"The Compassionate Sexist? How Benevolent Sexism Promotes and Undermines Gender Equality in The Workplace

Journal of Personality and Social Psychology © 2016 American Psychological Association
2016, Vol. 111, No. 5, 706 –727 0022-3514/16/$12.00 http://dx.doi.org/10.1037/pspi0000072
The Compassionate Sexist? How Benevolent Sexism Promotes and

Undermines Gender Equality in the Workplace
Ivona Hideg D. Lance Ferris

Wilfrid Laurier University The Pennsylvania State University
Although sexist attitudes are generally thought to undermine support for employment equity (EE)
policies supporting women, we argue that the effects of benevolent sexism are more complex. Across 4
studies, we extend the ambivalent sexism literature by examining both the positive and the negative
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
effects benevolent sexism has for the support of gender-based EE policies. On the positive side, we show
This document is copyrighted by the American Psychological Association or one of its allied publishers.
that individuals who endorse benevolent sexist attitudes on trait measures of sexism (Study 1) and
individuals primed with benevolent sexist attitudes (Study 2) are more likely to support an EE policy, and
that this effect is mediated by feelings of compassion. On the negative side, we find that this support
extends only to EE policies that promote the hiring of women in feminine, and not in masculine, positions
(Study 3 and 4). Thus, while benevolent sexism may appear to promote gender equality, it subtly
undermines it by contributing to occupational gender segregation and leading to inaction in promoting
women in positions in which they are underrepresented (i.e., masculine positions).
Keywords: gender equality and diversity, benevolent sexism, affirmative action/employment equity,
compassion, masculinity and femininity
Supplemental materials: http://dx.doi.org/10.1037/pspi0000072.supp
Women continue to face barriers in the workplace that hinder nies in the European Union, and 15.9% of board seats for Financial
their career advancement. This situation is particularly pronounced Post 500 companies in Canada (Catalyst, 2013a, 2013b; European
for high ranking positions, with women holding only 16.9% of Commission, 2014). Moreover, approximately 40% of Financial
board of directors seats for Fortune 500 companies in the United Post 500 companies had no women at all on their board of
States, 17.8% of board seats for the largest publicly listed compa- directors (Catalyst, 2013a). To address these inequalities, gender-
based employment equity (EE)1 policies, or policies that aim to
reduce discrimination and increase the hiring of women, have been
implemented in many countries worldwide (Sowell, 2004; Yang,
This article was published Online First July 14, 2016. D’Souza, Bapat, & Colarelli, 2006). While EE policies have pos-
Ivona Hideg, Lazaridis School of Business and Economics, Wilfrid
itive and socially beneficial goals which presumably should garner
Laurier University; D. Lance Ferris, Smeal College of Business, The
Pennsylvania State University.
support from employees, a large body of research shows negative
We thank Andrew Carton and Anne Wilson for their helpful comments employee reactions to EE policies (see Harrison, Kravitz, Mayer,
on previous versions of this article, as well as John Trougakos for helping Leslie, & Lev-Arey, 2006, for a meta-analysis). These negative
us collect data in his classes and Maddie D’Agata, Teodora Makaji, Peter reactions are problematic for governments and organizations that
Fisher, and Anja Krstic for their assistance with data collection. We also implement EE policies because negative reactions undermine the
thank the participants of the industrial-organizational psychology seminar success of EE policies in promoting gender diversity and equality
series at the University of Waterloo; management seminar series at the in the workplace (cf. Crosby, Iyer, & Sincharoen, 2006).
University of Western Australia, the Hong Kong Polytechnic University, One of the main factors undermining support for EE policies is
and the University of Guelph; and the Gender, Diversity, and Inclusion prejudiced attitudes such as sexism (e.g., Konrad & Hartmann,
Symposium at RMIT University at Melbourne. Finally, we also thank the
2001; Son Hing et al., 2011; Tougas, Brown, Beaton, & Joly,
participants of Diversity Publishing Workshop organized by the Gender
and Diversity in Organization division at the Academy of Management in
1995). Researchers typically postulate a straightforward predic-
2013 for their developmental feedback, and Católica Lisbon School of tion: The more sexist an individual’s attitudes, the less likely the
Business & Economics for their support as the Ivona Hideg was visiting individual is to support the EE policy. And not surprisingly, this
their Organizational Behavior group while working on the final revisions prediction holds: Past work has repeatedly shown that sexist atti-
for the paper. This research was supported in part by a research grant tudes decrease support for gender-based EE policies (Harrison et
awarded to both authors from the Social Sciences and Humanities Research al., 2006).
Council of Canada (435-2014-1147) and Early Researcher Award from the
Ontario Ministry of Research and Innovation awarded to the first author.
Correspondence concerning this article should be addressed to Ivona 1
In the United States, the equivalent term is affirmative action. Given
Hideg, Lazaridis School of Business and Economics, Wilfrid Laurier that three of our four studies were conducted in the context of Canadian
University, 75 University Avenue West, Waterloo, Ontario, Canada N2L employment equity policies, we use the term employment equity through-
3C5. E-mail: ihideg@wlu.ca out this article.
706
BENEVOLENT SEXISM AND COMPASSION 707
In this article, we argue for a more complex perspective on the we argue that these findings are limited given that there are various
effects sexism has on EE policy support. While acknowledging the forms sexism can take. In particular, ambivalent sexism theory
numerous previous findings showing a straightforward negative suggests that there are two complementary forms of sexism: hos-
relation between sexism and EE policy support, we draw on tile and benevolent sexism (Glick & Fiske, 1996, 2001; Glick et
ambivalent sexism theory (Glick & Fiske, 1996, 1997) to contend al., 2000). Hostile sexists, or what most people consider “sexists”
that this is a byproduct of past work almost always conceptualizing to be (Glick & Fiske, 1996), view women as inferior to men in
sexism as hostile sexism, or feelings of antipathy and hostility general and as trying to control men through feminist ideology and
toward women (cf. Allport, 1954). Conversely, past work on sexual seduction (Glick & Fiske, 1996, 1997). Past research has
reactions to EE policies has almost always ignored benevolent found that hostile sexism is related to adopting negative stereo-
sexism, or subjectively positive attitudes toward women character- types and evaluations of women (e.g., Glick & Fiske, 1996; Glick
ized by a sense of protection, idealization, and affection for et al., 2000; Masser & Abrams, 2004), as well as imbalanced
women. We suggest that distinguishing between hostile and be- traditional gender role norms such as the notion that a wife should
nevolent forms of sexism is critical, as their effects on support for assist her husband’s career even at the cost of her own career, and
EE policies likely differ. In particular, we posit that EE policies that a wife should do the housework and be responsible for
may be particularly likely to evoke a sense of compassion in nurturing children (Chen, Fiske, & Lee, 2009). Overall, past re-
benevolent sexists, or a sympathetic emotional reaction which search has shown that hostile sexism relates to a host of negative
arises when seeing others that are vulnerable and need help (Goetz,
attitudes and evaluations of women.
Keltner, & Simon-Thomas, 2010). This sense of compassion, in
Benevolent sexists, on the other hand, have subjectively positive
turn, should evoke a desire to help women via support for EE
attitudes toward women but that ultimately characterize women as
policies.
wonderful, yet weak (Glick & Fiske, 1996, 1997). For example,
However, our focus on compassion also allows us to outline a
benevolent sexists believe that women are delicate and should be
boundary condition that illustrates how benevolent sexism ulti-
protected and taken care of by men, that while women, compared
mately undermines occupational gender equality. In particular, our
model suggests that, should benevolent sexists not experience to men, have traits that men generally lack such as sensitivity to
compassion, benevolent sexism should not lead to EE policy others’ feelings, they may not have traits necessary to lead and
support; in line with this, we argue the compassion benevolent govern important social institutions and that women should be
sexists experience is limited to EE policies that promote female valued and protected because they fulfill men’s romantic needs
employment in jobs that adhere to the stereotypical gender roles (Glick & Fiske, 1996, 1997).
(i.e., jobs perceived to be feminine such as teachers or nurses). As In contrast to findings for hostile sexism, past research has
such, while benevolent sexism appears to promote gender equality found positive effects of benevolent sexism such as its association
via support for gender-based EE policies, it subtly suffocates with positive stereotypes and evaluations of women (e.g., Glick &
gender equality by keeping women in occupational “gender ghet- Fiske, 1996; Glick et al., 2000). Indeed, in comparison to hostile
tos.” In what follows, we outline our theoretical logic in more sexism, benevolent sexism is more prevalent and socially accepted
detail, and subsequently present four studies designed to test our in today’s society, and women are just as likely as men to hold
predictions. benevolent sexist attitudes (Becker, 2010; Glick & Fiske, 2001);
both men and women oftentimes do not even perceive benevolent
sexism as gender discrimination (e.g., Barreto & Ellemers, 2005;
EE Policies and Ambivalent Sexism
Becker, 2010), and women may prefer a profile of a benevolent
EE policies have become important tools for reducing gender sexist man over a profile of a nonsexist man (Kilianski & Rudman,
inequality worldwide (Tougas & Beaton, 1993; Yang et al., 2006). 1998). At the same time, this subjective positivity of benevolent
Yet, despite the fact that EE policies assist organizations in ad- sexism masks how it reinforces gender inequality. For example,
dressing past discrimination, promoting social justice, and increas- past research has found that treating women in a benevolent sexist
ing the pool of job candidates, employees tend to show low levels manner undermines women’s cognitive performance (Dardenne,
of support for EE policies (Harrison et al., 2006). A meta-analysis Dumont, & Bollier, 2007), that benevolent sexist managers are
by Harrison et al. (2006) identified prejudiced attitudes against more likely to assign women to less challenging roles in the
racial minorities (racism) and women (sexism) as one of the most workplace (King et al., 2012), and that endorsement of benevolent
influential factors in predicting negative reactions to EE policies.2 sexism is related to support of status quo (Becker & Wright, 2011;
In particular, they found sexism has a strong negative relation with Jost & Kay, 2005).
favorable attitudes toward EE policies (␳ ⫽ ⫺.518); indeed, the
relation between sexism and policy support was the strongest of all
2
examined individual difference variables (including self-interest, In addition to individual differences (e.g., sexism), Harrison et al.
belief in discrimination, experience of discrimination, and political identified a policy type as an important factor influencing support for EE.
Namely, they found that when EE policies give preference to disadvan-
ideology and affiliation). This is in line with an array of studies taged group members with less or no regard to said members’ job quali-
showing that sexist attitudes are related to less favorable attitudes fications (i.e., strong preference policies), employees’ reactions become
toward EE policies (e.g., Konrad & Hartmann, 2001; Konrad & increasingly negative in comparison to policies that consider qualifications
Spitz, 2003; Tougas et al., 1995), and opposition to EE policy first and demographics second (i.e., weak preference policies).Given that
EE policies that consider qualifications after demographics are illegal in
implementation (Son Hing et al., 2011). most countries including Canada and the United States (Pyburn, Ployhart,
While these results highlight the importance of sexist attitudes & Kravitz, 2008), in our studies, we employ legal, weak preference EE
in predicting whether or not an individual will support EE policies, policies.
708 HIDEG AND FERRIS
Taken as a whole, research from an ambivalent sexism frame- extensive work by Heilman and colleagues has demonstrated that
work suggests that the effects of benevolent sexism on attitudes EE policies can cause both beneficiaries (e.g., Heilman & Alcott,
and evaluations of women tend to be complex. On the one hand, 2001; Heilman, Simon, & Repper, 1987) and nonbeneficiaries
women are viewed as possessing positive idealized qualities de- (e.g., Heilman, Block, & Lucas, 1992; Heilman, Block, &
serving of protection, affection, and (presumably) support; on the Stathatos, 1997) of EE policies to view the beneficiaries of EE
other hand, women are viewed as weak and lacking abilities. We policies as being low in competence.3 Thus, while benevolent
argue that this complex nature of benevolent sexism is reflected in sexists may in general not strongly demean women’s competence,
the extent to which benevolent sexists support EE policies. In the presence of an EE policy may make negative appraisals of
particular, and in contrast the unilaterally negative effect hostile women’s competence more salient.
sexism has on support for EE policies, we argue that benevolent From the perspective of the stereotype content model, viewing
sexists will feel a sense of compassion for women and support women as warm yet incompetent means benevolent sexists are
programs (i.e., EE policies) that seek to protect them. However, likely to experience the emotion of pity when considering EE
this support is qualified by benevolent sexist views of women as policies meant to enhance the employability of women. Pity rep-
weak and incompetent, meaning EE policies should only be sup- resents an ambivalent emotion comprising feelings of both sadness
ported when they seek to place women in gender-appropriate and compassion, and which sometimes leads to helping pitied
positions. Below, we outline (and, in Studies 1 and 2, test) how others (also referred to as active facilitation) and sometimes leads
benevolent sexism may have the aforementioned positive effects to inaction and avoidance of pitied others (also referred to as
on support for EE policies via feelings of compassion. Subse- passive harm; see Cuddy et al., 2007). In particular, feelings of
quently, we outline (and, in Studies 3 and 4, test) how this sadness and disrespect toward a group are likely to lead to dis-
seemingly positive effect of benevolent sexism on support for tancing from the group, while feelings of compassion toward a
women actually serves to reinforce gender inequality by limiting group is more associated with subsequent helping behaviors to-
women to “feminine” jobs. ward the group (Goetz et al., 2010; Oveis, Horberg, & Keltner,
2010). We suggest that benevolent sexists will be particularly
How EE Policies Evoke Compassion for likely to experience compassion (and be inclined to help women)
Benevolent Sexists given women, unlike other pitied groups such as orphans or
tsunami victims, cannot easily be avoided or distanced from given
Past research suggests that there are two main dimensions along their ubiquitous presence and given most individuals would not
which groups can be perceived: competence and warmth (Cuddy, want to avoid or distance themselves from women (who despite
Fiske, & Glick, 2007; Fiske, Cuddy, Glick, & Xu, 2002). In being pitied are still generally highly valued in their own right;
particular, groups viewed as occupying high status roles are seen Glick & Fiske, 1996; Guttentag & Secord, 1983). As such, we
as competent, whereas groups who traditionally occupy low status expect that benevolent sexists are likely to experience the emotion
roles are not; groups viewed as competitive are seen as lacking of compassion toward women when considering EE policies.
warmth, whereas groups seen as cooperative are seen as warm The experience of compassion should, in turn, be related to
(Cuddy et al., 2007; Fiske, Xu, Cuddy, & Glick, 1999; Fiske et al., greater support for EE policies that are helping women secure jobs.
2002). The combination of perceptions of warmth and perceptions Past theorizing posits that compassion acts as a prosocial motivator
of competence, in turn, evokes emotional responses: for example, to help others through its other-orientation, or concern for others
those viewed as highly competent but low in warmth evoke feel- and their hardships (Goetz et al., 2010; Miller, Grimes, McMullen,
ings of envy, while those viewed as low in both warmth and & Vogus, 2012). By directing one’s attention toward, and increas-
competence evoke feelings of contempt (Cuddy et al., 2007). ing one’s awareness and understanding of, others’ vulnerable
Given that past research on intergroup affect and stereotypes circumstances, compassion leads to engaging in more helping
suggests that emotions arising from the combination of warmth behaviors designed to alleviate unfavorable conditions of others
and competence perceptions are the key mediators of the effect of (Atkins & Parker, 2012; Dutton, Worline, Frost, & Lilius, 2006;
stereotypes (such as benevolent sexism) on behaviors (Cuddy et Kanov et al., 2004). In particular, by increasing understanding and
al., 2007; Fiske et al., 2002; Mackie, Devos, & Smith, 2000), a appreciation of others’ vulnerable circumstances, compassion en-
pertinent question becomes “How warm and competent do benev- ables people to connect with others and instill a desire to alleviate
olent sexists view women?” With respect to warmth, past research others’ suffering (Batson & Shaw, 1991). We thus argue that in
has consistently shown that benevolent sexism is associated with response to the feelings of compassion aroused when viewing EE
the appraisal of women as being warm (Becker, Glick, Ilic, & policies, benevolent sexists will seek to help out women by sup-
Bohner, 2011; Delacollette, Dumont, Sarlet, & Dardenne, 2013).
With respect to competence, there are indications that benevolent
3
sexism may be related to perceiving women as low in competence. In line with Heilman’s work, a meta-analysis by Leslie, Mayer, and
Indeed, a definition of benevolent sexism suggests that benevolent Kravitz (2014) suggests that the presence of an EE policy is negatively
associated with perceptions of women’s competence. Interestingly, Leslie
sexist view women as wonderful, yet weak (Glick & Fiske, 1996, et al. also found that a presence of an EE policy is negatively associated
