio & Gaertner, 2000, who highlight a similar form of subtle
(e.g., Carli, 1999; Eagly et al., 1992), the association between being male and being a
leader(Koenig,Eagly,Mitchell,&Ristikari, 2011) canprovidelegitimacyto meninleader
roles, affording them the benefit of the doubt on occasions where their leader behaviour
diverges from the ideal. As reported above (Carli, 1990 Study 2), men’s use of assertive or
tentative speech didnot affecttheirinfluenceorlikeability(forsimilar findings, seeWiley
&Eskilson, 1985). Similarly,Eagly et al.(1992) providedmeta-analytic evidenceshowing
men are granted a wider scope of acceptable leadership styles. Numerous other studies
have shown that prescriptions for men’s behaviour in leadership and other workplace
roles are less restrictive than prescriptions for women’s behaviour (e.g., Carli, LaFleur, &
Loeber, 1995; Foschi, Sigerson, & Lembesis, 1995; Heilman & Chen, 2005; Okimoto &
notmen)may have changedovertime, resulting indiminished prejudice towardsagentic
women leaders,weproposethat women’s agenticbehaviour in leadership willno longer
be evaluated more negatively than that of men. In contrast, our central and novel
prediction is that women who behave in non-agentic ways in leader roles will now be
evaluated more negatively than both agentic women and non-agentic men. As recent
research has linked some forms of agentic behaviour with (non-)communality (e.g.,
Rudman & Glick, 1999, 2001) we will show that this prejudice can be distinguished from
attributions of communality to leaders, and hence represents a form of prejudice that
specifically relates to the agency displayed by women in leader roles.
To test our predictions of prejudice towards women who fail to behave agentically in
leader roles, Study 1 used a method based on Carli’s (1990, Study 2) classic and widely
cited study, which demonstrated men’s disapproval of women’s agentic behaviour. In
students for using college buses), delivered by male or female students in an agentic
(assertive) or non-agentic (tentative) style. In this study, we replicated this speech style
manipulation, but used politicians as targets to emphasize a leadership context, and used
speeches advocating for action on climate change to maximize the relevance of the topic
1Equivalent resultswere notshown for female participants,who dislikedthe agentic(self-promoting) female. However, as Eagly
negatively to women’s agentic behaviour.
220Renata Bongiorno et al.
to contemporary political contexts. The assertive version of the transcript was written to
indicate dominance, confidence, and strength – agentic attributes that are more
stereotypic of men and those with status. The tentative version was written to indicate
deference, hesitancy, and a lack of confidence – non-agentic attributes that are more
stereotypic of women and those with low status (Lakoff, 1976; for a meta-analysis, see
Leaper & Robnett, 2011).
The study had a 2 (speech style: assertive, tentative) 92 (leadergender: male, female)
9 2 (participant gender: male, female) between-participants design. Prejudice was
measured through two key leader evaluations. The first was likeability, to capture the
sense in which the leader was considered easy to connect with and similar to the self in
important ways, with lower likeability indicating an increased likelihood that the leader
wouldbesociallyexcluded. The second was influence, to capture leader persuasiveness,
with lower influence indicating a diminished capacity of the leader to convince others of
Contrary to role congruity theory predictions that women’s agentic behaviour in the
leader role elicits disapproval (Eagly & Karau, 2002), and Carli’s (1990) findings for male
participants, prejudice towards tentative female leaders was expected. Thus, it was
predicted that there would be a speech style 9 leader gender interaction, with tentative
female leaders regarded as less likeable and achieving less influence than assertive female
leaders and tentative male leaders. Assertive female leaders were predicted to achieve
similar levels of likeability and influence to assertive male leaders, and speech style was
not expected to impact upon the likeability or influence of male leaders.
