Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257663863

Mind the Gap: Framing of Women’s Success and Representation in STEM


Affects Women’s Math Performance under Threat

Article  in  Sex Roles · April 2012


DOI: 10.1007/s11199-012-0252-1

CITATIONS READS

27 540

3 authors, including:

David M Marx Radmila Prislin


San Diego State University San Diego State University
33 PUBLICATIONS   1,433 CITATIONS    51 PUBLICATIONS   2,052 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Radmila Prislin on 08 January 2015.

The user has requested enhancement of the downloaded file.


Sex Roles (2013) 68:454–463
DOI 10.1007/s11199-012-0252-1

ORIGINAL ARTICLE

Mind the Gap: Framing of Women’s Success and Representation


in STEM Affects Women’s Math Performance under Threat
Emily S. Shaffer & David M. Marx & Radmila Prislin

Published online: 19 December 2012


# Springer Science+Business Media New York 2012

Abstract Interventions designed to combat the negative based approach highlighting women’s advances in STEM to
effects of stereotype threat have primarily taken an alleviate stereotype threat.
individual-based approach. The current study sought to ex-
pand upon these strategies by taking a group-based ap- Keywords Stereotype threat . Social progress . STEM .
proach to reduce stereotype threat effects. Specifically, we Math performance
investigated whether the success and numerical representa-
tion of women in STEM positively impacts women’s math
performance and affective reactions. We hypothesized that Introduction
1) women under threat (control) would perform worse than
men; 2) there would be a larger performance difference for Although men have historically dominated the fields of
women than men when exposed to the success and balanced science, technology, engineering, and math (STEM), recent
representation of women in STEM compared to the control statistics show that U.S. male and female high school stu-
condition; 3) there would be a larger performance difference dents are now performing equally well in mathematics clas-
for women than men between the balanced condition and ses (Hill et al. 2010). However, in high-stakes testing
the unbalanced condition where women are portrayed as situations, such as the Scholastic Assessment Test (SAT)
successful, but not equally represented in STEM. For this or Graduate Record Exam (GRE), women continue to
study, male (n056) and female (n066) U.S. undergraduates underperform relative to men on the quantitative sections
from a large southern California state university read infor- (The College Board, 2011; Willingham & Cole, 1997). Over
mation about women’s success and representation in STEM the years, an increasing amount of research has been devot-
(or no information), completed a math exam under stereo- ed toward understanding this persistent gender gap in stan-
type threat conditions, and then expressed their threat-based dardized test performance (e.g., Fredricks & Eccles, 2002;
concerns. Results revealed that women performed worse Frome & Eccles, 1998; Schmader & Johns, 2003). One
than men in the control condition. Women in the balanced factor that may play a role is stereotype threat, which is
condition performed better than women in the control and defined as the situational pressure posed by the prospect that
unbalanced conditions. Men’s performance was unaffected a poor performance would be judged through the lens of a
by the balance or imbalance of women in STEM. Women’s negative group-relevant stereotype (Steele, 1997; Steele &
affective reactions largely mirrored the performance results. Aronson, 1995; Steele et al. 2002; these and all other studies
This study provides compelling evidence for using a group- cited were conducted in the U.S. unless otherwise noted).
Since stereotype threat was first documented, a number
of interventions designed to reduce the negative effects of
threat have been proposed. Among them, the use of role
E. S. Shaffer (*)
Department of Psychology, Tulane University, models in the U.S. (Marx & Roman, 2002; McIntyre et al.
2007 Percival Stern Hall, 2003) and the Netherlands (Marx et al. 2005), self-
New Orleans, LA 70118, USA affirmation (e.g., Martens et al. 2006), and alternative attri-
e-mail: eshaffer@tulane.edu
butions for the difficulty one may experience during a test
D. M. Marx : R. Prislin (Ben-Zeev et al. 2005; Johns et al. 2005) have all proven
San Diego State University, San Diego, USA effective among participant samples in the U.S. and Europe.
Sex Roles (2013) 68:454–463 455