1997). Moreover, empirical research has shown that women who with perceptions of women’s warmth. We suggest that finding is qualified
are treated in a benevolently sexist manner are perceived as less by benevolent sexism, that is, benevolent sexists would be more likely to
competent (Good & Rudman, 2010). appraise women as high in warmth even when presented with EE policies.
Consistent with this, in a separate study of 89 participants, we found
There is also reason to believe that within the context of EE support for the notion that benevolent sexists do indeed maintain their
policies, benevolent sexists—and, indeed, all individuals—are perception of women as warm, even when presented with an EE policy. For
likely to view women as being low in competence. In particular, more details, please contact the first author.
porting the EE policies that will assist women in improving their study took place. Thus, our sample consisted of real job applicants
lot in life. who were applying for 4-month full-time jobs. Participants first
completed an online survey at the beginning of a semester assess-
The Present Studies ing their benevolent and hostile sexist attitudes and demographics.
One month after completing the online survey, participants were
To test our model we conducted four studies in the context of given an in-class presentation of a gender-based EE policy that
Canadian (Studies 1–3) and American (Study 4) gender-based EE was ostensibly being developed at the university to be applied to
policies. In Study 1, we examined whether individuals who en- students’ job placements (see below) and were told that their input
dorse benevolent sexist attitudes on trait measures of benevolent is sought as the EE policy is relevant to them. Following the
sexism are more likely to support the EE policy, and if this effect presentation, the participants first completed a measure assessing
is mediated by compassion. In Study 2, we sought to establish the their state experience of compassion, and then measures assessing
causal effects of benevolent sexism by testing whether priming their attitudes and behavioral intentions to promote the proposed
benevolent sexist stereotypes (compared to a control condition) EE policy. The data from the online survey and the in-class survey
leads to more compassion and support for the EE policy. In Study were matched using student identification numbers, which partic-
3 and 4 we sought to identify a key boundary condition of the ipants provided in both surveys.
positive effect of benevolent sexism on support for the EE policy: Materials. Participants read about an EE policy for women
whether the EE policy promotes the hiring of women in feminine ostensibly under development at their university. This policy was
or masculine jobs. Further, given pity is viewed as comprising both developed and used in previous research (for the exact wording of
compassion and sadness but our theoretical development argues the policy, see Appendix A in Hideg & Ferris, 2014) and involved
compassion in particular is what would be invoked for benevolent hiring students into co-op jobs through the university’s co-op
sexists in the context of gender-based EE policies, in all four
center. The EE policy first stated that women are currently under-
studies we also sought to rule out sadness as an alternate expla-
represented in managerial and professional positions; the concept
nation for our results.
of EE policies and their legislation in Canada were then described.
A comparison of hiring rates for men and women in managerial
Study 1: Individual Differences in Endorsement of and professional positions in the region followed. Specifically,
Benevolent Sexism Are Related to Compassion and participants read that the proportion of men and women in these
Support for a Gender-Based EE Policy positions were 65% and 35%, respectively. The policy then sug-
gested that for women to be hired according to their rates of
In Study 1, we measured participants’ individual differences in
university graduation and other related factors, the rate of their
endorsement of benevolent sexism and examined the effect of
hiring should be 55% instead of 35%. Thus, the policy proposed
benevolent sexism on support for a gender-based EE policy, as
that co-op companies should favor hiring women over men to
well as compassion’s mediating effect, in the context of a Cana-
begin to redress the unequal representation of women in the
dian gender-based EE policy for students’ co-operative (co-op)
workforce—with the caveat that this preferential hiring should
jobs. Co-op jobs are full-time, paid, semester-long job placements
only occur when the candidates possessed equal qualifications.5
related to students’ field of study, and are an integral part of
Measures. The measures reported below used a 7-point Likert
students’ education where students gain practical skills and work
response scale ranging from 1 (strongly disagree) to 7 (strongly
experience (Linn, Ferguson, & Egart, 2004). As such, these jobs
agree) unless otherwise noted.
are valuable for students’ careers and an EE policy that applies to
Benevolent and hostile sexism. We measured benevolent and
their co-op hiring would be seen as very relevant and influential on
hostile sexism with the 22-item Ambivalent Sexism Inventory
their career. In line with past research (Harrison et al., 2006;
Hideg, Michela, & Ferris, 2011), we conceptualized policy support developed by Glick and Fiske (1996). This scale has two dimen-
as involving favorable attitudes toward the EE policy, and behav- sions: benevolent sexism (e.g., “Women should be cherished and
ioral intentions to promote the EE policy. As such, we tested
whether participants’ endorsement of benevolent sexist attitudes is 4
Our sample size was limited by our access to the two sections of
related to more favorable attitudes toward, and behavioral inten- course. However, using GⴱPower software (Faul, Erdfelder, Lang, &
tions that promote, the EE policy, as well as whether those rela- Buchner, 2007), we estimated that our power to detect main effects with a
tions are mediated by the experience of compassion. medium effect size (f 2 ⫽ 0.15) in multiple regression analyses with a
significance level of ␣ ⫽ .05 with our sample size of 115 was 98%.
5
In Canada, an equal-qualification or a tiebreak policy is in line with
Method Employment Equity Act of 1995, which requires organizations to monitor
their workforce for representation of the protected groups, and design and
Participants and procedure. Participants were 115 business implement EE programs that will accelerate and help hiring of members
undergraduate students (52 men and 63 women; age: M ⫽ 20.25, from protected groups by hiring based on demographics after the qualifi-
cations for the position have been met (Jain, Sloane, Horwitz, Taggar, &
SD ⫽ 3.13) enrolled in an organizational behavior course at a large Weiner, 2003). Similarly, in the United States, a tiebreak policy is consis-
Canadian university who received course research credit for par- tent with federal regulations based on Executive Order 11246, which
ticipation.4 Sixty-seven participants identified as East Asian, 24 as requires organizations to monitor their workforce for representation of
Southeast Asian, 10 as Caucasian, two as Middle Eastern, one as protected groups (Sowell, 2004). If underrepresentation of protected
groups is detected then remedial actions must be taken and can range from
West Indian, and eight as mixed (three unreported). These partic- strategies that include active recruitment of underrepresented groups to
ipants were enrolled in co-op education and were applying for giving additional weight to race or gender when making employment
co-op jobs for the next semester during the semester when this decisions (Crosby, Iyer, Clayton, & Downing, 2003).
protected by men”) and hostile sexism (e.g., “Women seek to gain .001 (d ⫽ 0.79). Also, in line with past research that suggests that
power by getting control over men”). Two composite scores were the gender gap in benevolent sexism is smaller than for hostile
computed, one for benevolent sexism by averaging the 11 benev- sexism (Glick & Fiske, 1996), there were no differences between
olent sexism items, and one for hostile sexism by averaging the 11 men’s and women’s endorsement of benevolent sexism, t(113) ⫽
hostile sexism items. Although the focus of our studies is on 1.05, p ⫽ .269 (d ⫽ 0.20). Consistent with past research on
benevolent sexism, past research documents a significant positive reactions to EE policies (e.g., Harrison et al., 2006), women,
relation between benevolent and hostile sexism; consistent with compared to men, had more favorable attitudes, t(113) ⫽ – 4.37,
past work (Glick & Fiske, 1996, 1997), to test the unique effects p ⬍ .001 (d ⫽ 0.82), and endorsed more behavioral intentions
of benevolent sexism we control for hostile sexism in our analyses. promoting the EE policy, t(113) ⫽ ⫺3.38, p ⫽ .001 (d ⫽ 0.64).
Emotions. Participants rated to what extent they felt “com- Given that our preliminary results showed that hostile sexism
passionate,” “touched,” and “sympathetic” (Oveis et al., 2010) and was positively correlated with benevolent sexism and that both
“sad” at the present moment on a 5-point Likert scale ranging from gender and hostile sexism influenced our outcome variables (i.e.,
1 (very slightly or not at all), 2 (a little bit), 3 (moderately), 4 support for the EE policy), we controlled for hostile sexism and
(quite a bit), to 5 (extremely). A composite compassion score was gender in our analyses below. We note that controlling for either
computed by averaging the first three items (per Oveis et al., hostile sexism or gender did not affect the significance of our
2010), and a higher score indicated higher experience of compas- results, but we nevertheless control for them in our analyses here
sion.6 As noted previously, we used a one-item sadness measure to (and in all subsequent studies) in line with best practices suggest-
rule out the possibility that sadness underlies the relation between ing that inclusion of control variables is warranted (and needed) if
benevolent sexism and support for a gender-based EE policy. control variables are theoretically relevant to the phenomenon
Attitudes toward the EE policy. We measured attitudes to- studied and are statistically related to dependent variables (e.g.,
ward the EE policy with a six-item scale. Sample items include Becker, 2005; Becker et al., 2016; Spector & Brannick, 2011).
“Employment equity as presented in today’s class would constitute Benevolent sexism, compassion, and support for the EE
a good policy” and “I would not like to work at an organization policy. To test whether benevolent sexists are more likely to
that implements this proposed employment equity policy” (re- experience compassion in response to an EE policy, and more
verse-coded; for all items, see Appendix in Kravitz & Platania, likely to have positive attitudes toward and behavioral intentions to
1993). A composite score was computed by averaging the six promote an EE policy, we conducted regressions controlling for
items, and a higher score indicated more favorable attitudes toward hostile sexism and gender (see Table 2 for complete regression
the EE policy. analyses). In line with our predictions, benevolent sexism was
Behavioral intentions. We measured behavioral intentions
positively related to compassion (b ⫽ .42, p ⫽ .002, f 2 ⫽ 0.10),
with an eight-item scale assessing to what extent participants
attitudes (b ⫽ .32, p ⫽ .013, f 2 ⫽ 0.06), and behavioral intentions
would be willing to implement a range of behaviors that would
(b ⫽ .44, p ⫽ .007, f 2 ⫽ 0.07). Thus, individuals who endorsed
promote the EE policy, such as “Volunteer for one day at an
benevolent sexism experienced more compassion and showed
information booth to create public awareness about the employ-
more support for the gender-based EE policy.
ment equity policy” and “Distribute and post flyers at the [univer-
Mediation analyses. We next tested whether compassion me-
sity] promoting the employment equity policy” (for all items, see
diates the relation between benevolent sexism and support for the
Appendix in Hideg et al., 2011). A composite score was computed
EE policy. To rule out the possibility that sadness underlies the
by averaging the eight items, and a higher score indicated a higher
relation between benevolent sexism and support for a gender-
endorsement of behavioral intentions.
based EE policy, we tested a multiple mediator model in which we
included both compassion and sadness as mediators of the relation
Results of benevolent sexism to attitudes and behavioral intentions. To test
a multiple mediator model we used a bias-corrected bootstrapping
Preliminary analyses. Table 1 presents means and standard
deviations by gender (and overall) for each variable, Cronbach’s procedure with 10,000 samples (Hayes, 2013; Preacher & Hayes,
alphas, and partial and zero-order correlations. As seen in Table 1, 2008) while controlling for hostile sexism and gender in the
in line with past research, benevolent sexism and hostile sexism analyses.
were positively correlated, r ⫽ .34, p ⬍ .001 indicating that they Although there was a significant relation between benevolent
are complementary components of sexism (Glick & Fiske, 1996; sexism and sadness (b ⫽ .29, p ⫽ .019, f 2 ⫽ 0.05), there was only
Glick et al., 2000). Also, in line with past research examining a significant indirect effect of benevolent sexism on attitudes
sexism and EE policy support using hostile sexism measures, we through compassion (indirect effect ⫽ .14, 95% confidence inter-
found that hostile sexism was negatively related to attitudes, val [CI] [.06, .27]), and not through sadness (indirect ef-
r ⫽ ⫺.27, p ⫽ .005, and marginally negatively to behavioral fect ⫽ ⫺.04, 95% CI [⫺.17, .01]; see Figure 1, Panel A). Similarly
intentions, r ⫽ ⫺.16, p ⫽ .097. Gender did not moderate the there was only a significant indirect effect of benevolent sexism on
relation between benevolent sexism (or hostile sexism) and sup- behavioral intentions through compassion (indirect effect ⫽ .16,
port for the EE policy. 95% CI [.05, .32]), but not through sadness (indirect ef-
Before proceeding to test our hypotheses, we examined gender fect ⫽ ⫺.04, 95% CI [⫺.16, .01]; see Figure 1, Panel B). These
differences in a) endorsement of sexist attitudes, and b) support for
a gender-based EE policy because past research indicates gender 6
One limitation of this measure was that it did not specify the target of
differences in these domains. Consistent with past research, men emotion. We address this issue by specifying the target of emotions (i.e.,
endorsed hostile sexism more than women, t(113) ⫽ 4.20, p ⬍ women) in Studies 3 and 4.
Table 1
Means, Standard Deviations, Cronbach’s Alphas, and Partial and Zero-Order Correlations (Study 1)
Variable Men (n ⫽ 52) Women (n ⫽ 63) Overall (n ⫽ 115) 1 2a 3b 4 5 6
1. Gender
2. Benevolent sexism 4.42 (.64) 4.28 (.73) 4.34 (.70) ⫺.10 (.74)
3. Hostile sexism 4.35 (.56) 3.86 (.68) 4.08 (.67) ⫺.38ⴱⴱⴱ .34ⴱⴱⴱ (.74)
4. Compassion 2.65 (.86) 2.49 (.99) 2.56 (.93) ⫺.09 .28ⴱⴱ .02 (.76)
5. Attitudes 4.82 (.98) 5.56 (.84) 5.23 (.98) .38ⴱⴱⴱ .23ⴱ ⫺.27ⴱⴱ .29ⴱⴱ (.82)
6. Behavioral intentions 3.89 (1.05) 4.61 (1.20) 4.28 (1.18) .30ⴱⴱ .26ⴱⴱ ⫺.16† .30ⴱⴱ .52ⴱⴱⴱ (.84)
7. Sadness 1.60 (.93) 1.57 (.82) 1.58 (.87) ⫺.01 .22ⴱ .14 .25ⴱⴱ ⫺.05 .01
Note. Gender is coded as 0 ⫽ men and 1 ⫽ women. Columns labelled “men,” “women,” and “overall” present means and standard deviations (standard
deviations are in parentheses). For compassion and sadness, the scale ranged from 1 to 5; and for benevolent and hostile sexism, attitudes, and behavioral
intentions, the scale ranged from 1 to 7. Cronbach’s alphas are presented in parentheses.
a
Correlations in this column are partial correlations controlling for hostile sexism (the exception is a zero-order correlation between benevolent and hostile
sexism). b Correlations in this column are partial correlations controlling for benevolent sexism.
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
results indicate that compassion, not sadness, underlies the relation reduce concerns regarding common method variance issues (Pod-
between benevolent sexism and support for the gender-based EE sakoff, MacKenzie, & Podsakoff, 2012). However, we still cannot
policy. infer causality, rendering it possible that, for example, compas-
sionate individuals are more likely to be benevolent sexists. In
Discussion Study 2, we sought to address this limitation and to provide more
evidence for the positive effect of benevolent sexism on support
Study 1 provides initial empirical evidence that sexism—
for an EE policy by priming benevolent sexist stereotypes.
namely, benevolent sexism— can have positive effects on support
Priming stereotypes refers to incidental activation of knowledge
for EE policies. In particular, we found that the greater an indi-
structures such as stereotype contents (Janiszewski & Wyer, 2014;
vidual’s benevolent sexist attitudes, the more favorable their atti-
Wheeler & Petty, 2001). Stereotype priming research argues that
tudes toward the EE policy and the more likely they were to
salient and cognitively accessible stereotypes automatically prompt
indicate behavioral intentions to promote the EE policy. We fur-
ther found evidence that compassion mediated the effect of benev- people to act in a way that promotes those stereotypes, because
olent sexism on attitudes and behavioral intentions. However, we stereotypes are deeply ingrained in a society and hence lead to
should note that our correlational design precludes causality infer- automatic behaviors (Bargh, Chen, & Burrows, 1996; Devine, 1989).
ences, a point to which we return in our general discussion. Research has shown that activation of stereotypes can, in certain
Nevertheless, Study 1 provides evidence that previously postulated circumstances, influence the attitudes and behaviors even of those
negative effects of sexist attitudes on the support for gender individuals who personally do not endorse those stereotypes (Jost &
diversity may not be straightforward and that in some instances Kay, 2005; Wheeler & Petty, 2001). For example, nonprejudiced
sexist attitudes may actually appear to enhance support for an EE individuals have been shown to act similar to prejudiced individuals
policy. when exposed to stimuli that prime prejudicial stereotypes (Devine,
One limitation, however, associated with Study 1 is that it was 1989). Such effects occur because even though an individual may not
a correlational study, which introduces concerns over common be prejudiced, they are frequently familiar with different stereotypes
method variance and causality. By assessing our predictor (benev- others hold (particularly when such prejudices are prevalent in society
olent sexism) and outcomes variables (attitudes and behavioral and media) and when exposed to a stimuli that primes the stereotype,
intentions) at two different time points separated by 1 month, we they may automatically behave in a way that resembles prejudiced
Table 2
Regression Analyses Result: The Effect of Benevolent Sexism on Compassion, Attitudes, and Behavioral Intentions That Promote the
Gender-Based Employment Equity Policy (Study 1)
Compassion Attitudes Behavioral intentions