One hundred and eighty-five participants (47% female: Mage= 28.3 years, SD = 12.49)
completed the study. Participants were approached on an Australian University campus
or via email if they had joined an online research database available to students and
graduates of the University. Participants approached on campus received five Australian
dollars, or those enrolled in psychology could opt for psychology research participation
credit. Participants from the online research database received credit points (worth
approximately 5 Australian dollars) towards a shopping voucher.
Materials and procedure
Participants were informed that the study was about effective communication, and were
randomly allocated to experimental conditions. All participants read a transcript of a
(David/Andrew Hayes). The content of the speech was at the strong end of proposals to
address climate change, to ensure room for influence even among those who already
supported climate change action. The transcript included statements about the threat
posed by climate change, the importance of developed nations acting on climate change,
and advocated the implementation of strong climate change policies, including deep,
reading the transcript, participants completed a series of measures, and were debriefed.
The study took most participants 15 min.
Prejudice towards tentative women221
The assertive version of the transcript contained italicized words to indicate strong
emphasis, and contained no indication of hesitation or uncertainty (e.g., ‘Scientists
believe a total increase of 2 degrees significantly increases the chance of dangerous
climate change…If the warming continues, sea levels could rise by 6 m, devastating
coastal areas and creating millions of refugees’). In the tentative version, no words were
italicized, and numerous hedges (e.g., ‘probably’, ‘you know’), qualifiers (e.g., ‘I think
now believe that a total increase of ah 2 degrees significantly increases the chance of you
know dangerous climate change…If the warming continues, I think that sea levels could
rise by 6 m, um devastating coastal areas and creating millions of refugees’). The
transcripts were identical in every other respect.
In addition to demographic questions, participants responded to the following key
This was measured using five items, including how likeable the leader was (adapted from
Carli, 1990), whetherparticipants could see themselves getting along with the leader in a
social setting (adapted from Diekman, 2007), and how similar they felt to the leader (e.g.,
similar to this speaker’; 1 = not at all, 9 = very much; a = .91).
Following Carli (1990), the influence measure captured participants’ level of agreement
with the position proposed by the leader, using a single item (‘I think this country should
commit to reducing its carbon emissions by 50% by 2020’; 1 = completely disagree,
9 = completely agree).
Agency and communality
Six items adapted from Bem (1974) and Eagly and Karau (2002) assessed the agency and
communality of leaders (1 = not at all, 9 = very much). Three items assessed agency
(dominant, forceful, confident; a = .89) for use as a manipulation check. Three items
assessed communality (friendly, sensitive, warm; a = .85) to check that agentic female
leaders did not violate prescriptions for women’s communality (cf. Rudman & Glick,
1999, 2001), and to include as a covariate to show that results for agency were
independent of perceived leader communality.
Comprehension and suspicion checks
Participants indicated whether the politician believed this country should support a 50%
reduction in emissions among developed countries by 2020 by circling ‘yes’ or ‘no’.
They also indicated whether the politician was a man or a woman. To test for suspicion,
at the end of the questionnaire participants were also asked to indicate whether the
politician described in the transcript was real.
222 Renata Bongiorno et al.
All participants correctly identified the gender of the politician and passed the
comprehension check. Eighteen participants failed the suspicion check by indicating
they knew the politician was not real and were excluded, leaving a final sample of 167
(50% female, Mage= 27.58 years, SD = 12.19). A 4 (leader name) 9 2 (speech style)
analysis of variance, conducted separately for agency and communality, revealed no
effects involving leader name in either analysis (all ps > .22), so leader name was not
included in subsequent analyses.
effective, F(1, 159) = 275.39, p < .001, gp2= .63, with assertive leaders (M = 6.46,
SD = 1.16) perceived as considerably more agentic than tentative leaders (M = 3.39,
SD = 1.32).Qualifyingthismaineffectwasasignificantinteractionbetweenspeechstyle
and participant gender, F(1, 159) = 6.05, p = .015, gp2= .04. Follow-up comparisons
revealed that male participants rated assertive speakers (M = 6.33, SD = 1.07) as consid-
erably more agentic than tentative speakers (M = 3.72, SD = 1.45), F(1, 81) = 88.48,
p < .001,however,thisdifferencewasgreaterforfemaleparticipants(assertivespeakers:
M = 6.57, SD = 1.24; tentative speakers: M = 3.03, SD = 1.06), F(1, 82) = 192.62,
p < .001. No other effects were significant.