Whereas these approaches proved successful, each incorpo- may persist and continue to have adverse consequences for
rated an individual approach to what is widely considered a those who are negatively stereotyped. In support of this
group-based phenomenon (see Prislin, 2009). Although the reasoning, Shapiro (2011) asserts that different types of
immediate impact of stereotype threat typically occurs at the stereotype threats will require different interventions.
individual-level (e.g., decreased math performance), the Further documenting the importance of a group-based inter-
threat itself is not necessarily solely directed at the individ- vention against stereotype threat are the findings showing
ual. Instead, threat can originate from more pervasive and that individuals under threat, relative to when they are not
subtle cues about one’s group gathered from one’s environ- under threat, are more focused on their group than individ-
ment and daily experiences (Murphy et al. 2007). In fact, ual identity (Marx et al. 2012a; Marx et al., 2005; cf. Rydell
one of the early conceptualizations of threat (Steele, 1997) et al. 2009).
was that of “a threat in the air” (p. 613). Hence, we felt it To our knowledge, the extant literature has focused al-
important to examine whether a less individually directed, most exclusively on individually-based interventions. An
group-based approach would also serve as an effective way exception is a study conducted in the U. K. by Rosenthal
to disarm the adverse effects of stereotype threat. Instead of and Crisp (2006) that focuses on “blurring inter-group
placing the onus on the individual to disarm the threat, boundaries” (p. 501) or reducing the distinctions between
another effective strategy may be to focus on changing men and women. This strategy makes the stereotype less
perceptions of the negative, group-wide impact of the threat relevant and therefore less threatening to women. Its effec-
itself. tiveness was examined in a series of studies that encouraged
The need for an alternative approach seems critical in British college women to consider ways in which there was
light of recent theorizing by Shapiro and Neuberg (2007) intergroup overlap between men and women before com-
who argue that stereotype threat may originate from multi- pleting a math exam under threat conditions. Results
ple sources. Specifically, they contend that not only con- showed that women who were encouraged to think about
cerns at the individual level, but also those at the group level the similarities between men and women performed signif-
may be responsible for stereotype threat effects. Group-level icantly better on a math exam than those who were not
concerns refer to the apprehension regarding how one may encouraged to blur these inter-group boundaries. Blurring
be judged by both ingroup and outgroup members if one group boundaries may be effective because individuals are
falters. In the current research, we propose that interventions no longer focused on what sets them apart from members of
focused on the group-level may be a more widely effective another group. Although this method proved beneficial,
way to combat the negative effects of stereotype threat. For there may be a hidden consequence brought on by encour-
the purposes of this research, we sought to investigate the aging the elimination of group boundaries—one’s specific
efficacy of this strategy on women’s performance in the group identity may become less distinct. In other words, for
math domain. We propose that this strategy will not only women to perform well on the math exam, they need to
impact U.S. college women’s math performance, but may reduce what is unique to their group. In our view, a more
also ameliorate their concerns about confirming the negative practical intervention should have a positive impact on
gender stereotype regarding women’s math ability (i.e., performance while also allowing the identities of stereo-
threat-based concerns; Marx et al., 2005). typed individuals to remain unchanged. That is, individuals
should not need to adapt their identity to become of less
Group-Based vs. Individual-Based Strategies value or importance in order to reduce the effects of stereo-
type threat. We argue that such an intervention may depend,
The experience of stereotype threat is multifaceted. Indeed, in part, on their group’s representation within the stereo-
as Shapiro and Neuberg (2007) note, the concern one feels typed domain (i.e., “are women becoming equally repre-
in a stereotype threat situation may stem, not only from sented, or are women remaining underrepresented?”).
potential self-insights, but also from what a person may
convey to ingroup and outgroup members. Thus, in addition The Role of Group-Based Success in Stereotype Threat
to the concern that one will confirm the stereotype in their Situations
own eyes, it is the concern that one’s performance will
confirm the stereotype in the eyes of ingroup members, Due to the salience of women’s gender identity in stereotype
outgroup members, or both that give rise to threat. This threat situations, information regarding their group’s
multifaceted nature of stereotype threat is not reflected in achievement may be particularly impactful. For instance,
the frequently developed, individual-based intervention Brewer and Weber (1994) showed that exposure to success-
strategies. As a result, group-level concerns that feed ste- ful ingroup members can be both enhancing and threatening
reotype threat (i.e., ingroup and outgroup threat) may not be to the self under certain conditions. In a negatively stereo-
fully addressed. If so, then group-based sources of threat typed situation, it may be particularly important to present
456 Sex Roles (2013) 68:454–463