Predictors b/␤ 95% CI b/␤ 95% CI b/␤ 95% CI
Step 1 R2 .01 .15ⴱⴱⴱ .09ⴱⴱ

Gender ⫺.10/⫺.05 [⫺.48, .28] .70/.36ⴱⴱⴱ [.34, 1.06] .76/.32ⴱⴱ [.30, 1.22]
Hostile sexism .15/.10 [⫺.41, .43] ⫺.09/⫺.06 [⫺.37, .19] .09/.05 [⫺.25, .43]
ⴱⴱ ⴱ ⴱⴱ
Step 2 ⌬R2 .09 .05 .06
Benevolent sexism .42/.31ⴱⴱ [.20, .64] .32/.23ⴱ [.06, .58] .44/.26ⴱⴱ [.17, .71]
Total R2 .10ⴱⴱ .20ⴱ .15ⴱⴱ
Note. CI ⫽ confidence interval. Gender is coded as 0 ⫽ men and 1 ⫽ women.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
SD ⫽ 3.27) at a midsize Canadian university who received course

research credit for participation.7 Thirty-nine participants identi-
fied as Caucasian, 27 as East Asian, 15 as South Asian, six as
Southeast Asian, two as Middle Eastern, one as Hispanic, and five
as mixed (10 unreported). Participants came to a laboratory and
were randomly assigned to one of two conditions: a benevolent
sexism condition where they read benevolent sexist statements, or
a control condition where they read positive but “gender-
stereotype-neutral” statements (see below for a description of
experimental manipulation). To cover the purpose of the study
participants were told that they were participating in a memory
study and that their task was to memorize given statements and
that they would be later tested on their memory of those state-
ments. Following the exposure to statements, all participants read

the same EE policy as in Study 1. After reading the EE policy, the

participants first completed the same measures as in Study 1, and
then manipulation checks.
Experimental manipulation. We followed Becker and Wright’s
(2011) manipulations to create our experimental (benevolent sexism)
and control groups. In the benevolent sexism condition, we activated
benevolent sexist stereotypes by having participants read and mem-
orize the following six statements from Glick and Fiske’s (1996)
Benevolent Sexism Scale: “Women should be cherished and pro-
tected by men”; “Secretly, most women yearn for a man whose arms
they can find protection and security”; “Women have a way of caring
Figure 1. Multiple mediator models of the relation between benevolent
sexism and attitudes (Panel A) and behavioral intentions (Panel B), via that men are not capable of in the same way”; “Women compared to
participants’ compassion and sadness (Study 1). Values are unstandardized men, tend to have a superior moral sensibility”; “Men are incomplete
coefficients. EE ⫽ employment equity; ns ⫽ nonsignificant. † p ⬍ .10. without women”; and “Many women would like a man who conquers
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. their heart.” These statements were used and validated in Becker and
Wright’s (2011) paper. To establish that the effects of activated
behavior even if it is not their intention (although this is not always the benevolent sexist stereotypes are due to their stereotypical content
case; for reviews, see Devine, 1989; Jost & Kay, 2005; Wheeler & rather than any generally positive attitudes toward women, it was
Petty, 2001). important that our control condition present participants with state-
As noted previously, benevolent sexist perceptions of women as ments that were still favorable toward women but that did not activate
being warm yet in need of protection are prevalent in Western society stereotypical views about women (Becker & Wright, 2011; Jost &
(Becker, 2010; Glick & Fiske, 2001); thus, even individuals who do Kay, 2005). Thus, in line with past research (Becker & Wright, 2011;
not typically subscribe to benevolently sexist views should be sus- Jost & Kay, 2005), participants in the control condition read and
ceptible to activation of those views through priming. In Study 2, we memorized five “gender stereotype-neutral” statements that were pos-
sought to take advantage of this susceptibility to priming and ran- itive in nature but unrelated to stereotypes held toward women (e.g.,
domly assigned participants either to a condition where benevolent “As compared to men, women are more reliable”).8
sexist stereotypes were activated by having participants read and Dependent measures. We used the same measures for com-
memorize statements containing benevolently sexist views or to a passion (␣ ⫽ .90), sadness, attitudes (␣ ⫽ .85), and behavioral
condition where participants read control statements that were posi-
tive in nature but not intended to activate benevolent sexist stereo-
types. In line with Study 1, we expected participants primed with
7
Our power analysis using GⴱPower software (Faul et al., 2007) indi-
benevolent sexism, compared to the control condition, would be more cated that the sample size needed for detecting main effects of medium
effect sizes in analyses of covariance with a significance level of ␣ ⫽ .05
likely to experience compassion and support EE policies; moreover, and 80% power would be 128. However, due to limits of our subject pool,
we expected compassion would mediate the effect of benevolent we only managed to recruit 105 students and thus have an estimated power
sexism (compared to the control condition) on support for the EE of 72%. We address power issues in Study 4 with a larger sample size.
8
policy. Given that Becker and Wright’s (2011) paper used gender stereotype-
neutral statements tested and validated in German society, and what is
considered gender stereotype-neutral differs over time and cultures (Hoff-
Study 2: Priming Benevolent Sexism Predicts man & Hurts, 1990), we conducted a pilot study to find attributes that are
Compassion and Support for a Gender-Based perceived as gender stereotype-neutral in a contemporary Canadian soci-
EE Policy ety. In a different sample of 155 undergraduate business students (93 men
and 62 women), we found that the following five attributes did not differ
significantly from the neutral point on a 5-point scale (1 ⫽ typical female,
Method 2 ⫽ rather female, 3 ⫽ neutral, 4 ⫽ rather male, 5 ⫽ typical male):
resourceful, predictable, refined, reliable, and conscientious. We used these
Participants and procedure. Participants were 105 business five attributes to create control condition statements. We thank Julia
undergraduate students (52 men and 53 women; age: M ⫽ 20.31, Becker for originally suggesting this pilot study to us.
intentions that promote the EE policy (␣ ⫽ .88) as in Study 1; we invokes feelings of compassion and enhances support for the EE
also used the same 7-point Likert response scales ranging from 1 policy. To test these predictions, we conducted analyses of cova-
(strongly disagree) to 7 (strongly agree) for attitudes and behav- riance testing for differences between the two conditions (benev-
ioral intentions and the same 5-point Likert scale ranging from 1 olent sexism vs. control condition) while controlling for gender as
(very slightly or not at all) to 5 (extremely) for compassion and in Study 1. It should be noted that there were no significant
sadness. interactions between condition and gender in predicting compas-
Manipulation checks. In line with previous research (Becker sion, attitudes, or behavioral intentions, and that the conclusions of
& Wright, 2011), at the end of the study participants were pre- our analyses remain the same regardless of whether we include or
sented with the 11 statements used in the two conditions, and were exclude gender in our analyses.
asked to select sentences that they had seen at the beginning of the Means and standard deviations of all measures are presented in
study. We examined whether participants identified the correct Table 3. As expected, participants in the benevolent sexism con-
sentences and used this as a manipulation check to ensure that dition experienced more compassion, F(1, 102) ⫽ 5.28, p ⫽ .024,
participants carefully read and memorized sentences. Selection of 95% CI for mean differences [⫺0.88, ⫺0.07] (␩p2 ⫽ .05), and
a sentence was coded as 1 and a nonselection as 0. We summed endorsed more behavioral intentions, F(1, 102) ⫽ 4.72, p ⫽ .032,
selected sentences for each condition creating two manipulation 95% CI for mean differences [⫺0.65, ⫺0.03] (␩p2 ⫽ .04), than
check scores, a benevolent sexism score and a control score for participants in the control condition. However, there were no
each participant. The benevolent sexism score varied between 0 differences between the two conditions in predicting attitudes, F(1,
(no sentences selected) and 6 (six sentences selected), and the 102) ⫽ .02, p ⫽ .903, 95% CI for mean differences [⫺2.01, 2.27]
control score varied between 0 and 5. (␩p2 ⬍ .001).
Mediation analyses. We further predicted that compassion
Results would mediate the effect of the benevolent sexism condition
Manipulation checks. To test whether participants could re- (compared to the control condition) on attitudes and behavioral
call sentences that they had read at the beginning of the study, we intentions. As in Study 1, we initially planned to rule out the
conducted t tests examining differences in benevolent sexism possibility that sadness underlies the effect of benevolent sexism
scores and control scores between the two conditions. In particular, condition (vs. control condition) on support for a gender-based EE
we tested whether participants in the benevolent sexism condition policy by testing a multiple mediator model. However, given that
were more likely to select benevolent sexism sentences, and par- there were no differences in experienced sadness between partic-
ticipants in the control condition were more likely to select control ipants in the benevolent sexism condition and the control condi-
sentences. As expected, participants in the benevolent sexism tion, F(1, 102) ⫽ 1.02, p ⫽ .314 (␩p2 ⫽ .01), sadness could not
condition were more likely to select benevolent sexism sentences explain the significant observed differences between conditions in
(M ⫽ 5.36, SD ⫽ 0.88) than participants in the control condition support for the gender-based EE policy; consequently, we only
(M ⫽ .02, SD ⫽ 0.21), t(103) ⫽ 39.38, p ⬍ .001 (d ⫽ 8.30). Also, tested compassion as a mediator. Figure 2 presents the results of our
as expected, participants in the control condition were more likely analyses testing the mediating effect of compassion on the effect of
to select control sentences (M ⫽ 4.78, SD ⫽ 0.52) than participants benevolent sexism on behavioral intentions (while controlling for
in the benevolent sexism condition (M ⫽ 0.23, SD ⫽ 0.46), gender, as in Study 1). As expected, there was a significant indirect
t(103) ⫽ 47.08, p ⬍ .001 (d ⫽ 9.27). In summary, participants in effect of benevolent sexism on behavioral intentions through compas-
both the benevolent sexism condition and the control condition sion (indirect effect ⫽ .24, 95% CI [.04, .47]).
showed significantly higher recognition of the statements that they Although, as described above, there were no direct effects of the
saw at the beginning of the study compared to the other condition. benevolent sexism condition (compared to the control condition)
This indicates that participants followed task instructions and on attitudes, more recent treatments of mediation suggest this link
carefully read and memorized statements at the beginning of the might not be necessary and that an indirect effect may nonetheless
study. exist (e.g., MacKinnon & Fairchild, 2009; Shrout & Bolger, 2002).
Benevolent sexism prime, compassion, and support for the In line with this recent literature on mediation, we tested for an
EE policy. The purpose of Study 2 was to test whether priming indirect effect and it was significant (indirect effect ⫽ .14, 95% CI
benevolent sexist stereotypes, compared to a control condition, [.02, .28]). Thus, there was an indirect effect of the benevolent
Table 3
Means and Standard Deviations (Study 2)
Benevolent sexism prime Control prime

Variables Men (n ⫽ 33) Women (n ⫽ 28) Overall (n ⫽ 61) Men (n ⫽ 19) Women (n ⫽ 25) Overall (n ⫽ 44)
Compassion 2.22 (.91) 2.64 (1.31) 2.42 (1.12) 1.84 (.77) 2.04 (1.04) 1.95 (1.06)
Attitudes 4.55 (1.01) 5.56 (.79) 5.01 (1.04) 4.60 (1.14) 5.56 (.96) 5.14 (1.14)
Behavioral intentions 3.23 (1.21) 5.01 (1.04) 4.05 (1.43) 2.93 (1.33) 4.31 (1.17) 3.71 (1.41)
Sadness 1.58 (.87) 1.39 (.57) 1.49 (.74) 1.58 (.96) 1.72 (1.02) 1.66 (.99)
Note. Standard deviations are presented in parentheses. For compassion and sadness, the scale ranged from 1 to 5; for attitudes and behavioral intentions,
the scale ranged from 1 to 7.
point is that benevolent sexists should believe that women