leaders (M = 5.27, SD = 1.32) to be seen as more communal than tentative leaders
(M = 4.86, SD = 1.33), F(1, 159) = 3.89, p = .05, gp2= .02. No other effects were
significant. Thus, agentic female leaders were not perceived as non-communal.
solely to agency (for correlations, see Appendix).
Tests of hypotheses
of covariance were used for likeability and influence, with communality as a covariate.
The predicted interaction between speech style and leader gender was significant,2F(1,
158) = 3.98, p = .048, gp2= .03, qualifying a main effect for speech style, F(1,
158) = 11.21, p = .001, gp2= .07, and leader gender, F(1, 158) = 6.19, p = .014,
gp2= .04. The means are shown in Figure 1. There was also a significant main effect for
participant gender, F(1, 158) = 11.74, p = .001, gp2= .07, with likeability rated higher
by male than by female participants (M = 5.38, SE = .13; M = 4.75, SE = .13, respec-
tively). The three-way interaction was not significant (p = .224), indicating that
participant gender did not moderate the main findings. No other effects were significant.
To interpret the speech style 9 leader gender interaction, comparisons were made
within leader gender and within speech style, controlling for communality. Comparing
within leader gender, assertive female leaders were more likeable than tentative female
2The interaction remained significant, p = .018 after the communality covariate was omitted.
Prejudice towards tentative women223
leaders, F(1, 77) = 10.42, p = .002, gp2= .12, but there was no significant difference in
the likeability of assertive and tentative male leaders, F(1, 84) = 0.86, p = .357.
Comparing within speech style, tentative male leaders were more likeable than tentative
female leaders, F(1, 74) = 8.30, p = .005, gp2= .10, but there was no significant
differenceinthelikeabilityofassertivemaleandassertivefemaleleaders,F(1,87) = 0.22,
p = .637.
The predicted interaction between speech style and leader gender was significant,3F(1,
158) = 6.49, p = .012, gp2= .04. No other effects were significant, including the three-
way interaction (p = .970). To interpret the significant interaction, comparisons were
made within leader gender and within speech style, controlling for communality. These
supported predictions, with means shown in Figure 1. Comparing within leader gender,
F(1, 77) = 6.66, p = .012, gp2= .01, but there was no significant difference in influence
between assertive and tentative male leaders, F(1, 84) = 0.49, p = .486. Comparing
within speech style, participants were significantly more influenced by tentative male
than tentative female leaders, F(1, 74) = 7.66, p = .007, gp2= .10, while assertive male
and assertive female leaders did not significantly differ in influence, F(1, 87) = 0.39,
p = .534.
Study 1 found clear support for the prediction that prejudice is now directed towards
women who fail to behave agentically in leader roles. Social penalties were directed at
tentative female leaders, who were shown to be less likeable and less influential than
assertive female leaders and tentative male leaders. Assertive female leaders were as
likeable and influential as assertive male leaders, and the likeability and influence of male
leaders was unaffected by the assertiveness of their speech.
Figure 1. Mean scores (+SE) for Study 1 for likeability and influence as a function of leader gender and
3The interaction remained significant, p = .007 after the communality covariate was omitted.
224Renata Bongiorno et al.
This subtle form of prejudice towards women who fail to behave agentically in leader
indicate that female politicians will only be as effective as male politicians where they
consistently demonstrate confidence, strength and decisiveness. Where their behaviour
diverges from these male-stereotypic agentic ideals, they risk becoming ineffective and
who were afforded greater leeway in the leadership style they adopted.