individuals with positive information about their group’s to engage in more environmentally-friendly behaviors than
achievement in the relevant domain in order to defend when the outgroup was portrayed more positively. Applying
against threat. Information regarding positive group-based this reasoning to the current research, we argue that when
achievement may allow those who are targeted by the ste- provided with positive information about women’s progress
reotype to incorporate the successes of ingroup members and improved representation relative to men in STEM-
into the self. Thus, a woman who learns about the achieve- related fields, women may feel inspired to enact these
ments of her group in STEM fields may recognize that achievements of their group. Information that emphasizes
because she is a woman she can perform well and may social progress could reassure women that their group is
begin to think “we can do it” (Marx et al., 2005, p. 432). moving into these fields and may also decrease any concern
This point is further underscored by Blanton, Crocker, and they may have regarding performance in STEM.
Miller (2000) who showed that in a stereotyped context, the Conversely, information that emphasizes that women’s rep-
success of an ingroup member is beneficial when that resentation in STEM still lags behind that of men may only
ingroup member’s success disconfirms the stereotype. further exacerbate their concerns about performing well. In
Specifically, these researchers showed that self-esteem essence, women may be doubly threatened in this situation
among African American participants was higher when they —first they are aware of the negative stereotype associated
overheard that an ingroup confederate (i.e., a fellow African with their group, and second they learn that in spite of all
American) scored particularly well on a math exam than efforts at improvement, their group still has yet to achieve
when the same ingroup confederate was reported to have not equal representation in STEM, further cementing their infe-
achieved this level of success. These researchers also point rior status in these areas.
out that exposure to information regarding successful
ingroup members may have important implications for re- Group-Level Change and Stereotyped Concerns
ducing stereotype threat.
In addition to reducing the negative effects of stereotype threat
The Role of Group-Wide Representation in STEM on performance, we sought to determine whether an interven-
tion focused at the group level would likewise be effective at
Notwithstanding the impact of information about success of alleviating women’s concerns about confirming a stereotype
an ingroup member (a woman), it may be equally important associated with their group. Previous research has shown that
to provide comparative information about the representation even when performance is enhanced via exposure to female
of ingroup (women) and outgroup (men) members in a role models, women still expressed some concern about con-
stereotyped domain. This information may not only have firming the negative stereotype regarding women’s ability in
an immediate effect on test performance, but may also speak math (e.g., Marx et al. 2005; Marx et al. 2012b). It is possible
to the content of the stereotype itself. Whereas information that this continuing concern occurs because a single instanti-
about the ingroup that disconfirms the stereotype can be ation of an ingroup member who disconfirms the stereotype is
useful—especially in an immediate testing situation, infor- insufficient to alleviate these threat-based concerns. After all,
mation that is favorable to the ingroup in relation to the a single role model or individual who defies a stereotype does
outgroup may have an additional and more lasting positive not necessarily invalidate a stereotype associated with one’s
effect. For instance, positive information regarding the group (Weber and Crocker 1983). It stands to reason that a
ingroup’s increased representation in the relevant domain reduction of such stereotyped concerns may require an
has the potential to make the stereotype less impactful. This approach that is focused at the group-level. To this end, we
may occur because the increase in representation may send argue that exposing women to messages that highlight both
the message to fellow ingroup members that they are not the success of their group and their group’s progress in a
only capable in this domain, but also that they will not be domain where they have been underrepresented in the past
judged in terms of the stereotype if they struggle. Altering may not only improve performance, but also has the potential
perceptions of women’s representation in STEM to reflect a to alter women’s (and men’s) threat-based concerns.
more balanced presence of men and women may effectively
alter perceptions of the field itself to be less threatening Present Research
(Murphy et al., 2007; Walton and Cohen 2007). In fact,
work by Rabinovich, Morton, Postmes, and Verplanken Providing women with information about their group’s suc-
(2011) conducted with a British sample of participants cess and progress in STEM-related fields, in comparison to
showed that the valence of group-based information can men in those fields, allows women to see a more compre-
have differential effects on behavioral outcomes. In their hensive picture of their standing in this domain. If this
study, information that portrayed the participants’ own information is presented in a way that allows women to
group in a positive light led participants to report intentions draw positive conclusions with regard to both their ingroup
Sex Roles (2013) 68:454–463 457