(particularly in comparison to men) are not suited for positions
of power or status in our society (Glick et al., 2000). In other
words, benevolent sexists view women as being warm and in
need of help— but also as being inherently limited in their
abilities. These views may also have implications for the types
of employment positions benevolent sexists view women as
suitable for. We argue that EE policies that seek to place women
in jobs that violate traditional gender roles are likely to alter
how benevolent sexists perceive women, and hence the com-
Figure 2. Mediation of the relation between benevolent sexism condition passion they experience.
(vs. control condition) and behavioral intentions that promote the gender- In particular, gender stereotypes suggest that women are
based employment equity policy, via participants’ compassion (Study expected to be warm, affectionate, helpful, kind, nurturing, and
2).Values are unstandardized coefficients. ns ⫽ nonsignificant. ⴱ p ⬍ .05. gentle; in contrast, men are expected to be assertive, aggressive,
ⴱⴱ
p ⬍ .01.
independent, and decisive (Bakan, 1966). Benevolent sexists
are especially likely to see women in this light (Glick et al.,
2000). Thus, from the view of a benevolent sexist, women
sexism condition (compared to the control condition) on attitudes should be well-suited for feminine occupations characterized by
via compassion. Taken together, these results provide evidence behaviors such as being affectionate, nurturing, and helpful, but
that activating benevolent sexist stereotypes leads to higher expe-
particularly unsuitable for masculine occupations characterized
rience of compassion, which in turn leads to more support for a
by behaviors such as being assertive, independent and aggres-
gender-based EE policy.
sive. If an EE policy is attempting to place women in employ-
ment roles that are more masculine in nature, the EE policy may
Discussion also alter how benevolent sexists view the women helped by
Study 2 establishes causality for the effect of benevolent sexism such policies. When women seek to occupy what are tradition-
on compassion and support for a gender-based EE policy. More- ally viewed as masculine roles, they are less likely to be
over, this study shows that it is not just any positive attitudes perceived as fitting the feminine stereotype of being warm and
toward women that influence support for EE policies, but it is cooperative, but rather are viewed as being competitive and
positive and stereotypical attitudes that cause these effects. uncaring. For example, past research has found that women who
Namely, we found that participants primed with benevolent sexist occupy traditionally male roles are seen as cold and calculating
statements, compared to participants primed with positive, but not (Rudman & Glick, 2001).
stereotypical statements about women, experienced more compas- Consequently, EE policies that seek to place women in tradi-
sion and expressed more behavioral intentions to promote the EE tionally male roles are likely to lead benevolent sexists to view
policy. We also found that compassion mediated the effect of women assisted by such policies as contrary to how women
benevolent sexism on behavioral intentions. Although we did not “should be” and a violation of their perceptions of what makes
find direct effects of benevolent sexism on attitudes toward the EE women different from men (i.e., warm and nurturing). As a result,
policy, we found evidence for indirect effects of benevolent sexism EE policies seeking to place women in traditionally male roles
on attitudes via compassion. should be less likely to evoke compassion from benevolent sexists
To this point, we have outlined how benevolent sexism can lead because such policies are perceived as assisting competitive, in-
to increased support for EE policies by inducing experiences of dependent women—that is, women who are not warm and nurtur-
compassion. On the one hand, these findings support our conten-
ing. Thus, while we have argued that benevolent sexists are likely
tion that a more complex and nuanced view of sexism—specifi-
to experience compassion when considering EE policies in gen-
cally, accounting for its two forms—is required to truly understand
eral, when considering an EE policy that promotes women into
how sexist attitudes influence support for EE policies. On the other
masculine roles benevolent sexists are unlikely to experience that
hand, these findings seemingly fly in the face of the rather dismal
sense of compassion because such women violate the necessary
statistics we report at the beginning of this article regarding female
representation in traditional positions of power and influence, as conditions for experiencing compassion.
well as ambivalent sexism theory’s contention that benevolent Notably, the EE policy used in Study 1 and 2 was in reference
sexism is ultimately harmful to women. In other words, if benev- to students’ co-op job placements in general. That is, it was not
olent sexism is widespread (as suggested by past research, i.e., specified in the EE policy whether the employment of women was
Becker, 2010; Glick & Fiske, 2001) and it seemingly encourages encouraged in more male or female stereotyped positions. Al-
EE policies designed to advance women’s employment prospects, though such cues as to the type of position were absent, given
why are so few women advancing to the highest levels of organi- co-op positions are typically entry level positions it is unlikely that
zations? our study participants viewed the EE policy as being designed to
As one solution for this conundrum, consider also that be- place women in powerful or high status positions—positions pre-
nevolent sexist ideology holds that women are warm, but are dominantly perceived as masculine jobs (Glick, Wilk, & Perreault,
also the “weaker” sex. Benevolent sexists view women as being 1995). Thus, it is reasonable to assume that in the absence of
fragile and in need of assistance; a consequence of this view- explicit statements of the type of job, participants may not have
viewed co-op positions as very masculine.9 However, in Study 3, masculine. Based on these findings, we manipulated whether a
we sought to more explicitly manipulate whether the EE policy gender-based EE policy for students’ co-op hiring promoted the
promoted the hiring of women in more masculine jobs or more hiring of female students in masculine or feminine positions, by
feminine jobs to establish a possible boundary effect of our find- using finance and human resource management positions, respec-
ings. tively.
In addition to testing this prediction, in Study 3 we also sought Measures. Using 7-point Likert response scales ranging from
to address a limitation of our compassion measure in Study 1 and 1 (strongly disagree) to 7 (strongly agree), we used the same
2, in that it did not indicate the target of the compassion felt (i.e., measure to assess benevolent sexism (␣ ⫽ .76) and hostile sexism
toward women). In Study 3, we address this issue by referencing (␣ ⫽ .87) as in Study 1, and the same measures to assess attitudes
the target of the compassion in our compassion measure. Finally, (␣ ⫽ .91) and behavioral intentions (␣ ⫽ .90) as in Study 1 and 2.
given that in Study 1 and Study 2 we used a one-item measure of To assess compassion, participants rated to what degree they feel
unknown reliability when ruling out sadness as an alternate expla- or possess “compassion,” “tender feelings,” and “sympathy” to-
nation for our findings, in Study 3 we use a more psychometrically ward women at the present moment on a 5-point Likert scale
sound three-item measure of sadness. ranging from 1 (very slightly or not at all) to 5 (extremely; ␣ ⫽ .90;
Oveis et al., 2010). To assess sadness, participants rated to what

Study 3: Benevolent Sexism and Job Type extent they felt “sad,” “blue,” and “downhearted” (␣ ⫽ .95;
Watson & Clark, 1994) using the same 5-point Likert scale as
Method compassion. To assess whether participants in the finance condi-
tion perceived the co-op positions as more masculine and less
Participants and procedure. Participants were 90 business feminine than participants in the human resource management
undergraduate students (33 men and 57 women; age: M ⫽ 20.41, condition, participants rated whether the co-op positions for which
SD ⫽ 1.00) at a midsize Canadian university who received course the EE policy would apply are seen as “masculine” and “feminine”
research credit for participation.10 Forty-four participants identi- on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly
fied as Caucasian, 15 as East Asian, 11 as South Asian, three as agree).
Southeast Asian, two as West Indians, one as Middle Eastern, and
five as mixed (nine unreported). The procedure consisted of two
parts. In the first part, participants completed an online survey Results
assessing their benevolent and hostile sexist attitudes and demo-
Preliminary analyses. Means and standard deviations of all
graphics. In the second part, 2 weeks later, in an online survey
measures are presented in Table 4. In line with past research and
participants were randomly assigned to read one of the two ver-
our Study 1 results, benevolent sexism and hostile sexism were
sions of a gender-based EE policy that was ostensibly being
positively correlated, r ⫽ .41, p ⬍ .001. Thus, to parse out unique
developed at the participants’ university and were told that their
effects of benevolent sexism on support for the EE policy, we
input is sought as the policy is relevant to them. In one version, the
controlled for hostile sexism in our hypotheses tests.
EE policy promoted the hiring of female students in more mascu-
Before proceeding to test hypotheses, we examined gender
line co-op jobs. In another version, the EE policy promoted the
differences in a) endorsement of sexist attitudes, and b) support for
hiring of female students in more feminine co-op jobs (see below
gender-based EE policies. Consistent with past research and our
for details on position type manipulation). Following the presen-
Study 1 results, men endorsed more hostile sexism (M ⫽ 4.20,
tation, the participants first completed a measure assessing their
SD ⫽ 0.70) than women (M ⫽ 3.53, SD ⫽ 0.99), t(88) ⫽ 3.44, p ⬍
state experience of compassion toward women, and then measures
.001 (d ⫽ 0.78). Also, in line with past research that suggests that
assessing their attitudes and behavioral intentions to promote the
gender gap in benevolent sexism is smaller than for hostile sexism
proposed gender-based EE policy. The data from the two surveys
(Glick & Fiske, 1996), there was only a marginal difference
were matched using student identification numbers, which partic-
between men’s (M ⫽ 4.09, SD ⫽ 0.67) and women’s (M ⫽ 3.76,
ipants provided in both surveys.
Position type manipulation. We used the same gender-based
EE policy as in Study 1 and 2, but we added and manipulated 9
In a separate sample of 298 students, we tested whether co-op positions
information about the position type the EE policy was specifically in general (i.e., with no specific industry mentioned) were seen as mascu-
targeting. In one version the policy applied to more masculine line on a 7-point Likert scale ranging from 1 (strongly disagree) to 7
(strongly agree), compared to typical masculine co-op positions (i.e.,
positions; in another version the policy applied to more feminine finance). As expected, we found that general co-op positions were seen as
positions. To determine what co-op positions students perceived as less masculine (M ⫽ 4.41, SD ⫽ 1.25) than finance co-op positions (M ⫽
more masculine as opposed to more feminine, we conducted a pilot 5.26, SD ⫽ 1.37), t(296) ⫽ 11.46, p ⬍ .001.
test with a separate sample of 138 undergraduate business students
10
Our power analysis using GⴱPower software (Faul et al., 2007) with
(76 men and 62 women). We tested to what degree co-op positions an effect size of f 2 ⫽ 0.08 (the average effect size obtained in Study 1) with
a significance level of ␣ ⫽ .05 and 80% power would be 101. However,
in different business areas (finance, accounting, marketing, and due to limits of our subject pool, we only managed to recruit 90 students
human resource management)11 are perceived to be masculine and and thus have an estimated power of 76% to detect two-way interactions
feminine. Our pilot study (please see the online Supplemental between benevolent sexism and job type. We address this lower power
Materials for complete results) showed that undergraduate busi- issue in Study 4 with a larger sample size.
11
Students in our samples were business undergraduate students, and
ness students perceived finance positions to be the most masculine their four areas of specialization included finance, accounting, marketing,
and the least feminine; whereas they perceived human resource and human resource management. As such, their co-op experiences were
management positions to be the most feminine and the least mostly in these four areas of business.
Table 4
Human resource management positions Finance positions

Benevolent sexism 4.09 (.72) 3.75 (.72) 3.86 (.73) 4.09 (.65) 3.78 (.98) 3.91 (.86)
Hostile sexism 4.48 (.66) 3.49 (1.01) 3.79 (1.02) 3.97 (.66) 3.59 (.97) 3.76 (.86)
Compassion 2.87 (1.06) 3.04 (.93) 2.99 (.97) 2.59 (.97) 3.28 (.94) 2.98 (1.01)
Attitudes 3.96 (1.60) 5.42 (.77) 4.97 (1.27) 3.91 (1.08) 5.36 (1.36) 4.72 (1.43)
Sadness 1.26 (.54) 1.78 (.98) 1.62 (.79) 1.54 (.62) 1.99 (1.33) 1.79 (1.09)
Note. Standard deviations are presented in parentheses. For compassion and sadness, the scale ranged from 1 to 5; for benevolent and hostile sexism,
attitudes, and behavioral intentions, the scale ranged from 1 to 7.
SD ⫽ 0.82) endorsement of benevolent sexism, t(88) ⫽ 1.96, p ⫽ and position type given the strong positive correlation between
.05 (d ⫽ 0.44). benevolent and hostile sexism.12 In the first step of the regression
Consistent with past research on reactions to EE policies and our we entered control variables: gender, hostile sexism (centered),
Study 1 and 2 results, women, compared to men, had more and the interaction term between hostile sexism and position type.
favorable attitudes (women: M ⫽ 5.39, SD ⫽ 1.04; men: M ⫽ In the second step, we entered main effects of benevolent sexism
3.93, SD ⫽ 1.32), t(88) ⫽ ⫺5.82, p ⬍ .001 (d ⫽ 1.23), and (centered) and position type. In the third step, we entered the
endorsed more behavioral intentions promoting the EE policy interaction between benevolent sexism and position type.
(women: M ⫽ 4.27, SD ⫽ 1.23; men: M ⫽ 2.63, SD ⫽ 1.30), In predicting compassion toward women, there was a significant
t(88) ⫽ ⫺5.99, p ⬍ .001 (d ⫽ 1.30). As such, we controlled for interaction between benevolent sexism and position type,
gender in our analyses. It should be noted that the conclusions of b ⫽ ⫺.68, t(83) ⫽ ⫺2.41, p ⫽ .018 (f 2 ⫽ 0.07; see Table 5 for a
our analyses and hypotheses tests remain the same regardless of full regression model). To interpret the interaction, we graphed it
whether we include or exclude gender in our analyses. at high (⫹1 SD) and low (⫺1 SD) levels of benevolent sexism. In
We also tested for potential condition differences between mas- line with our predictions, a simple slope analysis revealed that the
culine and feminine positions on benevolent and hostile sexism. As more individuals endorsed benevolent sexism the more compas-
expected, there were no condition differences on either benevolent sion toward women they experienced when the EE policy pro-
sexism, t(88) ⫽ .51, p ⫽ .761, or hostile sexism, t(88) ⫽ .18, p ⫽ moted the hiring of women in human resource management posi-
.857. tions, b ⫽ .71, t(83) ⫽ 3.28, p ⫽ .001; but not in finance positions,
Manipulation checks. Participants in the finance condition, b ⫽ .03, t(83) ⫽ .17, p ⫽ .868 (see Figure 3). We further examined
as expected, found the co-op position to be more masculine (M ⫽ differences in compassion toward women when the EE policy
5.22, SD ⫽ 1.47) than participants in the human resource man- promoted the hiring of women in human resource management
agement condition (M ⫽ 3.51, SD ⫽ 1.34), t(88) ⫽ ⫺5.76, p ⬍ versus finance positions within high and low levels of benevolent
.001 (d ⫽ 1.22). Also as expected, participants in the human sexism. As expected, individuals high on benevolent sexism ex-
resource management condition found the co-op positions to be perienced marginally more compassion toward women when the
more feminine (M ⫽ 5.25, SD ⫽ 1.05) than participants in the EE policy promoted the hiring of women in human resource
finance condition (M ⫽ 3.63, SD ⫽ 1.20) t(88) ⫽ 6.71 p ⬍ .001 management positions compared to finance positions, b ⫽ .50,
(d ⫽ 1.44). In addition, we tested whether benevolent sexism t(83) ⫽ ⫺1.68, p ⫽ .097. We found the opposite effect for
moderated the success of our manipulation check and we found no individuals low on benevolent sexism: they experienced margin-
significant interactions between benevolent sexism and position ally less compassion toward women when the EE policy promoted
type in predicting either how masculine, b ⫽ ⫺.11, t(83) ⫽ ⫺.25, the hiring of women in human resource management positions
p ⫽ .800 (f 2 ⬍ 0.001), or feminine, b ⫽ .22, t(83) ⫽ .68, p ⫽ .501 compared to finance positions, b ⫽ .56, t(83) ⫽ 1.89, p ⫽ .062.
(f 2 ⫽ 0.005), the job types were perceived. Thus, our manipulation In predicting attitudes, there was a significant interaction be-
of masculine and feminine job type by using finance and human tween benevolent sexism and position type, b ⫽ ⫺.99,
resource management positions, respectively, was successful. t(83) ⫽ ⫺2.98, p ⫽ .004 (f 2 ⫽ 0.11; see Table 5). As expected, the
The moderating effect of job type. We predicted that the more individuals endorsed benevolent sexism the more favorable
positive effect of benevolent sexism on compassion and attitudes attitudes toward the EE policy they had when the EE policy
and behavioral intentions that promote a gender-based EE policy promoted the hiring of women in human resource management
would extend to EE policies that promote the hiring of women in positions, b ⫽ .67, t(83) ⫽ 2.65, p ⫽ .010, but not in finance
feminine (i.e., human resource management) positions, but not in
masculine (i.e., finance) positions. To that end, we conducted
12
hierarchical moderated regression analyses testing for an interac- Without controlling for hostile sexism and the interaction between
tion between benevolent sexism and position type in predicting hostile sexism and position type, our interactions became marginal or
nonsignificant (.079 ⱖ p ⱕ .125). However, in Study 4 (with a larger
compassion toward women, and attitudes and behavioral inten- sample size), controlling for hostile sexism and the interaction between
tions. In line with Study 1 and 2, we controlled for hostile sexism hostile sexism and position type did not influence the significance of our
and gender, as well as for the interaction between hostile sexism results.
Table 5
Regression Analyses Result: Benevolent Sexism ⫻ Position Type Interaction in Predicting Compassion Toward Women, Attitudes, and
Behavioral Intentions That Promote the Gender-Based Employment Equity Policy (Study 3)
Compassion toward women Attitudes Behavioral intentions

Predictors b/␤ 95% CI b/␤ 95% CI b/␤ 95% CI
2 ⴱⴱⴱ ⴱⴱⴱ
Step 1 R .05 .31 .29
Gender .43/.21† [⫺.03, .89] 1.31/.47ⴱⴱⴱ [.78, 1.85] 1.59/.52ⴱⴱⴱ [.99, 2.19]
Hostile sexism ⫺.02/⫺.02 [⫺.32, .28] ⫺.33/⫺.23† [⫺.67, .01] ⫺.10/⫺.07 [⫺.48, .28]
Hostile Sexism ⫻ Position .13/.07 [⫺.33, .59] .33/.14 [⫺.21, .87] .08/.03 [⫺.52, .68]
Step 2 ⌬R2 .05† .003 .01
Benevolent sexism .31/.25ⴱ [.03, .59] .09/.06 [⫺.25, .43] .20/.11 [⫺.18, .88]
Position .03/.02 [⫺.39, .45] ⫺.08/⫺.03 [⫺.58, .42] .18/.06 [⫺.36, .72]
Step3 ⌬R2 .06ⴱ .07ⴱⴱ .03ⴱ
Benevolent Sexism ⫻ Position ⫺.68/⫺.40ⴱ [⫺1.24, ⫺.12] ⫺.99/⫺.42ⴱⴱ [⫺1.65, ⫺.33] ⫺.77/⫺.30ⴱⴱ [⫺1.53, ⫺.01]
Total R2 .16ⴱⴱ .38ⴱⴱ .33ⴱⴱⴱ