These findings make a clear case that both women and men now react negatively to
non-agentic behaviour from women in leader roles. However, given the novelty of this
finding, we conducted a replication study using a different issue, leaders who were more
proximal to the target to allow for a broader measureof likeability,and using audio rather
than written presentations to manipulate leader agency.
Study 2 changed the issue from climate change to the topic of increasing university
tuition fees, which were being proposed in universities in the region at the time of the
study and had led to student protests.Although Study 1 wasbased on Carli’s (1990, Study
2) general approach, this study matched Carli’s methodology even more closely on
targets (students), message (unpopular), and mode of presentation (audio recordings).
Hence, participants listened to audio recordings of male or female student advocates
presenting an unpopular proposal to increase tuition fees, using either assertive or
The study had a 2 (speech style: assertive, tentative) 92 (leadergender: male, female)
9 2 (participant gender: male, female) between-participants design. It was again
predicted that there would be a speech style 9 leader gender interaction, with tentative
female leaders regarded as less likeable and influential than assertive female leaders and
tentative male leaders. Assertive female leaders were predicted to be as likeable and
influential as assertive male leaders and the likeability and influence of male leaders was
not expected to vary based on the assertiveness of their speech.
Sixty-six participants from an Australian University (56% female: Mage= 20.4 years,
SD = 3.3)completedthestudy.Themajorityofparticipantswererecruitedfromfirstyear
psychology and received course credit. The remaining participants were approached on
Materials and procedure
An identical procedure to Study 1 was followed, except that participants were informed
that they would hear the opinion of a student advocate over whether they supported a
informed that the rise in fees, if implemented, would only apply to new students and not
about the need to reduce student-staff ratios, attract high-quality teachers, upgrade and
enhance access to information technology, and to improve parking facilities.
Prejudice towards tentative women 225
Participants listened to the speech using individual audio players and headsets.
Audiotapes were made by two male and two female actors reading the speech in an
forthright manner, while in the tentative version, the student advocate sounded more
hesitant and uncertain, with the script adding numerous hedges, qualifiers, and
hesitations (e.g., ‘[I’m no expert of course, but it seems to me that] The revenue raised
from this fee increase will [you know] allow the university to maintain its reputation as
[um] an institution of excellence…’).
This study used an expanded set of items relative to Study 1. In addition to general
demographic questions, the questionnaire contained the following key measures.
In this study, interaction with the leader was more plausible than that in Study 1, so we
included extra items relating to attitudes to social interactions with the leader, using an
‘I could see myself enjoying meeting this student advocate in person’; a = .82).
A multi-item scale of influence was adapted from Blankenship and Holtgraves (2005),
includingthreeitems(Ithinkraisingfeesby25%wouldbe,1 = bad/foolish/undesirable,
9 = good/wise/desirable; a = .88).
Agency and communality
A 4-item scale was used to assess leader agency (dominant, forceful, assertive, acts as a
leader; a = .92) and a 4-item scale assessed leader communality (friendly, sensitive,
sympathetic, generous; a = .76).
All participants correctlyidentified the student advocate’s genderand position on the fee
increase. A 4 (individual speaker) 9 2 (speech style) analysis of variance for agency and
communality revealed no effects involving individual speakers (all ps > .28), so this
variable was not included in subsequent analyses.
effective - assertive leaders were attributed significantly more agency (M = 6.29,
226Renata Bongiorno et al.
SD = 1.63) than tentative leaders (M = 3.15, SD = 1.30), F(1, 58) = 67.79, p < .001,
gp2= .54. No other effects for agency were significant.
For communality, there was no significant difference between assertive (M = 5.36,
SD = 1.58) and tentative (M = 5.68, SD = 0.97) leaders’ communality ratings, F(1,
58) = 1.56, p = .217. Although no communality effects were significant, for the focal
analyses we continued to control for communality as a covariate to match Study 1 (for
correlations, see Appendix).