and their outgroup—that is, their ingroup can achieve suc- which, however, still leaves women under-
cess and are almost equal to the outgroup—then this may represented in STEM. Furthermore, men
lessen the burden associated with overcoming the negative are expected to perform better than women
stereotype. Accordingly, the present research sought to im- in the unbalanced condition.
plement a two-pronged, group-level approach to mitigate
For threat-based concerns we tested the following two
the negative effects that women experience under stereotype
primary hypotheses.
threat. Namely, this strategy aimed to reduce these effects by
not only showing that a woman can succeed in STEM, but Hypothesis 4: Women will have higher threat-based con-
also by highlighting how the gender gap in STEM has been cerns relative to men in the control condition,
closing. Such a strategy may be utilized when combating the thus demonstrating a typical stereotype threat
many different types of stereotype threat (Shapiro & effect on affective reactions.
Neuberg, 2007) because it can speak to both self and group Hypothesis 5: The difference between women’s threat-
threats. That is, this strategy incorporates information pre- based concerns in the balanced and unbal-
viously shown to be effective in individual-based strategies, anced conditions will be larger than the
but also includes information that may disarm group-level difference for men in these two conditions.
threats. Importantly, this strategy should improve women’s Additionally, women will show lower
performance, but not at the expense of men’s performance. threat-based concerns in the balanced con-
In this study, we tested five hypotheses that focused on dition than in the control condition.
the effectiveness of our group-based strategy for reducing
the effects of stereotype threat on performance and threat-
based concerns among participants of comparable aptitude Method
and identification with math. Because prior math aptitude
and identification with math can vary widely between par- Participants and Design
ticipants, we created a composite covariate to equate all
participants on these variables. This was important as ste- One hundred twenty two undergraduates (56 male, 66
reotype threat and strategies aimed at reducing it operate to female; see Table 1) from San Diego State University
affect individuals with similar math-related backgrounds. (SDSU) participated in exchange for partial fulfillment of
Each of the five hypotheses was tested after first establish- a psychology course requirement. A post hoc power
ing that there was a significant interaction of participant analyses revealed that with our sample size (N0122)
gender with our experimental manipulations. and an expected moderate effect-size (η2 0.05), we had
approximately a 70 % chance of detecting a significant
Hypothesis 1: Men will perform better than women in the effect. Participants were recruited via an online recruit-
control condition where participants were ment system (SONA), which listed all available studies
provided with neutral information regard- being conducted in the psychology department.
ing the history of academics at SDSU, thus Participants were unaware about the purpose of the study
demonstrating a typical stereotype threat prior to arrival at the laboratory. Further details about
effect in performance. participants’ demographic characteristics are contained in
Hypothesis 2: Compared to the control condition, there Table 1. For this study we used a 2 (Participant Gender:
will be a larger math performance differ- male, female) x 3 (STEM Representation: balanced, un-
ence for women relative to men in the balanced, control) between-participants design.
balanced condition where participants are
provided with information showing that Procedure
women can and do achieve success in
STEM, and that the progress of women is Participants were greeted by an experimenter and then
altering the representation of their group in shown to one of two separate cubicles that contained a
these fields. Furthermore, women are desktop computer. They were then told that the study was
expected to perform as well as men in the about academics and student life and consisted of a series of
balanced condition. unrelated tasks that would be presented on the computer. To
Hypothesis 3: There will be a larger math performance manipulate stereotype threat, participants were first pre-
difference for women relative to men be- sented with information that past research has shown gender
tween the balanced and unbalanced condi- differences in math favoring men, and that they would be
tion where they are provided with completing a math exam later in the session (Marx et al.
information about the success of women, 2012a, b; Rydell et al., 2009; Spencer et al., 1999).
458 Sex Roles (2013) 68:454–463

Table 1 Participant gender, age, and race/ethnicity across experimental conditions

Balanced Unbalanced Control

Men Women Men Women Men Women

Total N (%) 17 (43.6 %) 22 (56.4 %) 23 (52.3 %) 21 (47.7 %) 16 (41.0 %) 23 (59.0 %)