Note. CI ⫽ confidence interval. Gender is coded as 0 ⫽ men and 1 ⫽ women. Position is coded as 0 ⫽ human resource management and 1 ⫽ finance.
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
positions, b ⫽ ⫺.32, t(83) ⫽ ⫺1.48, p ⫽ .144 (see Figure 4). human resource management positions compared to finance positions,
Further, as expected, individuals high on benevolent sexism had b ⫽ .77, t(83) ⫽ 1.96, p ⫽ .054.
more favorable attitudes toward the EE policy promoting the Moderated mediation. Finally, we proposed an overall mod-
hiring of women in human resource management positions com- erated mediation model (see Figure 6) in which the positive effect
pared to finance positions, b ⫽ ⫺.87, t(83) ⫽ ⫺2.46, p ⫽ .016. We of benevolent sexism on attitudes and behavioral intentions via
also found that individuals low on benevolent sexism had margin- compassion extends to EE policies that promote the hiring of
ally less favorable attitudes toward the EE policy promoting the women in feminine positions, but not in masculine positions. To
hiring of women in human resource management positions com- assess this hypothesis, we used Hayes’s (2013) PROCESS macro
pared to finance positions, b ⫽ .68, t(83) ⫽ 1.95, p ⫽ .054. (Model 8), which tests for moderated mediation using two regres-
In predicting behavioral intentions, there was a significant interac- sion models and bias-corrected bootstrapping technique (with
tion between benevolent sexism and position type, b ⫽ ⫺.77, 10,000 samples) to compute conditional indirect effects. The first
t(83) ⫽ ⫺2.05, p ⫽ .043 (f 2 ⫽ 0.05; see Table 5). As expected, the regression model estimates the interaction between benevolent
more individuals endorsed benevolent sexism the more behavioral sexism and position type in predicting compassion toward women
intentions that promote the EE policy they endorsed when the EE (we also controlled for gender, hostile sexism, and the interaction
policy promoted the hiring of women in human resource management between hostile sexism and position type in our moderated medi-
positions, b ⫽ .66, t(83) ⫽ 2.28, p ⫽ .025; but not in finance ation analyses). The second regression model estimates the effect
positions, b ⫽ ⫺.12, t(83) ⫽ ⫺.48, p ⫽ .632 (see Figure 5). However, of compassion toward women on support for the EE policy (atti-
contrary to our expectations, individuals high on benevolent sexism tudes and behavioral intentions) while controlling for the effect of
did not endorse more behavioral intentions supporting the EE policy benevolent sexism, position type, and their interaction.
promoting the hiring of women in human resources management In the first regression, as described above, there was a signifi-
positions compared to finance positions, b ⫽ ⫺.44, t(83) ⫽ ⫺1.09, cant interaction between benevolent sexism and position type in
p ⫽ .278. In line with findings for compassion and attitudes, individ- predicting compassion toward women. In the second regression,
uals low on benevolent sexism endorsed marginally fewer behavioral there was a significant effect of compassion toward women on
intentions supporting the EE policy promoting the hiring of women in both attitudes, b ⫽ .63, t(82) ⫽ 5.75, p ⬍ .001, and behavioral
5
7
Attitudes toward the EE Policy
Compassion toward Women
Human Resource 6
4
Management Human Resource
5 Management
Positions
3 Positions
4
Finance Positions
Finance positions
3
2
2
1 1
Low Benevolent High Benevolent Low Benevolent High Benevolent
Sexism Sexism Sexism Sexism
Figure 3. An interaction between benevolent sexism and position type in Figure 4. An interaction between benevolent sexism and position type in
predicting compassion toward women (Study 3). predicting attitudes toward the employment equity (EE) policy (Study 3).
intentions, b ⫽ .55, t(82) ⫽ 4.10, p ⬍ .001. Finally, the conditional

indirect effect was significant for human resource management
positions when predicting both attitudes (conditional indirect ef-
fect ⫽ .45, 95% CI [.18, .83]) and behavioral intentions (condi-
tional indirect effect ⫽ .39, 95% CI [.14, .48]), but not for finance
positions (conditional indirect effect for attitudes ⫽ .02, 95% CI
[⫺.25, .33]; conditional indirect effect for behavioral intentions ⫽
.02, 95% CI [⫺.22, .29]). Thus, benevolent sexism was related to
favorable attitudes and higher endorsement of behavioral inten-
tions that promote the gender-based EE policy via compassion
toward women only when the policy promoted the hiring of
women in human resource management positions but not in fi- Figure 6. The tested moderated mediation model in Study 3 and 4. EE ⫽
nance positions. employment equity.
Additional analyses. In line with Study 1 and Study 2, to rule

out the possibility that sadness underlies the effect of benevolent

jobs). While this contributes to the external validity of our find-
sexism and position type on support for a gender-based EE policy,
ings, it may also have influenced our results because both male and
we tested for an interaction between benevolent sexism and posi-
female participants may have viewed the EE policy as harming or
tion type in predicting feelings of sadness. As expected, there was
helping their job prospects. This level of self-interest may color
no significant interaction between benevolent sexism and position
reactions to the EE policy in unanticipated ways. To address this
type in predicting sadness, b ⫽ ⫺.29, t(83) ⫽ ⫺1.04, p ⫽ .301
concern, in Study 4 we test the boundary effect of job type in an
(f 2 ⫽ 0.01). This result indicates that sadness did not underlie the
employee sample that had no personal stakes in a proposed EE
effect of benevolent sexism and position type on support for the
policy. Further, in Study 4 we also extend our investigation to
EE policy. Although we do not have sufficient power to reliably
additional masculine and feminine jobs to increase our stimulus
test whether participant gender qualified our two-way interactions,
sampling, that is, using multiple instances of a stimulus category (in
we address this issue in Study 4.
our case masculine and feminine job types). By doing so, we increase
the external and construct validity of our findings (Wells & Wind-
Discussion schitl, 1999). Study 4 also seeks to replicate the findings of Study 3,
Study 3 demonstrated that the positive effect of benevolent an important step in establishing the robustness of our effects (Nosek,
sexism on support for gender-based EE policies extends only to EE Spies, & Motyl, 2012). Finally, although Study 3 found no evidence
policies that promote the hiring of women in feminine roles. In for three-way interactions between benevolent sexism, job type, and
particular, Study 3 provided empirical evidence for a moderated participant gender, one possible explanation raised by reviewers is
mediation model in which greater benevolent sexism elicits greater that Study 3 was not sufficiently powered to detect such interactions.
compassion toward women, which in turn leads to more favorable Thus, in Study 4 we sought to recruit a larger sample to ensure
attitudes and to a higher endorsement of behavioral intentions that adequate power to examine such interactions.
support the EE policy, but only when the policy promoted the
hiring of women in feminine (human resource management), but Study 4: Additional Evidence for a Boundary
not masculine (finance), positions. Effect of Job Type
To this point, we have tested the effect of benevolent sexism on
support for EE policies in samples of participants who would be Method
directly influenced by the proposed EE policy (i.e., co-op job
applicants where the EE policy applied to their potential co-op Participants and procedure. Participants were 713 U.S. em-
ployees (335 women, 378 men; age: M ⫽ 34.43, SD ⫽ 9.69)13
7 13
We recruited our sample for Study 4 in two waves via MTurk. Our
Behavioral Intentions
6 initial sample consisted of 223 employees. While this sample had sufficient
Human Resource power to detect predicted two-way interactions, it did not have adequate
Management power to detect three-way interactions. Using GⴱPower software, we esti-
5 Positions
mated that the sample size needed for 80% power to detect three-way
4 interactions with a significance level of ␣ ⫽ .05 and an effect size of f 2 ⫽
Finance Positions 0.009 (i.e., the average effect size obtained in Study 3 for our three-way
3 interactions) would be 875. As such, we recruited an additional 600
participants; however, 110 participants did not correctly answer two open-
2 ended manipulation check items, that is, “What was the name of the policy
you read about?” and “What was the main goal of the policy you read
1 about?” These participants were thus excluded from analyses. Thus, our
Low Benevolent High Benevolent final sample was 713, which provided 72% power to detect three-way
Sexism Sexism
interactions with an average effect size of f 2 ⫽ 0.009. Although somewhat
below the ideal 80% power threshold, this sample size does provide us with
Figure 5. An interaction between benevolent sexism and position type in enough participants to detect a three-way interaction when predicting
predicting behavioral intentions that promote the employment equity pol- attitudes based on the effect size in Study 3 (80% power for f 2 ⫽ 0.02,
icy (Study 3). meaning 395 participants would be needed).
recruited using Amazon’s Mechanical Turk (MTurk), an online 1, 2, and 3. Using a 5-point Likert scale ranging from 1 (very
platform for web-based survey and experimental data collections slightly or not at all) to 5 (extremely), we used the same measure
(Paolacci & Chandler, 2014). Five hundred fifty-nine participants of compassion (␣ ⫽ .90) and sadness (␣ ⫽ .95) as in Study 3. Our
identified as Caucasian, 45 as African American, 37 as Hispanic, manipulation check consisted of one item assessing to what degree
22 as East Asian, 11 as Southeast Asian, seven as South Asian, the job position for the EE policy was seen as masculine or
four as Middle Eastern, and 20 as mixed (eight unreported). feminine on a 7-point scale ranging from 1 (feminine) to 7 (mas-
Participants completed an online survey posted on MTurk and culine).
were compensated $2.00 for their participation. Participants were
informed that they would complete two unrelated studies. In the
Results
first study, participants completed a benevolent and hostile sexism
measure. In the second study, participants were randomly pre- Preliminary analyses. Means and standard deviations of all
sented with one of four versions of a gender-based EE policy that measures are presented in Table 6. In line with past research and
would ostensibly be implemented at a company not related to the the results of Study 1 and 3, benevolent sexism and hostile sexism
participants. The EE policy promoted the hiring of women in either were positively correlated, r ⫽ .38, p ⬍ .001. Men also endorsed
more masculine positions (depending on the version, either a more hostile sexism (M ⫽ 3.49, SD ⫽ 1.27) than women (M ⫽
finance manager or a corporate sales manager job) or more fem- 2.94, SD ⫽ 1.37), t(711) ⫽ 5.62, p ⬍ .001 (d ⫽ 0.42), as well as
inine positions (depending on the version, either a human resource more benevolent sexism (men: M ⫽ 3.67, SD ⫽ 1.21; women:
manager or a customer service representative job; see below for M ⫽ 3.32, SD ⫽ 1.27), t(711) ⫽ 3.75, p ⬍ .001 (d ⫽ 0.28).
details on position type manipulations). Following the presentation Finally, consistent with past research and our Study 1, 2, and 3
of the EE policy, participants completed the same measures as in results, women, compared to men, had more favorable attitudes
Study 3. (women: M ⫽ 5.39, SD ⫽ 1.35; men: M ⫽ 4.82, SD ⫽ 1.40),
Position type manipulation. As in Study 3, we embedded our t(711) ⫽ ⫺5.51, p ⬍ .001 (d ⫽ 0.42), and endorsed more behav-
position type manipulation in the gender-based EE policy. The ioral intentions promoting the EE policy (women: M ⫽ 4.54, SD ⫽
proposed policy, however, did not apply to participants’ own job 1.67; men: M ⫽ 3.85, SD ⫽ 1.72), t(711) ⫽ ⫺5.36, p ⬍ .001 (d ⫽
search and hiring (as was the case in our previous studies), but 0.41). As such, we controlled for hostile sexism, gender, and the
rather was framed as being for the hiring process in a company not interaction between hostile sexism and position type in our anal-
related to them called INDSCO (this name was taken from previ- yses, which is also in line with our analyses in previous studies.
ous research on EE using a company not familiar to participants; Manipulation checks. As expected, the masculine position of
James, Brief, Dietz, & Cohen, 2001). There were four versions of finance manager was perceived as more masculine (M ⫽ 4.16,
the EE policy, with two versions proposing the hiring of women in SD ⫽ 1.07) than either of the two feminine positions: human
masculine positions—that is, finance manager or corporate sales resource manager (M ⫽ 3.73, SD ⫽ 1.14), t(709) ⫽ 3.65, p ⫽ .002
manager—and two versions proposing the hiring of women in (d ⫽ 0.39), or customer service representative (M ⫽ 3.54, SD ⫽
feminine positions—that is, human resource manager or customer 1.10), t(709) ⫽ 5.31, p ⬍ .001 (d ⫽ 0.57). The finance manager
service representative. These four positions were chosen because position was perceived as marginally less masculine than the other
past research by Johnson, Podratz, Dipboye, and Gibbons (2010) masculine position, corporate sales manager, (M ⫽ 4.45, SD ⫽
has identified these positons as masculine and feminine, respec- 1.13), t(709) ⫽ ⫺2.42, p ⫽ .073 (d ⫽ 0.26). The position of
tively (please see the online Supplemental Materials for pilot study corporate sales manager was perceived as more masculine than
results testing masculinity and femininity perceptions for these either the human resource manager, t(709) ⫽ 6.10, p ⬍ .001 (d ⫽
four positions). 0.63); or the customer service representative, t(709) ⫽ 7.79, p ⬍
Measures. Using 7-point Likert response scales ranging from .001 (d ⫽ 0.82). Further, the difference in masculinity perceptions
1 (strongly disagree) to 7 (strongly agree), we used the same between the human resource manager and the customer service
measure to assess benevolent sexism (␣ ⫽ .91) and hostile sexism representative was not significant, t(709) ⫽ 1.61, p ⫽ .377 (d ⫽
(␣ ⫽ .94) as in Study 1 and 3 and the same measures to assess 0.17). Given that the differences in masculinity perceptions be-
attitudes (␣ ⫽ .93) and behavioral intentions (␣ ⫽ .94) as in Study tween the two masculine positions (finance manager and corporate
Table 6
Feminine positions Masculine positions