Tests of hypotheses
To test hypotheses, likeability, and influence were examined using a 2 (speech style) 9 2
(leader gender) 9 2 (participant gender)4analyses of covariance, with communality
entered as a covariate.
The predicted interaction between speech style and leader gender was identified,5F(1,
57) = 10.37,p = .002,gp2= .15,qualifyingasignificantmaineffectforspeechstyle,F(1,
57) = 11.47, p = .001, gp2= .17. Means are shown in Figure 2. The three-way interac-
tionwasnotsignificant(p = .468)indicatingthatparticipantgenderdidnotmoderatethe
expected two-way interaction. No other effects were significant.
Thespeech style byleadergender interactionwasexaminedbyseparate comparisons
within leader gender and within speech style, controlling for communality. Comparing
within each leader gender, as predicted, tentative female leaders were less likeable than
assertive female leaders, F(1, 30) = 17.92, p < .001, gp2= .37, but there was no
significant difference in the likeability of assertive and tentative male leaders, F(1,
30) = 0.001, p = .972. Comparing within each speech style, tentative female leaders
Figure 2. Mean scores (+SE) for Study 2 for likeability and influence as a function of leader gender and
4Participant gender was included as a factor to match the analytic approach used in Study 1. However, due to relatively smaller
cell sizes, we replicated the main analyses without this factor. The predicted interaction for likeability remained significant,
p = .003, and the interaction for influence became marginal p = .067.
5The interaction remained significant, p = .043 after the communality covariate was omitted.
Prejudice towards tentative women227
werelesslikeablethantentativemale leaders,F(1,28) = 4.60,p = .041,gp2= .14,while
assertive female leaders in this case were rated higher on likeability than assertive male
leaders, F(1, 32) = 5.07, p = .031, gp2= .14.
The predicted interaction between speech style and leader gender was significant,6F(1,
57) = 4.07, p = .048, gp2= .07. No other effects were significant, including the three-
way interaction (p = .813). To interpret the two-way interaction, comparisons were
made within leader gender and within speech style, controlling for communality.
female leaders significantly less influential than assertive female leaders, F(1, 30) = 6.27,
p = .018, gp2= .17, but no difference between tentative and assertive male leaders, F(1,
30) = 0.01, p = .911. Comparing within each speech style also showed the predicted
differences, with tentative female leaders significantly less influential than tentative male
leaders, F(1, 28) = 8.64, p = .007, gp2= .24, but no significant difference in influence
between assertive male and assertive female leaders, F(1, 32) = 0.07, p = .788.
Study 2 replicated and extended the findings from Study 1 using a different target, topic,
and mode of presentation, providing even greater evidence of prejudice towards women
who fail to behave agentically in leader roles. Tentative female leaders were found to be
less likeable and less influential than both assertive female leaders and tentative male
leaders. Consistent with predictions, assertive female leaders achieved a similar level of
influence to assertive male leaders, and were rated even more favourably than assertive
male leaders in terms of their likeability. This finding further emphasizes that women’s
agentic behaviour is no longer a source of prejudice (see Diekman, 2007; Heilman &
Okimoto, 2007, for similar findings showing no prejudice based on women’s agentic
Theresultsofthecurrentstudies demonstrateasubtleform ofprejudice towards women
who fail to show agency in leader roles. Consistent with predictions, tentative women
were less likeable and achieved less influence compared to both assertive women and
tentative men. Prejudice towards tentative women was shown by both male and female
participants, and occurred regardless of whether leaders were public figures or peers,
whether they expressed popular or unpopular views, and for both written and audio
presentations, suggesting the effect is pervasive. Prejudice towards women’s agentic
Thus, women who spoke assertively were at least as likeable and influential as assertive
men. As further evidence of men’s greater scope of behaviour in leader roles, the
assertiveness of men’s speech did not affect their likeability or influence.
6The interaction becamemarginal,p = .079after the communality covariatewas omitted,however,comparisons continuedto
show predicted differences (ps < .027).