Age 19.00 18.86 19.22 18.67 19.25 19.17
SD (1.22) (1.13) (1.17) (.97) (1.13) (1.72)
Race
White 4 (23.5 %) 11 (50.0 %) 15 (65.2 %) 11 (52.4 %) 6 (37.5 %) 10 (43.5 %)
Latino/a 6 (35.3 %) 6 (27.3 %) 3 (13.0 %) 5 (23.8 %) 4 (25.0 %) 7 (30.4 %)
Asian 6 (35.3 %) 2 (9.1 %) 2 (8.7 %) 3 (14.3 %) 4 (25.0 %) 4 (17.4 %)
African-American 1 (5.9 %) 1 (4.5 %) 0 (.0 %) 1 (4.8 %) 0 (.0 %) 0 (.0 %)
Middle-Eastern 0 (.0 %) 0 (.0 %) 0 (.0 %) 0 (.0 %) 1 (6.2 %) 1 (4.3 %)
Other 0 (.0 %) 2 (9.1 %) 3 (13.0 %) 1 (4.8 %) 1 (6.2 %) 1 (4.3 %)

There were no significant differences in these demographic variables across conditions

Next, participants read a short bogus newspaper article in math?” The scale was anchored with the terms (1) not at
that served as our manipulation of STEM representation (see all important and (7) very important (α0.78). To control for
Appendix). In the balanced condition, participants read an prior math ability, we also collected participants’ self-
article that highlighted the number of women in STEM fields reported math SAT scores. To equate all participants on
has significantly increased over the years, such that women math identification and math ability, we created a composite
are now in nearly equal numbers to men in STEM careers. The covariate based on their self-reported math SAT and MIQ
article in the unbalanced condition was nearly identical to the scores (for a similar procedures, see Marx & Roman, 2002).
aforementioned article except that participants read about how Specifically, to express each variable in a common metric,
the presence of women in STEM fields has increased over the we calculated z-scores and averaged across the standardized
years, but women still remain underrepresented in comparison variables. This composite variable was used as a covariate in
to men. In addition to information regarding women’s STEM all subsequent analyses.
representation, both articles provided examples of women
who have achieved success in the relevant domain as a way Main Dependent Measures
to reinforce that women are capable of succeeding in STEM,
despite being equally or unequally represented in these Math Performance
domains. Participants in the control condition read about the
history of academics at SDSU. Participants completed ten GRE-like math problems that have
After reading their respective articles, participants took a been used in prior work on stereotype threat (e.g., Marx et al.
math exam under stereotype threat conditions. Afterwards, 2012a, b). Participants were given 10 min to complete the
they were asked a number of follow-up questions pertaining math exam. Performance was determined by the number of
to their threat-based concerns and basic demographic infor- questions answered correctly, with a possible range from 0 to
mation. Once participants completed these measures, they 10. Consistent with standardized testing procedures, calcula-
were thanked and debriefed. tors were not allowed, but scratch paper was provided.

Prescreen Measures Threat-Based Concerns

Math Identification and Math Ability This measure (Marx et al. 2012a, b; Marx et al., 2005;
Rydell et al. 2010; see also, Marx & Goff, 2005) consists
All participants answered a prescreen questionnaire prior to of three items that assessed how concerned participants were
participation in the study. This questionnaire included a about confirming a group-relevant stereotype (α0.63). For
modified version of the Math Identification Questionnaire instance, they were asked to rate how much they agreed with
(MIQ; Brown & Josephs, 1999), which assesses the extent statements such as, “I worry that my ability to perform well
to which one is identified with the math domain. For exam- on math tests is affected by my gender.” The scale was
ple, participants indicated how much they agreed with three anchored with the terms (1) strongly disagree and (7)
questions, including “How important is it for you to do well strongly agree. The three items were averaged to form a
Sex Roles (2013) 68:454–463 459