Benevolent sexism 3.60 (1.25) 3.35 (1.21) 3.48 (1.24) 3.74 (1.17) 3.29 (1.34) 3.53 (1.27)
Hostile sexism 3.59 (1.30) 2.84 (1.31) 3.24 (1.36) 3.39 (1.24) 3.03 (1.43) 3.23 (1.34)
Compassion 2.77 (1.06) 3.16 (1.07) 2.95 (1.08) 2.82 (1.05) 3.16 (1.16) 2.98 (1.11)
Attitudes 4.83 (1.38) 5.51 (1.25) 5.15 (1.36) 4.81 (1.42) 5.26 (1.44) 5.03 (1.44)
Sadness 1.30 (.69) 1.25 (.60) 1.28 (.65) 1.39 (.83) 1.20 (.52) 1.30 (.71)
Note. Standard deviations are presented in parentheses. For compassion and sadness, the scale ranged from 1 to 5; for benevolent and hostile sexism,
attitudes, and behavioral intentions, the scale ranged from 1 to 7.
sales manager) were only marginal (and that in a separate pilot ginally significant, b ⫽ .36, t(703) ⫽ 1.93, p ⫽ .054 (f 2 ⫽ 0.005;
study reported in the online Supplemental Materials were not see Table 7). To follow-up on this marginal three-way interaction,
significant), and the differences in masculinity perceptions be- we tested two-way interactions between benevolent sexism and
tween the two feminine positions (human resource manager and position type for men and women. A two-way interaction between
customer service representative) were nonsignificant, we collapsed benevolent sexism and position type was significant for men,
data within each position type. We used this dichotomous variable b ⫽ ⫺.41, t(703) ⫽ ⫺3.10, p ⫽ .002, but not for women,
(masculine vs. feminine job positions) in our subsequent analyses. b ⫽ ⫺.05, t(703) ⫽ ⫺.38, p ⫽ .703. Following up on the
Further, as in Study 3 we found no significant interactions significant two-way interaction for men, simple slope analyses
between benevolent sexism and position type in predicting our showed that there was a significant positive relation between
manipulation check, b ⫽ ⫺.10, t(706) ⫽ ⫺1.45, p ⫽ .147 (f 2 ⫽ benevolent sexism and endorsement of behavioral intentions pro-
0.003). In line with Study 3, we also tested for potential differences moting the hiring of women in feminine positions, b ⫽ .56,
between masculine and feminine positions on benevolent and t(703) ⫽ 6.21, p ⬍ .001, but not in masculine positions, b ⫽ .15,
hostile sexism. As expected, there were no condition differences t(703) ⫽ 1.52, p ⫽ .128 (see Figure 9 for a complete three-way
on either benevolent sexism, t(711) ⫽ ⫺.52, p ⫽ .607, or hostile interaction). Further, men high on benevolent sexism endorsed
sexism, t(711) ⫽ .12, p ⫽ .904. more behavioral intentions supporting the EE policy promoting the
The moderating effect of job type. As in Study 3, we con- hiring of women in feminine versus masculine positions,
ducted hierarchical moderated regression analyses testing for an b ⫽ ⫺1.36, t(703) ⫽ ⫺3.27, p ⫽ .001; there were no such
interaction between benevolent sexism and position type in pre- differences for men low on benevolent sexism, b ⫽ ⫺.33,
dicting compassion toward women, and attitudes and behavioral t(703) ⫽ ⫺.86, p ⫽ .391. Thus, men high on benevolent sexism
intentions. In addition, we added gender as a full factor to test were in particular likely to support an EE policy hiring women in
three-way interactions between benevolent sexism, position type, feminine, but not in masculine, positions (whereas women seemed
and gender. As in Study 3, we also controlled for hostile sexism to support the hiring of women in both positions; see Figure 9).
and the interaction between hostile sexism and position type.14 Although our proposed two-way interaction between benevolent
In predicting compassion toward women, the three-way inter- sexism and position type was significant (see Table 7), it was
action between benevolent sexism, position type, and gender, was qualified by the above described marginally significant three-way
not significant, b ⫽ .09, t(703) ⫽ .80, p ⫽ .426 (f 2 ⫽ 0.001). As interaction.
predicted there was a significant interaction between benevolent Moderated mediation. As in Study 3, we used Hayes’s PRO-
sexism and position type, b ⫽ ⫺.20, t(704) ⫽ ⫺3.23, p ⫽ CESS macro (Model 8) to test the overall moderated mediation
.001(f 2 ⫽ 0.01; see Table 7). In line with our predictions and Study
model (we did not test three-way interactions between benevolent
3 results, the more individuals endorsed benevolent sexism the
sexism, position type, and gender as a part of our moderated
more compassion toward women they experienced when the EE
mediation model given the three-way interaction did not predict
policy promoted the hiring of women in feminine positions, b ⫽
our mediator, compassion). In the first regression, as described
.50, t(704) ⫽ 9.69, p ⬍ .001. Contrary to Study 3 results, we found
above, there was a significant interaction between benevolent
that benevolent sexism was positively related to compassion ex-
sexism and position type in predicting compassion toward women.
perience even in masculine positions, b ⫽ .29, t(704) ⫽ 5.49, p ⬍
In the second regression, there was a significant effect of compas-
.001 (see Figure 7). Further, as expected, individuals high on
sion toward women on both attitudes, b ⫽ .52, t(705) ⫽ 12.23, p ⬍
benevolent sexism experienced more compassion toward women
.001, and behavioral intentions, b ⫽ .75, t(705) ⫽ 14.36, p ⬍ .001.
when the EE policy promoted the hiring of women in feminine
Given that simple slopes were significant for both feminine and
versus masculine positions, b ⫽ –.31, t(704) ⫽ ⫺2.50, p ⫽ .013;
masculine positions in a two-way interaction predicting compas-
there were no such differences for individuals low on benevolent
sion toward women (our mediator in the moderated mediation
sexism, b ⫽ .20, t(704) ⫽ 1.51, p ⫽ .132.
model), we tested conditional indirect effects at high (⫹1 SD) and
In predicting attitudes, the three-way interaction between be-
nevolent sexism, position type, and gender, was not significant, low (⫺1 SD) levels of benevolent sexism. The conditional indirect
b ⫽ .10, t(703) ⫽ .71, p ⫽ .480 (f 2 ⫽ 0.001). As predicted there effect was significant for high benevolent sexism when predicting
was a significant interaction between benevolent sexism and po- both attitudes (conditional indirect effect ⫽ ⫺.29, 95% CI
sition type, b ⫽ ⫺.23, t(704) ⫽ ⫺2.94, p ⫽ .003 (f 2 ⫽ 0.01; see [⫺.56, ⫺.04]) and behavioral intentions (conditional indirect ef-
Table 7). As expected, the more individuals endorsed benevolent fect ⫽ ⫺.42, 95% CI [⫺.80, ⫺.07]), but not for low benevolent
sexism the more favorable attitudes toward the EE policy they had sexism (conditional indirect effect for attitudes ⫽ .01, 95% CI
when the EE policy promoted the hiring of women in feminine [⫺.20, .21]; behavioral intentions ⫽ .01, 95% CI [⫺.29, .30]).
positions, b ⫽ .31, t(704) ⫽ 4.80, p ⬍ .001, but not in masculine Thus, individuals high (and not low) on benevolent sexism were
positions, b ⫽ .08, t(704) ⫽ 1.15, p ⫽ .250 (see Figure 8). Further, more likely to support the EE policy hiring women in feminine
as expected, individuals high on benevolent sexism had more versus masculine positions due to experience of compassion to-
favorable attitudes toward the EE policy promoting the hiring of ward women.
women in feminine versus masculine positions, b ⫽ ⫺.39, Additional analyses. In line with our Study 3 results, there
t(704) ⫽ ⫺2.55, p ⫽ .012; there were no such differences for was no significant interaction between benevolent sexism and
individuals low on benevolent sexism, b ⫽ .18, t(704) ⫽ 1.12, p ⫽
.262. 14
Controlling or not controlling for hostile sexism and the interaction
In predicting behavioral intentions, the three-way interaction between hostile sexism and position type did not materially affect the
between benevolent sexism, position type, and gender was mar- significance of our results.
Table 7
Regression Analyses Result: Benevolent Sexism ⫻ Position Type ⫻ Participant Gender Interaction in Predicting Compassion Toward
Women, Attitudes, and Behavioral Intentions That Promote the Gender-Based Employment Equity Policy (Study 4)
Compassion toward women Attitudes Behavioral intentions

Predictor b/␤ 95% CI b/␤ 95% CI b/␤ 95% CI
2 ⴱⴱⴱ ⴱⴱⴱ ⴱⴱⴱ
Step 1 R .21 .27 .22
Benevolent sexism .29/.33ⴱⴱⴱ [.21, .37] .19/.17ⴱⴱⴱ [.11, .27] .32/.23ⴱⴱⴱ [.22, .42]
Position .01/.004 [⫺.13, .15] ⫺.14/⫺.05 [⫺.32, .04] ⫺.20/⫺.06† [⫺.44, .04]
Gender .27/.12ⴱⴱⴱ [.11, .43] .34/.12ⴱⴱⴱ [.16, .52] .46/.13ⴱⴱⴱ [.22, .70]
Hostile sexism ⫺.36/⫺.44ⴱⴱⴱ [⫺.44, ⫺.28] ⫺.50/⫺.48ⴱⴱⴱ [⫺.60, ⫺.40] ⫺.63/⫺.49ⴱⴱⴱ [⫺.87, ⫺.39]
Hostile Sexism ⫻ Position .01/.005 [⫺.11, .13] ⫺.09/⫺.06 [⫺.23, .05] .10/.05 [⫺.08, .28]
Step 2 ⌬R2 .03ⴱⴱⴱ .01ⴱ .01ⴱ
Benevolent Sexism ⫻ Position ⫺.20/⫺.16ⴱⴱ [⫺.32, ⫺.05] ⫺.23/⫺.15ⴱⴱ [⫺.39, ⫺.07] ⫺.24/⫺.12ⴱ [⫺.50, ⫺.04]
Benevolent Sexism ⫻ Gender ⫺.22/⫺.18ⴱⴱⴱ [⫺.34, ⫺.10] ⫺.01/⫺.01 [⫺.15, .13] ⫺.09/⫺.05 [⫺.27, .09]
Position ⫻ Gender .12/.05 [⫺.17, .41] ⫺.07/⫺.02 [⫺.44, .30] .22/.05 [⫺.23, .67]
Step3 ⌬R2 .001 .001 .004†

Benevolent Sexism ⫻ Position ⫻ Gender .09/.06 [⫺.15, .33] .10/.05 [⫺.15, .35] .36/.14† [⫺.01, .73]
Total R2 .24ⴱⴱⴱ .28ⴱⴱⴱ .23ⴱⴱⴱ
Note. CI ⫽ confidence interval. Gender is coded as 0 ⫽ men and 1 ⫽ women. Position is coded as 0 ⫽ feminine positions and 1 ⫽ masculine positions.
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
position type in predicting sadness, b ⫽ ⫺.05, t(704) ⫽ 1.10, p ⫽ for both feminine and masculine positions, men supported only the
.273. This result indicates that sadness did not underlie the effect EE policy for feminine, and not masculine, positions.
of benevolent sexism and position type on support for the EE There was also one unexpected finding: benevolent sexism was
policy. positively related to the experience of compassion even when the
EE policy promoted the hiring of women in masculine positions.
Discussion This positive relation between benevolent sexism and the experi-
ence of compassion for masculine positions was statistically
Study 4 provides additional evidence for the conditional nature weaker than the positive relation between benevolent sexism and
of the positive effect of benevolent sexism on support for EE the experience of compassion in feminine positions, and stands in
policies by showing that benevolent sexists support the hiring of contrast to the findings of Study 3. One reason for this unexpected
women in feminine, but not masculine, positions. Extending our finding could be due to the fact that the participants in Study 4
Study 3 results, our results held even for individuals who did not were an employed sample with more work experience, and hence
have any personal stakes in the proposed EE policy and across have more exposure to women struggling in the workplace in
multiple types of feminine and masculine positions. Finally, in stereotypically masculine positions. In turn, this may translate to
Study 4 we were able to adequately test any potential effects of feeling more compassion toward women in general, albeit not
participant gender. We did not find a significant three-way inter- translating to more support for EE policies. Alternately, it could be
action between benevolent sexism, position type, and gender when due to the fact that compassion triggered in masculine positions
predicting compassion toward women and attitudes. However, was for different reasons than compassion triggered in feminine
there was a marginal three-way interaction when predicting be- positions. Namely, it could be that participants experienced com-
havioral intentions indicating that benevolently sexist women were passion because they felt that it would be difficult for women to be
likely to endorse behavioral intentions supporting the EE policy in masculine positions (potentially because these positions are too
5 7
Attitudes toward the EE Policy
Compassion toward Women
6
4 Feminine Feminine
Positions 5 Positions
3 4
Masculine Masculine
Positions 3 Positions
2
2
1 1
Low Benevolent High Benevolent Low Benevolent High Benevolent
Sexism Sexism Sexism Sexism
Figure 7. An interaction between benevolent sexism and position type in Figure 8. An interaction between benevolent sexism and position type in
predicting compassion toward women (Study 4). predicting attitudes toward the employment equity (EE) policy (Study 4).
7 have the blatant negative effects that hostile sexism does, yet it
subtly undermines gender equality by only supporting the place-
6 ment of women in gender-appropriate positions, making it even
more dangerous for gender equality as its effects may go unno-
Behavioral Intentions
5 ticed. Similarly, our results suggest that benevolent sexism leads to

(1) Masculine
Positions, Women inaction when it comes to promoting equality and diversity where
4 (2) Masculine it matters the most: where a significant underrepresentation of
Positions, Men
women in traditionally masculine jobs exists.
3 (3) Feminine Positions,
Women
(4) Feminine Positions,
2 Men Contributions to the Literature on Sexist Attitudes
1 Our research contributes to the literature on sexist attitudes in
Low Benevolent High Benevolent several ways. First, in our theorizing we outline the emotional
Sexism Sexism
consequence of benevolent sexism, that is, the experience of
Figure 9. A three-way interaction between benevolent sexism, position compassion, and how the experience of compassion influences
type, and gender in predicting behavioral intentions that promote the consequent support for EE policies. The finding that benevolent
employment equity policy (Study 4). sexism leads to more support for EE policies contributes to recent
research findings suggesting that benevolent sexism is related to
beliefs in gender equality of income and employment opportunities
challenging for women and ultimately they may fail or because (Sibley & Perry, 2010), but more importantly it identifies the
potentially they may not be happy in positions where they will be underlying mechanism. By outlining the experience of compassion
minority). This might have occurred in this sample in particular as the mediating process, our research also leads to the insight
since there was no self-interest implicated (i.e., the EE policy regarding a critical boundary condition of the positive effect of
would not influence their own hiring prospects as it did in the benevolent sexism: compassion experienced by benevolent sexists
previous three studies) and hence they might not have felt threat- is limited only to EE policies that promote the hiring of women in
ened in the masculine positions. Of course, these explanations are more feminine positions that do not violate traditional gender
speculative, and future research is needed to replicate and further roles.
understand this effect. Second, our research contributes to resolving an apparent par-
adox in the sexism literature in which on one hand, benevolent
General Discussion sexism is related to an array of positive and beneficial outcomes
In this paper, we provided a more complete account of how for women such as positive evaluation of women (Glick & Fiske,
sexist attitudes influence support for gender-based EE policies by 1996), and on the other hand is related to detrimental outcomes for
examining the effect of benevolent sexism on support for EE women such as lower gender equality indices at the national level
policies. By drawing on and integrating work on ambivalent sex- (Glick et al., 2000) and providing less challenging developmental
ism and applying it to the literature on reactions to EE policies, we work assignments (King et al., 2012). By identifying the mecha-
show that the previously postulated negative influence of sexism nism underlying reactions of benevolent sexists, it appears that
on support for EE policies is not that simple. Past EE research has seemingly positive reactions of benevolent sexists to women and
uniformly held that sexism should have straightforward negative gender issues are due to the experience of compassion but that
effects on the promotion of gender equality and diversity; how- benevolent sexists only experience compassion if the EE policy
ever, this past research has only conceptualized sexism in terms of places women in traditional gender roles.
negative attitudes toward women and has overlooked that sexist Our findings thus illustrate how benevolent sexists may fuel
attitudes may also be positive, that is, benevolent sexism. Counter gender inequality and gender segregation at the workplace by
to past research on hostile sexism, across four studies we find that encouraging the employment of women in “gender-appropriate”
the effect of benevolent sexism on support for gender-based EE jobs, and not promoting their access to more masculine jobs such
policies can be positive, but the effect is complex. Namely, be- as upper management, leadership, and other positions in typically
nevolent sexism has a positive effect on support for EE policies male domains. In line with this suggestion, a recent study by King
due to experiences of compassion, but only for EE policies that et al. (2012) showed that benevolent sexist attitudes limit women’s
promote the hiring of women in more feminine positions, and not exposure to more challenging developmental assignments in orga-
in more masculine positions. Thus, even the seemingly positive nizations, which in turn may limit women’s opportunities to obtain
effects of benevolent sexism subtly erode gender equality in the high level positions in organizations. Further, this assertion is well
workplace. aligned with past research that suggests that in Western countries,
It is interesting to note that in line with Masser and Abrams’s where there is a high rate of participation of women in the
(2004) finding that benevolent sexism was not associated with workplace, women are disproportionately represented in lower-
evaluations of women applying for managerial (i.e., masculine) level positions such as customer service roles and underrepre-
positions (while hostile sexism was negatively associated with sented in higher-level positions such as managerial positions (Re-
evaluations of women), we found that benevolent sexism was not skin, 1993). Even in traditionally female employment sectors such
associated with support for EE policies promoting women in as call centers, women are disproportionately segregated into mass
masculine positions. Thus, benevolent sexism does not appear to production roles as opposed to mass customizations roles (i.e.,
professional roles) and team leader and manager roles (Scholarios A second limitation is that our samples in our first three studies
& Taylor, 2011). lacked the power to properly test for potential moderating effect of
participants’ gender on the relation between benevolent sexism
and support for EE policies. Nevertheless, in Study 4 we sought to
Contributions to the Literature on Compassion
address this issue with a larger sample size; in line with the
Our work also contributes to the literature on compassion and findings from our first three studies, in Study 4 we also did not find
moral judgment, which suggests that compassion is important for an interaction with gender when predicting compassion toward
moral judgment and action as it makes people attuned to unjusti- women and attitudes. However, there was a marginal three-way
fied harm experienced by others and consequent actions to allevi- interaction between benevolent sexism, position type, and gender
ate that harm and suffering (Goetz et al., 2010; Kaplan & Tivnan, when predicting behavioral intentions indicating that while women
2014; Oveis et al., 2010). While this literature suggests primarily were likely to endorse behavioral intentions promoting the EE
positive and beneficial effects of experiencing compassion, our policy hiring women in both feminine and masculine jobs, men
work suggests that experiences of compassion may also have only supported the EE policy hiring women in feminine job. Yet
negative outcomes. In particular, the experienced compassion of the effect size of this three-way interaction was small, and overall
benevolent sexists contributes to occupational gender segregation our results suggest that both men and women who are benevolent
and limits women’s opportunities, which contributes to gender sexists exhibit the same pattern of support for EE policies: they
inequality. support the hiring of women in feminine and not masculine jobs.
Interestingly, and in line with recent research (Shnabel, Bar- We believe that these findings that gender is not an overly influ-
Anan, Kende, & Bareket, 2015), one possible interpretation of our ential factor are illuminating and important as they show that both
work is that the compassion benevolent sexists experience may men and women may perpetuate the gender segregation in the
lead them to engage in dependency-oriented helping behaviors. workplace and may contribute to pushing women into traditional
Dependency-oriented help refers to providing a final solution to feminine positions (i.e., positions of lower power and status),
those in need rather them providing them with tools to succeed on while failing to help the hiring of women where needed the
their own or remove barriers to success (Nadler, 2015; Nadler & most—in more typical masculine positions (i.e., positions of power
Chernyak-Hai, 2014). Dependency-oriented help thus does not and status).
empower those who are helped, and reinforces their dependent Further, in all of our studies we sought to rule out sadness as an
position and the superiority of helpers. An EE policy may be seen alternate explanation for our results given that pity (an emotion
as a type of dependency-oriented help as EE policies do not invoked by perceptions of high warmth and low competence)
remove barriers (i.e., discrimination) to women’s success, but consists of both compassion and sadness, but our theory argued for
rather provide a final solution by placing women in jobs without compassion specifically as an underlying mechanism. In all four
addressing the root cause of the problem (Jackson & Esses, 2000). studies we showed that compassion and not sadness underlies the
relation between benevolent sexism and support for the EE policy.
However, the lack of findings for sadness may be due to floor
Limitations and Future Directions
effects, as the mean values for sadness were generally low in our
Some limitations of our work should also be noted. First, our studies. Nevertheless, given that we observed a significant relation
operationalization of masculine and feminine jobs may be con- between benevolent sexism and sadness in Study 1 (although
founded with other dimensions. More specifically, the masculine sadness did not underlie the relation between benevolent sexism
and feminine jobs that we tested in our Study 3 and 4 may also and support for the EE policy) it seems there was enough variance
differ in occupational prestige in addition to perceived masculinity in sadness to observe significant correlations, which may make
and femininity traits needed for success in the job. Thus, it may be floor effects potentially less likely as an explanation for the lack of
difficult to discern whether our Study 3 and 4 results were driven results regarding sadness.
by the masculinity-femininity of the jobs or the occupational It should also be noted that in our mediational models we cannot
prestige of the jobs. Past research suggests that gender-specific infer causality of compassion on support for EE policies given that
positions are naturally confounded with occupational prestige, in compassion and support were measured at the same time and from
that masculine jobs are highly and positively correlated with the same source. However, past research in which compassion was
occupational prestige (Glick, 1991; Glick et al., 1995). Thus, an induced provides some casual evidence for the effect of compas-
alternate perspective on our results is that benevolent sexists may sion on helping behaviors and care for others (e.g., Oveis et al.,
be likely to encourage and promote the employment of women in 2010; Piff, Kraus, Côté, Cheng, & Keltner, 2010). Nevertheless,
less prestigious positions, instead of more feminine positions. we encourage additional research manipulating compassion to
Relatedly, an alternative perspective may be that benevolent extend, and provide further support for, our findings.
sexists may be likely to encourage and promote the employment of A final limitation is that our benevolent sexism prime in Study
women in positions where they are the majority (i.e., less presti- 2 may lead to more compassion because it primes ideas about
gious positions), perhaps thinking women might be happier in a protecting and helping others more generally, rather than women
job where they are well represented. Yet regardless of their pos- in particular, which in turn could lead to more support for EE
sible reasons for only supporting EE policies for women in fem- policies. However, based on past research using this manipulation,
inine positions—including possibly having the best of intentions— there is good reason to believe that benevolent sexism is specific
the outcomes are the same: women are segregated into feminine to helping women in particular. For example, Shnabel et al. (2015)
jobs and gender equality is undermined and jeopardized in the showed that exposure to the same benevolent sexism prime as in
workplace. our Study 2 (compared to a control condition) is specifically
related to dependency-oriented helping behaviors toward women, Conclusion