228Renata Bongiorno et al.
Comparing the present findings with Carli (1990) findings suggests a shift in
prescriptive stereotypes of women’s agenticbehaviour – agentic behaviour from women
that was penalized. These findings signal a turnaround from times when endorsement of
traditional gender roles was greater and non-agentic behaviour from women was
rewarded, consistent with their lower status and subordinate position to men.
Nonetheless, this subtle prejudice towards women’s non-agentic behaviour presents
another barrier to equality that may prove more difficult to overcome. Unlike traditional
agentic behaviour is likely to appear fair and legitimate, as non-agentic behaviour is
inconsistent with expectations of how a leader should act. In this way, our findings share
similarities with the aversive racism literature (Dovidio & Gaertner, 2000), which posits
that because contemporary sensibilities and laws have made discrimination immoral and
In highlighting this subtle prejudice towards female leaders, our findings also expose
the continuing bias towards male leaders, who escaped similar social penalties for
equivalent non-agentic behaviour. These findings reflect past research showing men are
given greater leeway in leadership (Carli, 1990, Study 2; for a meta-analysis, see Eagly
et al., 1992), and are equally important to explaining men and women’s differential
leadership outcomes. This bias towards men is likely to result from the stronger
2011), which may heighten trust in men’s leadership ability. Although these findings
the ‘benefit of the doubt’ they are given in circumstances where their leader behaviour
diverges from the ideal is likely to give them a critical upper hand, including the
negativity towards female leaders from even a brief exposure to their non-agentic
behaviour suggests women are met with the unrealistic expectation that they should
always meet the behavioural ideals associated with that role, placing them in a more
precarious position as leaders.
To overcome this subtle prejudice towards women it is critical to understand its
underlying motivations. It is noteworthy that female and male participants displayed
female participants (e.g., Carli, 1990, Study 2). This alignment in reactions corresponds
with the narrowing gap in gender differences in liberal/feminist attitudes, which were
women in their liberal attitudes (Twenge, 1997). Although prejudice towards women’s
agentic behaviour in leadership may persist for those who continue to explicitly endorse
traditional gender roles (cf. Eagly & Karau, 2002), evidence of more liberal attitudes
non-agentic female leaders who are penalized. We propose two distinct motivational
processes for this subtle form of prejudice.
For people who genuinely support progress towards gender equality in society,
in ways that prove their equivalence to men in terms of agency and leadership suitability
projected to increase further in the future). Evidence of non-agentic behaviour from
because it could beused as an example ofwhywomen are not suited to leadership.
Prejudice towards tentative women229
1995), prejudice towards non-agentic women may be driven by a desire to find more
legitimate ways of limiting women’s time as leaders. That is, in the light of greater
normative acceptance of women in leadership, penalizing non-agentic women may be
preferred, because it can be portrayed as consistent with a meritocracy, but remains
disadvantageous to women because men avoid similar penalties.
While prejudice towards non-agentic women may therefore be motivated by both a
desire for continued social change or finding new ways to maintain gender inequality,
another possible explanation is that it reflects attentional processes. As women are
counter-stereotypical leaders, perceivers may pay closer attention to their leadership
of agency. However, our findings suggest such attentional processes are unlikely to
and female leaders across studies. Thus, evaluation processes provide a more plausible
explanation. That is, despite similar judgements about the level of agency of female and
male leaders, non-agentic behaviour was only evaluated more negatively in women.
Although only agency was manipulated in these studies, attributions of communality
were also measured and revealed a marginally significant finding for agentic leaders to
be seen as more communal (Study 1), or no different in communality (Study 2) than
women’s agentic behaviour with non-communality (for a discussion, see Heilman &
not necessarily lead to the perception that they are non-communal. Study 1 suggests that
aligned with communality, an association observed in similar circumstances by Judd,
James-Hawkins, Yzerbyt and Kashima (2005, Study 4). Such findings underscore the
importance of conceptualizing agency and communality as distinct dimensions, whose
relationship can be positive, negative or neutral (see Judd et al., 2005, for an exploration
of the factors governing this relationship). Through doing so, it is possible to grasp the
precise basis for (and changing nature of) prejudice towards women in leadership.