composite threat-based concern score with higher numbers difference between the balanced and control conditions would
reflecting greater concern. be larger for women (M05.48 vs. M04.15) than men (M0
Previous research using this scale has reported alpha 5.17 vs. M05.81). This hypothesis was supported, F(1, 115)0
levels from .70 to .83 (e.g., Marx et al., 2005; Marx & 5.40, p0.02, η2 0.04. That is, women did benefit from being
Goff, 2005; Rydell et al. 2010). Because our observed alpha exposed to the balanced information relative to when this
was lower than expected, we submitted the three scale items information was not provided (control condition), F(1,
to a principle components analysis with varimax rotation to 115)05.85, p<.02, η2 0.05, while men showed no difference
test whether they reflected the same underlying factor. between these two conditions, F(1, 115)00.99, p0.32. Follow
Results confirmed that all three items loaded together on up analyses revealed that within the balanced condition wom-
one factor with loadings greater than .71. en (M05.48) performed as well as men (M05.17), F(1, 115)0
0.27, p0.60. Hypothesis 3 predicted that women would per-
form better in the balanced (M05.48) relative to the unbal-
Results anced (M 04.30) condition, but that there would be no
performance difference for men in these two conditions (M0
Math Performance 5.17 vs. M05.66). This hypothesis was also supported, F(1,
115)04.17, p0.04, η2 0.03. Follow up analyses demonstrated
Participants’ math performance was assessed using a 2 that, consistent with a stereotype threat effect, men performed
(Participant Gender) × 3 (STEM Representation) Analysis better (M05.66) than women (M04.30) in the unbalanced
of Covariance (ANCOVA) on the total number of math condition, F(1, 115)05.97, p<.02, η2 0.05.
problems answered correctly, controlling for the covariate Taken together, these performance results provide evi-
(see Table 2). This analysis yielded an effect for the cova- dence for the effectiveness of a group-based strategy in
riate, F(1, 115)038.28, p<.01, and a main effect of gender, combating stereotype threat. Moreover, these results show
such that men (M05.55) performed better than women (M0 that highlighting the unequal representation of women in
4.64), F(1, 115)07.11, p<.01, η20.06. There was no main STEM, although seemingly necessary to inspire women to
effect of STEM Representation, F<1.00. We also found a overcome this underrepresentation, did not reduce stereo-
reliable Participant Gender by STEM Representation inter- type threat effects on performance. A more effective strategy
action, F(2, 115)03.19, p<.05, η2 0.03. To interpret this appears to be highlighting that women are becoming more
significant omnibus interaction, we conducted planned con- equally represented and successful in STEM, thus reinforc-
trasts that were informed by our hypotheses. ing women’s social progress in areas once dominated by
Hypothesis 1 predicted a typical stereotype threat effect on men. These results further illustrate that calling attention to
performance (Spencer et al., 1999). Results revealed that men women’s success and subsequent equal representation in
(M05.81) did perform better than women (M04.15) in the STEM was clearly beneficial for women without harming
control condition F(1, 115)07.64, p<.01, η2 0.06. We next the math performance of men, providing support for the
tested Hypothesis 2, which anticipated that the performance practicality of our approach for combating the effects of
stereotype threat.
Table 2 Mean (SD) math performance and threat-based concerns as a
function of STEM representation and participant gender Threat-Based Concerns

Balanced Unbalanced Control


To examine participants’ threat-based concerns we conducted
Men Women Men Women Men Women a 2 (Participant Gender) x 3 (STEM Representation)
ANCOVA, controlling for the covariate (see Table 2). This
Performance 5.17b 5.48b 5.66b 4.30a 5.81b 4.15a analysis revealed a significant effect for the covariate, F
SD (2.62) (1.99) (2.51) (1.81) (1.67) (1.94) (1,115) 04.56, p < .04, but no main effects, Fs < 1.00.
Threat 2.67b 2.06b 2.25b 2.66b 1.83a 2.56c Importantly, we found a significant Participant Gender by
SD (.98) (1.31) (1.07) (1.10) (.97) (1.13) STEM Representation interaction, F(2, 115)03.94, p0.02,
η2 0.05. To interpret this significant omnibus interaction,
The balanced condition stressed women’s success and increased nu-
planned comparisons were conducted that were informed by
merical representation in STEM. The unbalanced condition presented
information about women’s success in STEM, but also showed that our final two primary hypotheses.
women remained underrepresented with regard to men. The control As expected and consistent with past research (Marx et
condition detailed the history of academics at SDSU. Higher numbers al. 2012a; Marx et al., 2005; Rydell et al., 2009), we found
indicate better math performance and greater threat-based concerns.
All means within a condition that do not share a subscript differ at
support for Hypothesis 4. Specifically, women in the control
p<.05. Participants’ math performance could range from 0 to 10 and condition were more concerned about confirming a group-
their threat-based concern scores could range from 1 to 7 relevant stereotype (M02.56) than were men (M01.83), F
460 Sex Roles (2013) 68:454–463