that is, a kind of help that reinforces women’s dependent position;
whereas there were no condition difference (benevolent sexism In this research we examined the effect of a less understood, but
prime vs. control condition) in dependency-oriented helping be- quite prevalent, type of sexism, namely benevolent sexism, on
haviors toward men (see their Study 2b). These findings show that support for gender diversity in the context of gender-based EE
he benevolent sexism prime is related to helping behavior directed policies. We show that, counter to past research showing sexism
toward women and not toward people more generally. undermines gender diversity, one type of sexism, benevolent sex-
With respect to future research directions, although in our paper ism, appears beneficial for gender-based EE policies due to an
we have focused on employee reactions to gender-based EE pol- experience of compassion. However, this positive effect of benev-
icies, EE policies in Canada (and other countries) also promote the olent sexism extends only to EE policies that promote the hiring of
employment of other disadvantaged groups such as racial minor- women in more feminine positions, but not in more masculine
ities, disabled people, and aboriginal people. We expect that the positions. Thus, benevolent sexism ultimately undermines gender
diversity by contributing to gender segmentation in the workplace
theoretical model that we have developed and tested in this paper
and leading to inaction and lack of social change in promoting the

may also apply to benevolent prejudiced attitudes against other
employment of women in roles in which they are underrepre-

groups (i.e., racial minorities). Past work on benevolent prejudice
sented. Overall, our research points to the complex and subtle
has mostly focused on benevolent sexism because the interdepen-
effects of benevolent sexism on promotion of gender diversity and
dence between men and women is far greater than between any
contributes to a more comprehensive account of how sexist atti-
other in-groups and out-groups due to procreation needs (Glick &
tudes that are prevalent and socially accepted in our society may
Fiske, 1996). Yet, past research has also suggested that benevolent
appear beneficial for gender equality while subtly contributing to
prejudiced attitudes also exist regarding other groups such as racial
gender segregation in the workplace.
minorities and aboriginal people (Jackman, 1994; Werhun &
Penner, 2010). Thus, more broadly our work may contribute to
understanding of how benevolent prejudiced attitudes influence References
reactions to minority groups, and future work in this avenue is
encouraged. Allport, G. W. (1954). The nature of prejudice. Oxford, England: Addison
Wesley.
Atkins, P. W. B., & Parker, S. K. (2012). Understanding individual
Practical Implications for Governments compassion in organizations: The role of appraisals and psychological
flexibility. Academy of Management Review, 37, 524 –546. http://dx.doi
and Organizations
.org/10.5465/amr.2010.0490
Our proposed theoretical model offers important practical im- Bakan, D. (1966). The duality of human existence: An essay on psychology
plications for governments that mandate EE polices and for orga- and religion. Oxford, England: Rand McNally.
nizations that strive to successfully implement these policies. The Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social
behavior: Direct effects of trait construct and stereotype-activation on
theoretical model and empirical findings suggest that the effect of
action. Journal of Personality and Social Psychology, 71, 230 –244.
benevolent sexism may be particularly detrimental for promoting
http://dx.doi.org/10.1037/0022-3514.71.2.230
gender diversity in traditionally masculine positions in organiza- Barreto, M., & Ellemers, N. (2005). The burden of benevolent sexism:
tions. Benevolent sexism oftentimes is not perceived as sexism by How it contributes to the maintenance of gender inequalities. European
both men and women (Becker, 2010), and, as such, its effects may Journal of Social Psychology, 35, 633– 642. http://dx.doi.org/10.1002/
go undetected. Moreover, while organizations and managers may ejsp.270
appear to be quite supportive of gender-based EE policies, a more Batson, C. D., & Shaw, L. L. (1991). Evidence for altruism: Toward a
subtle undermining of these policies for traditionally nonfeminine pluralism of prosocial motives. Psychological Inquiry, 2, 107–122.
positions may take place. Given that EE polices and other diversity http://dx.doi.org/10.1207/s15327965pli0202_1
initiatives are designed not only to support higher employment Becker, J. C. (2010). Why do women endorse hostile and benevolent
sexism? The role of salient female subtypes and internalization of sexist
rates for women, but to actually promote higher quality employ-
contents. Sex Roles, 62, 453– 467. http://dx.doi.org/10.1007/s11199-
ments for women (i.e., encouraging the employment of women in 009-9707-4
male dominated, high-status positions), benevolent sexism may Becker, J. C., Glick, P., Ilic, M., & Bohner, G. (2011). Damned if she does,
undermine this goal. damned if she doesn’t: Consequences of accepting versus confronting
A way for organizations and governments to tackle this problem patronizing help for the female target and male actor. European Journal
is to raise awareness about this type of sexism and to communicate of Social Psychology, 41, 761–773. http://dx.doi.org/10.1002/ejsp.823
that even though benevolent sexism does not appear harmful but Becker, J. C., & Swim, J. K. (2012). Differential effects of addressing harm
rather well-meaning it is still very harmful to women in the versus pervasiveness of benevolent sexism. Social Psychology, 43, 127–
workforce. Supporting this notion, recent research has shown that 137. http://dx.doi.org/10.1027/1864-9335/a000091
raising awareness about harmful effects of more subtle forms of Becker, J. C., & Wright, S. C. (2011). Yet another dark side of chivalry:
Benevolent sexism undermines and hostile sexism motivates collective
prejudice such as benevolent sexism can effectively lead to reduc-
action for social change. Journal of Personality and Social Psychology,
tion in endorsement of such attitudes (Becker & Swim, 2012). 101, 62–77. http://dx.doi.org/10.1037/a0022615
Thus, organizations could offer training to managers and other Becker, T. E. (2005). Potential problems in the statistical control of
employees who implement and use EE policies in hiring, promo- variables in organizational research: A qualitative analysis with recom-
tions, and other HR functions to raise awareness about this subtle mendations. Organizational Research Methods, 8, 274 –289. http://dx
yet undermining form of sexism. .doi.org/10.1177/1094428105278021
Becker, T. E., Atinc, G., Breaugh, J. A., Carlson, K. D., Edwards, J. R., & terly, 21, 119 –135. http://dx.doi.org/10.1111/j.1471-6402.1997
Spector, P. E. (2016). Statistical control in correlational studies: 10 .tb00104.x
essential recommendations for organizational researchers. Journal of Glick, P., & Fiske, S. T. (2001). An ambivalent alliance. Hostile and
Organizational Behavior, 37, 157–167. http://dx.doi.org/10.1002/job benevolent sexism as complementary justifications for gender inequal-
.2053 ity. American Psychologist, 56, 109 –118.
Bem, S. L. (1974). The measurement of psychological androgyny. Journal Glick, P., Fiske, S. T., Mladinic, A., Saiz, J. L., Abrams, D., Masser, B., . . .
of Consulting and Clinical Psychology, 42, 155–162. http://dx.doi.org/ López, W. (2000). Beyond prejudice as simple antipathy: Hostile and
10.1037/h0036215 benevolent sexism across cultures. Journal of Personality and Social
Catalyst. (2013a). 2013 Catalyst census: Financial Post 500 women board Psychology, 79, 763–775.
directors. New York, NY: Author. Glick, P., Wilk, K., & Perreault, M. (1995). Images of occupations:
Catalyst. (2013b). 2013 Catalyst census: Fortune 500 women board direc- Components of gender and status in occupational stereotypes. Sex Roles,
tors. New York, NY: Author. 32, 565–582. http://dx.doi.org/10.1007/BF01544212
Chen, Z., Fiske, S. T., & Lee, T. L. (2009). Ambivalent sexism and Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: An
power-related gender-role ideology in marriage. Sex Roles, 60, 765–778. evolutionary analysis and empirical review. Psychological Bulletin, 136,
http://dx.doi.org/10.1007/s11199-009-9585-9 351–374. http://dx.doi.org/10.1037/a0018807