In the light of the findings for leader communality, the lack of prejudice towards
agentic female leaders should not be interpreted as acceptance of women’s non-
communal behaviour in leader roles. It is clear from other research that women continue
Okimoto, 2007; Heilman et al., 2004) or actual (Bowles et al., 2007; Heilman & Chen,
aggressive, ruthless or domineering (cf. Rudman & Glick, 1999, 2001), prejudiced
reactions towards women are likely to persist.
Moreover, because our findings for agentic female leaders are based on independent
evaluations of male and female leaders, it is not anticipated that they will translate into
male domains. Examples include where a woman outperforms a man in a test about
football knowledge, or the most effective way to punch an opponent (Rudman &
Fairchild, 2004). Such demonstrations are still likely to lead men to feel emasculated or
humiliated, eliciting greater disapproval when enacted by a woman than a man.
Furthermore, men’s greater leeway to be non-agentic in leadership roles is unlikely to
extend to contexts where evidence of their masculinity can be more easily questioned,
such as where they are advocating for a feminist cause (Anderson, 2009), or where they
230 Renata Bongiorno et al.
excel in ‘feminized’ roles not traditionally associated with male success (Heilman & Download full-text
Although this research used language to manipulate agency (an assertive vs. tentative
speech style), it is possible that people’s reactions to language differ from other
behaviours that indicate agency. Therefore, different behavioural indicators of agency
versus non-agency should be used in future research, such as describing a leader’s
decisiveness or indecisiveness in making business decisions. Although language and
communication are critical aspects of leadership behaviour, using non-linguistic
indicators of agency would enable further assessment of the generalizability of the
It would also be beneficial for future research to replicate these studies with different
samples. Interview data from corporate leaders consistent with pressure on women
to behave agentically to be accepted (Davies-Netzley, 1998), suggest prejudice towards
are common in experiments relating to reactions towards women’s agentic behaviour
(e.g.,Diekman,2007;Heilman&Okimoto,2007;Rudman&Glick, 1999),and enabled us
to increase the validity of comparisons with Carli’s (1990) research, replicating these
findings using corporate and other workplace populations would provide a valuable
other roles requiring agency, this view has become increasingly untenable, and most
people in contemporary western societies now explicitly endorse women’s right to be
and influential as equivalent men. While accepting agency in women is a step in the right
direction, this research exposes that women who fail to show agency in leadership are
swiftly singled out for disapproval, whereas this does not occur for men. Based on men’s
signs of confidence and strength will provide a considerable challenge. While a few
women will be able to meet this expectation, the majority who cannot remain
disadvantaged, with men avoiding similar penalties for equivalent non-agentic behav-
effort if gender equality is to be achieved.
We thank Ngaire Donaghue, Matthew Hornsey, Jolanda Jetten, Colin Leach, Barbara Masser,
Michelle Ryan, and Laurie Rudman for their valuable feedback on versions of this manuscript,
and acknowledge Craig McGarty for providing initial advice.
Adler, N. J. (1999). Global leaders: Women of influence. In G. Powell (Ed.), Handbook of gender &
work (pp. 239–261). Thousand Oaks, CA: Sage.
Anderson, V. N. (2009). What’s in a label? Judgments of feminist men and feminist women.
Psychology of Women Quarterly, 33, 206–215. doi:10.1111/j.1471-6402.2009.01490.x
Barreto, M., Ryan, M., & Schmitt, M. T. (Eds.) (2009). The glass ceiling in the 21st century:
Understanding barriers to gender equality. Washington, DC: American Psychological
Prejudice towards tentative women 231