(1,115)04.21, p0.04, η2 0.04. Next, we examined whether overcome the negative stereotype surrounding their
women’s threat-based concerns would be higher in the un- group; however, this did not appear to be the case. We,
balanced compared to the balanced condition, whereas thus, argue that providing information that highlights the
men’s threat-based concerns would not differ between these status quo (i.e., the overrepresentation of men) may only
two conditions (Hypothesis 5). We again found support for solidify perceptions of women’s status in STEM, making
this hypothesis, F(1,115)04.45, p<.04, η2 0.04, such that it difficult for women to envision change. This was
the difference in women’s threat-based concerns between evident in women’s similarly subpar performance in the
the balanced (M02.06) and unbalanced (M02.66) condi- unbalanced condition and in the control condition (i.e., a
tions was greater than men’s (M 02.67 vs. M 02.25). typical stereotype threat condition). We also found that
Follow up analyses revealed that for women, threat-based these group-based messages had parallel effects on wom-
concerns were somewhat reduced in the balanced condition en’s threat-based concerns. Women in the balanced con-
(M02.06) compared to the control condition (M02.56), F dition were slightly less concerned about confirming a
(1,115)02.35, p<.13, η2 0.02, suggesting that the balanced group-relevant stereotype than in the unbalanced condi-
information did allay their concerns to some extent. This tion where they were confronted with women’s underrep-
finding thus demonstrates that a group-level approach is resentation in STEM, even when provided with
somewhat beneficial in lowering women’s threat-based con- information showing that women can succeed in STEM.
cerns, something that has not been found in previous ste- Men’s performance did not differ as a function of wom-
reotype threat intervention work (see e.g., Marx et al. 2012a; en’s group-based success even when presented with in-
Marx et al., 2005). This finding also suggests that an even formation showing that women were gaining ground in
more targeted approach may be needed to fully reduce STEM fields. The equal representation of women in
women’s threat-based concerns. Interestingly, and unexpect- STEM increased threat-based concerns for men, yet, as
edly, men’s threat-based concerns were higher in the bal- noted above, these concerns did not adversely impact
anced compared to the control condition, F(1,115)04.87, p their performance. Importantly, this strategy, which
<.03, η2 0.04, implying that information regarding women’s proved useful and practical for women, does not come
equal representation and success in STEM may have raised at the expense of men’s performance. It is interesting to
men’s threat-based concerns. That is, men may have felt note that men did perform slightly less well in the
more negatively about their group’s place in the “STEM balanced condition than in the control condition. Future
hierarchy”, because of this positive portrayal of women. research may examine whether a manipulation that more
On the whole, these results echo the findings for perfor- explicitly threatens men’s status and numbers in the
mance: Just as women performed better when their group STEM fields would have an even greater negative impact
was portrayed as being equally represented in STEM, rather on their performance. However, as a first step in inves-
than unequally represented, their threat-based concerns were tigating this group-based strategy, we felt it important not
likewise reduced. Men, however, seemed to be more sensi- to sacrifice men’s performance for the benefit of women.
tive to variations in women’s representation in STEM, such Our findings also provide strong support for using group-
that their concerns were increased in the balanced condition based strategies in addition to individual-based strategies
relative to the control condition, a point we return to in the against the effects of threat. This is not to say that our
discussion. group-based intervention is more beneficial or effective than
individual-based strategies, such as role models. However,
when taking a group-based focus, providing information
Discussion regarding the success of women may only solve one side
of the equation. Showing that women are becoming more
This study highlights a novel strategy for combating the equally represented in STEM may be especially impactful
negative effects of stereotype threat: namely, presenting because it changes perceptions of the field itself. This par-
positive group-based information regarding women’s suc- ticular strategy not only had beneficial effects on women’s
cess and progress in STEM. When women were able to performance, but it is also the first strategy that showed
see that not only can women succeed in STEM, but also some utility in diminishing threat-based concerns. We argue
that their group is gaining ground in these fields, they that this occurred because highlighting the success and
performed as well as men despite being in a stereotype representation of one’s group speaks specifically to group-
threat situation. When the success and representation of level concerns. Rather than presenting a single instantiation
their group was portrayed as being unequal to men, of success in math, as with the use of role models, the
women’s performance suffered relative to men. One presentation of group-wide change seemed impactful
may assume this inequality in success and representation enough to reduce women’s threat-based concerns to some
would motivate women to perform well to defy or degree.
Sex Roles (2013) 68:454–463 461