Crosby, F. J., Iyer, A., Clayton, S., & Downing, R. A. (2003). Affirmative Good, J. J., & Rudman, L. A. (2010). When female applicants meet sexist
action. Psychological data and the policy debates. American Psycholo- interviewers: The costs of being a target of benevolent sexism. Sex
gist, 58, 93–115. Roles, 62, 481– 493. http://dx.doi.org/10.1007/s11199-009-9685-6
Crosby, F. J., Iyer, A., & Sincharoen, S. (2006). Understanding affirmative Guttentag, M., & Secord, P. (1983). Too many women? Beverly Hills, CA:
action. Annual Review of Psychology, 57, 585– 611. http://dx.doi.org/10 Sage.
.1146/annurev.psych.57.102904.190029 Harrison, D. A., Kravitz, D. A., Mayer, D. M., Leslie, L. M., & Lev-Arey,
Cuddy, A. J. C., Fiske, S. T., & Glick, P. (2007). The BIAS map: Behaviors D. (2006). Understanding attitudes toward affirmative action programs
from intergroup affect and stereotypes. Journal of Personality and in employment: Summary and meta-analysis of 35 years of research.
Social Psychology, 92, 631– 648. http://dx.doi.org/10.1037/0022-3514 Journal of Applied Psychology, 91, 1013–1036. http://dx.doi.org/10
.92.4.631 .1037/0021-9010.91.5.1013
Dardenne, B., Dumont, M., & Bollier, T. (2007). Insidious dangers of
Hayes, A. F. (2013). Introduction to mediation, moderation, and condi-
benevolent sexism: Consequences for women’s performance. Journal of
tional process analysis: A regression-based approach. New York, NY:
Personality and Social Psychology, 93, 764 –779. http://dx.doi.org/10
Guilford Press.
.1037/0022-3514.93.5.764
Heilman, M. E., & Alcott, V. B. (2001). What I think you think of me:
Delacollette, N., Dumont, M., Sarlet, M., & Dardenne, B. (2013). Benev-
Women’s reactions to being viewed as beneficiaries of preferential
olent sexism, men’s advantage, and the prescription of warmth to wom-
selection. Journal of Applied Psychology, 86, 574 –582. http://dx.doi
en. Sex Roles, 68, 296 –310. http://dx.doi.org/10.1007/s11199-012-
.org/10.1037/0021-9010.86.4.574
0232-5
Heilman, M. E., Block, C. J., & Lucas, J. A. (1992). Presumed incompe-
Devine, P. (1989). Stereotypes and prejudice: Their automatic and con-
tent? Stigmatization and affirmative action efforts. Journal of Applied
trolled components. Journal of Personality and Social Psychology, 56,
Psychology, 77, 536 –544. http://dx.doi.org/10.1037/0021-9010.77.4
5–18. http://dx.doi.org/10.1037/0022-3514.56.1.5
.536
Dutton, J. E., Worline, M. C., Frost, P. J., & Lilius, J. (2006). Explaining
Heilman, M. E., Block, C. J., & Stathatos, P. (1997). The affirmative action
compassion organizing. Administrative Science Quarterly, 51, 59 –96.
stigma of incompetence: Effects of performance information ambiguity.
European Commission. (2014, March). Gender balance on corporate
boards: Europe is cracking the glass ceiling. Retrieved from http://ec Academy of Management Journal, 40, 603– 625. http://dx.doi.org/10
.europa.eu/justice/gender-equality/gender-decision-making/index_en .2307/257055
.htm Heilman, M. E., Simon, M. C., & Repper, D. P. (1987). Intentionally
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). GⴱPower 3: A favored, unintentionally harmed? Impact of sex-based preferential se-
flexible statistical power analysis program for the social, behavioral, and lection on self-perceptions and self-evaluations. Journal of Applied
biomedical sciences. Behavior Research Methods, 39, 175–191. http:// Psychology, 72, 62– 68. http://dx.doi.org/10.1037/0021-9010.72.1.62
dx.doi.org/10.3758/BF03193146 Hideg, I., & Ferris, D. L. (2014). Support for employment equity policies:
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often A self-enhancement approach. Organizational Behavior and Human
mixed) stereotype content: Competence and warmth respectively follow Decision Processes, 123, 49 – 64. http://dx.doi.org/10.1016/j.obhdp
from perceived status and competition. Journal of Personality and .2013.11.002
Social Psychology, 82, 878 –902. http://dx.doi.org/10.1037/0022-3514 Hideg, I., Michela, J. L., & Ferris, D. L. (2011). Overcoming negative
.82.6.878 reactions of nonbeneficiaries to employment equity: The effect of par-
Fiske, S. T., Xu, J., Cuddy, A., & Glick, P. (1999). (Dis)respecting versus ticipation in policy formulation. Journal of Applied Psychology, 96,
(dis)liking: Status and interdependence predict ambivalent stereotypes of 363–376. http://dx.doi.org/10.1037/a0020969
competence and warmth. Journal of Social Issues, 55, 473– 491. http:// Hoffman, C., & Hurts, N. (1990). Gender stereotypes: Perception or
dx.doi.org/10.1111/0022-4537.00128 rationalization. Journal of Personality and Social Psychology, 58, 197–
Glick, P. (1991). Trait-based and sex-based discrimination in occupational 208. http://dx.doi.org/10.1037/0022-3514.58.2.197
prestige, occupational salary, and hiring. Sex Roles, 25, 351–378. Jackman, M. R. (1994). The velvet glove: Paternalism and conflict in
Glick, P., & Fiske, S. T. (1996). The Ambivalent Sexism Inventory: gender, class, and race relations. Berkeley, CA: University of California
Differentiating hostile and benevolent sexism. Journal of Personality Press.
and Social Psychology, 70, 491–512. http://dx.doi.org/10.1037/0022- Jackson, L. M., & Esses, V. M. (2000). Effects of perceived economic
3514.70.3.491 competition on people’s willingness to help empower immigrants.
Glick, P., & Fiske, S. T. (1997). Hostile and benevolent sexism: Measuring Group Processes & Intergroup Relations, 3, 419 – 435. http://dx.doi.org/
ambivalent sexist attitudes toward women. Psychology of Women Quar- 10.1177/1368430200003004006
Jain, H. C., Sloane, P. J., Horwitz, F. M., Taggar, S., & Weiner, N. (2003). social entrepreneurship. Academy of Management Review, 37, 616 – 640.
Employment equity and affirmative action: An international compari- http://dx.doi.org/10.5465/amr.2010.0456
son. Armonk, NY: M. E. Sharpe. Nadler, A. (2015). The other side of the help giving paradigm: Seeking and
James, E. H., Brief, A. P., Dietz, J., & Cohen, R. R. (2001). Prejudice receiving help. In D. A. Schroeder & W. G. Graziano (Eds.), The Oxford
matters: Understanding the reactions of Whites to affirmative action handbook of prosocial behavior (pp. 307–329). New York, NY: Oxford
programs targeted to benefit Blacks. Journal of Applied Psychology, 86, University Press.
1120 –1128. http://dx.doi.org/10.1037/0021-9010.86.6.1120 Nadler, A., & Chernyak-Hai, L. (2014). Helping them stay where they are:
Janiszewski, C., & Wyer, R. S., Jr. (2014). Content and process priming: Status effects on dependency/autonomy-oriented helping. Journal of
A review. Journal of Consumer Psychology, 24, 96 –118. http://dx.doi Personality and Social Psychology, 106, 58 –72. http://dx.doi.org/10
.org/10.1016/j.jcps.2013.05.006 .1037/a0034152
Johnson, S. K., Podratz, K. E., Dipboye, R. L., & Gibbons, E. (2010). Physical Nosek, B. A., Spies, J. R., & Motyl, M. (2012). Scientific utopia: II.
attractiveness biases in ratings of employment suitability: Tracking down Restructuring incentives and practices to promote truth over publishabil-
the “beauty is beastly” effect. The Journal of Social Psychology, 150, ity. Perspectives on Psychological Science, 7, 615– 631. http://dx.doi
301–318. http://dx.doi.org/10.1080/00224540903365414 .org/10.1177/1745691612459058
Jost, J. T., & Kay, A. C. (2005). Exposure to benevolent sexism and Oveis, C., Horberg, E. J., & Keltner, D. (2010). Compassion, pride, and
complementary gender stereotypes: Consequences for specific and dif- social intuitions of self-other similarity. Journal of Personality and
fuse forms of system justification. Journal of Personality and Social Social Psychology, 98, 618 – 630. http://dx.doi.org/10.1037/a0017628
Psychology, 88, 498 –509. http://dx.doi.org/10.1037/0022-3514.88.3 Paolacci, G., & Chandler, J. (2014). Inside the Turk: Understanding the
.498 Mechanical Turk as a participant pool. Current Directions in Psychological
Kanov, J. M., Maitlis, S., Worline, M. C., Dutton, J. E., Frost, P. J., & Lilius, Science, 23, 184 –188. http://dx.doi.org/10.1177/0963721414531598
J. (2004). Compassion in organizational life. American Behavioral Scientist, Piff, P. K., Kraus, M. W., Côté, S., Cheng, B. H., & Keltner, D. (2010).
47, 808 – 827. http://dx.doi.org/10.1177/0002764203260211 Having less, giving more: The influence of social class on prosocial
Kaplan, U., & Tivnan, T. (2014). Multiplicity of emotions in moral behavior. Journal of Personality and Social Psychology, 99, 771–784.
judgment and motivation. Ethics & Behavior, 24, 421– 443. http://dx.doi http://dx.doi.org/10.1037/a0020092
.org/10.1080/10508422.2014.888517 Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of
Kilianski, S. E., & Rudman, L. A. (1998). Wanting it both ways: Do method bias in social science research and recommendations on how to
women approve of benevolent sexism? Sex Roles, 39, 333–352. http:// control it. Annual Review of Psychology, 63, 539 –569. http://dx.doi.org/
dx.doi.org/10.1023/A:1018814924402 10.1146/annurev-psych-120710-100452
King, E. B., Botsford, W., Hebl, M. R., Kazama, S., Dawson, J. F., & Perkins, Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling
A. (2012). Benevolent sexism at work: Gender differences in the distribu- strategies for assessing and comparing indirect effects in multiple me-
tion of challenging developmental experiences. Journal of Management, 38, diator models. Behavior Research Methods, 40, 879 – 891. http://dx.doi
1835–1866. http://dx.doi.org/10.1177/0149206310365902 .org/10.3758/BRM.40.3.879
Konrad, A. M., & Hartmann, L. (2001). Gender differences in attitudes Pyburn, K. M., Jr., Ployhart, R. E., & Kravitz, D. A. (2008). The diversity-
toward affirmative action programs in Australia: Effects of beliefs, validity dilemma: Overview and legal context. Personnel Psychology,
interests, and attitudes toward women. Sex Roles, 45, 415– 432. http:// 61, 143–151. http://dx.doi.org/10.1111/j.1744-6570.2008.00108.x
dx.doi.org/10.1023/A:1014317800293 Reskin, B. (1993). Sex segregation in the workplace. Annual Review of
Konrad, A. M., & Spitz, J. (2003). Explaining demographic group differ- Sociology, 19, 241–270. http://dx.doi.org/10.1146/annurev.so.19
ences in affirmative action attitudes. Journal of Applied Social Psychol- .080193.001325
ogy, 33, 1618 –1642. http://dx.doi.org/10.1111/j.1559-1816.2003 Rudman, L. A., & Glick, P. (2001). Prescriptive gender stereotypes and
.tb01966.x backlash toward agentic women. Journal of Social Issues, 57, 732–762.
Kravitz, D. A., & Platania, J. (1993). Attitudes and beliefs about affirma- http://dx.doi.org/10.1111/0022-4537.00239
tive action: Effects of target and of respondent sex and ethnicity. Journal Scholarios, D., & Taylor, P. (2011). Beneath the glass ceiling: Explaining
of Applied Psychology, 78, 928 –938. http://dx.doi.org/10.1037/0021- gendered role segmentation in call centres. Human Relations, 64, 1291–
9010.78.6.928 1319. http://dx.doi.org/10.1177/0018726711416265
Leslie, L. M., Mayer, D. M., & Kravitz, D. A. (2014). The stigma of Shnabel, N., Bar-Anan, Y., Kende, A., & Bareket, O. (2015). Help to
affirmative action: A stereotyping-based theory and meta-analytic test of perpetuate traditional gender roles: Benevolent sexism increases engage-
the consequences for performance. Academy of Management Journal, ment in dependency-oriented cross-gender helping. Journal of Person-
57, 964 –989. http://dx.doi.org/10.5465/amj.2011.0940 ality and Social Psychology, 110, 55–75.
Linn, P. L., Ferguson, J., & Egart, K. (2004). Career exploration via Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-
cooperative education and lifespan occupational choice. Journal of Vo- experimental studies: New procedures and recommendations. Psycho-
cational Behavior, 65, 430 – 447. http://dx.doi.org/10.1016/j.jvb.2003.10 logical Methods, 7, 422– 445. http://dx.doi.org/10.1037/1082-989X.7.4
.002 .422
Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Sibley, C. G., & Perry, R. (2010). An opposing process model of benev-
Explaining offensive action tendencies in an intergroup context. Journal olent sexism. Sex Roles, 62, 438 – 452. http://dx.doi.org/10.1007/s11199-
of Personality and Social Psychology, 79, 602– 616. http://dx.doi.org/10 009-9705-6
.1037/0022-3514.79.4.602 Son Hing, L. S., Bobocel, D. R., Zanna, M. P., Garcia, D. M., Gee, S. S.,
Mackinnon, D. P., & Fairchild, A. J. (2009). Current directions in media- & Orazietti, K. (2011). The merit of meritocracy. Journal of Personality
tion analysis. Current Directions in Psychological Science, 18, 16 –20. and Social Psychology, 101, 433– 450. http://dx.doi.org/10.1037/
http://dx.doi.org/10.1111/j.1467-8721.2009.01598.x a0024618
Masser, B. M., & Abrams, D. (2004). Reinforcing the glass ceiling: The Sowell, T. (2004). Affirmative action around the world: An empirical
consequences of hostile sexism for female managerial candidates. Sex study. New Haven, CT: Yale University Press.
Roles, 51, 609 – 615. http://dx.doi.org/10.1007/s11199-004-5470-8 Spector, P. E., & Brannick, M. T. (2011). Methodological urban legends:
Miller, T. L., Grimes, M. G., McMullen, J. S., & Vogus, T. J. (2012). The misuse of statistical control variables. Organizational Research
Venturing for others with heart and head: How compassion encourages Methods, 14, 287–305. http://dx.doi.org/10.1177/1094428110369842
Tougas, F., & Beaton, A. M. (1993). Affirmative action in the work place: Journal of Applied Social Psychology, 40, 899 –916. http://dx.doi.org/
For better or for worse. Applied Psychology, 42, 253–264. http://dx.doi 10.1111/j.1559-1816.2010.00603.x
.org/10.1111/j.1464-0597.1993.tb00741.x Wheeler, S. C., & Petty, R. E. (2001). The effects of stereotype activation
Tougas, F., Brown, R., Beaton, A. M., & Joly, S. (1995). Neosexism: Plus on behavior: A review of possible mechanisms. Psychological Bulletin,
ça change, plus c’est pareil. Personality and Social Psychology Bulletin, 127, 797– 826. http://dx.doi.org/10.1037/0033-2909.127.6.797
21, 842– 849. http://dx.doi.org/10.1177/0146167295218007 Yang, C., D’Souza, G. C., Bapat, A. S., & Colarelli, S. M. (2006). A
Watson, D., & Clark, L. A. (1994). The PANAS-X: Manual for the positive cross-national analysis of affirmative action: An evolutionary psycho-
and negative affect schedule-expanded form. Iowa City, IA: University logical perspective. Managerial and Decision Economics, 27, 203–216.
of Iowa. http://dx.doi.org/10.1002/mde.1293
Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social
psychological experimentation. Personality and Social Psychology Bul-
letin, 25, 1115–1125. http://dx.doi.org/10.1177/01461672992512005 Received December 23, 2015
Werhun, C. D., & Penner, A. J. (2010). The effects of stereotyping and Revision received June 9, 2016
implicit theory on benevolent prejudice toward Aboriginal Canadians. Accepted June 14, 2016 䡲
New Policy for the Journal of Personality and Social Psychology

The Journal of Personality and Social Psychology is inviting replication studies submissions.
Although not a central part of its mission, the Journal of Personality and Social Psychology values
replications and encourages submissions that attempt to replicate important findings previously
published in social and personality psychology. Major criteria for publication of replication papers
include the theoretical importance of the finding being replicated, the statistical power of the
replication study or studies, the extent to which the methodology, procedure, and materials match
those of the original study, and the number and power of previous replications of the same finding.
Novelty of theoretical or empirical contribution is not a major criterion, although evidence of
moderators of a finding would be a positive factor.
Preference will be given to submissions by researchers other than the authors of the original finding,
that present direct rather than conceptual replications, and that include attempts to replicate more
than one study of a multi-study original publication. However, papers that do not meet these criteria
will be considered as well.
Submit through the Manuscript Submission Portal at (http://www.apa.org/pubs/journals/psp/) and

please note that the submission is a replication article. Replication manuscripts will be peer-
reviewed and if accepted will be published online only and will be listed in the Table of Contents
in the print journal. As in the past, papers that make a substantial novel conceptual contribution and
also incorporate replications of previous findings continue to be welcome as regular submissions.

"The Compassionate Sexist? How Benevolent Sexism Promotes and Undermines Gender Equality in The Workplace

Uploaded by

Copyright:

Available Formats

You might also like

"The Compassionate Sexist? How Benevolent Sexism Promotes and Undermines Gender Equality in The Workplace

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

"The Compassionate Sexist? How Benevolent Sexism Promotes and Undermines Gender Equality in The Workplace

Uploaded by

Copyright:

Available Formats

Journal of Personality and Social Psychology © 2016 American Psychological Association

2016, Vol. 111, No. 5, 706 –727 0022-3514/16/$12.00 http://dx.doi.org/10.1037/pspi0000072

The Compassionate Sexist? How Benevolent Sexism Promotes and

Ivona Hideg D. Lance Ferris

Supplemental materials: http://dx.doi.org/10.1037/pspi0000072.supp

Variable Men (n ⫽ 52) Women (n ⫽ 63) Overall (n ⫽ 115) 1 2a 3b 4 5 6

Compassion Attitudes Behavioral intentions

Step 1 R2 .01 .15ⴱⴱⴱ .09ⴱⴱ

SD ⫽ 3.27) at a midsize Canadian university who received course

ments. Following the exposure to statements, all participants read

the same EE policy as in Study 1. After reading the EE policy, the

Benevolent sexism prime Control prime

point is that benevolent sexists should believe that women

Oveis et al., 2010). To assess sadness, participants rated to what

Human resource management positions Finance positions

Compassion toward women Attitudes Behavioral intentions

Total R2 .16ⴱⴱ .38ⴱⴱ .33ⴱⴱⴱ

intentions, b ⫽ .55, t(82) ⫽ 4.10, p ⬍ .001. Finally, the conditional

Additional analyses. In line with Study 1 and Study 2, to rule

out the possibility that sadness underlies the effect of benevolent

Feminine positions Masculine positions

Compassion toward women Attitudes Behavioral intentions

Step3 ⌬R2 .001 .001 .004†

5 ticed. Similarly, our results suggest that benevolent sexism leads to

related to dependency-oriented helping behaviors toward women, Conclusion

and leading to inaction and lack of social change in promoting the

employment of women in roles in which they are underrepre-

http://dx.doi.org/10.1007/s11199-009-9585-9 351–374. http://dx.doi.org/10.1037/a0018807

New Policy for the Journal of Personality and Social Psychology

Submit through the Manuscript Submission Portal at (http://www.apa.org/pubs/journals/psp/) and

You might also like