Men’s threat-based concerns were higher in the bal- Coda


anced condition than in the control condition. In each of
these conditions, it was mentioned that women had at The results of this study extend research on stereotype threat
least made some progress in STEM. That is, in both in two important ways. First, we documented the beneficial
conditions, participants were presented with information effects of positive group-based information in reducing ste-
about a woman who has made achievements in STEM. reotype threat. This study provides a first step toward de-
However, only in the balanced condition were women veloping a comprehensive strategy that may target multiple
portrayed to be in equal numbers to men. It is possible types of stereotype threat. Whereas we agree with Shapiro
that men, who are positively stereotyped in this domain, and Neuberg (2007, p. 126) that there is no “silver bullet” in
could have interpreted any positive progress of women as combating stereotype threat, we believe that a comprehen-
potentially threatening to their place in the hierarchy. The sive strategy is useful when multiple threats may be present.
progress of women, albeit less significant in the unbal- Second, whereas most interventions have primarily focused
anced condition, may signal to men that they are losing on relieving threat at the individual level, this study provides
their stronghold in STEM. Because of this, men may feel strong evidence that taking a broader, group-based perspec-
extra pressure to perform well in order to assert their tive can mitigate the performance effects and also the threat-
position in the field. In fact, Brown and Josephs (1999) based concerns associated with stereotype threat.
have shown that there can be disadvantages associated When considering actions needed for further advancements
with being positively stereotyped in a particular domain. of women in STEM, caution should be used in order to present
Presumably, men can become worried about confirming an effective message. That is, highlighting the fact that women
the stereotype and living up to the positive performance of are not equal in number to men in STEM-related careers,
their group. Because of this extra pressure, men’s perfor- which on the surface may sound motivating, may in fact
mance can also suffer. Brown and Joseph’s findings may discourage women to a further extent and trigger a greater
help explain why we found an increase in men’s threat- threat. This is not to say that an effective tactic would be to
based concerns when exposed to a woman’s success and present an unrealistic picture that shows no disparity exists in
increased representation of women in STEM compared to STEM. To the contrary, a message regarding underrepresen-
the control condition. Although this concern may have tation may be an effective call to action to achieve equality.
been somewhat raised with exposure to positive informa- However, we contend that messages of underrepresentation,
tion regarding women’s progress in STEM, it was not specifically in immediate testing situations, may need to be
potent enough to impact men’s performance negatively. tempered with additional information showing that women
It is worth noting that the alpha for the threat-based are gaining ground and those women who are already in
concerns scale was relatively lower than in past research STEM fields are doing well.
using the same scale. However, due to the consistent pat-
terns of effects as well as the results of the principle com-
ponents analysis, we believe that this measure still provided Appendix
a valid assessment of participants’ threat-based concerns. It
is also worth noting that our study included conditions that Balanced Condition
consistently provided positive information regarding wom-
en’s success and varied the extent of their representation. We Recent Study Shows Men and Women Nearly Equal
are reluctant to believe that receiving any type of positive in STEM
information, regardless of group membership, would ac-
count for the effects in this study. In fact, Brewer and WASHINGTON, D.C. – Over the last 50 years, women
Weber (1994) have shown that, for numerical minorities, have been clearly outnumbered in Science, Technology,
positive information regarding the outgroup results in lower Engineering, and Mathematics (STEM) fields. However, a
evaluations of the self regardless of the resultant negative recent study by the Department of Education shows that this
consequences. That is, to minorities, this positive informa- gender imbalance is getting smaller.
tion only highlights the outgroup’s superiority and their own The Department of Education has been collecting statis-
group’s inferiority. Indeed, the results of our study docu- tics on the number of men and women in STEM fields for
ment that not all positive information is treated equally. In the last 25 years. The first study revealed that men out-
order for this positive information to be effective, it must numbered women 4 to 1 in these fields and careers. Over
speak directly to the concerns of the individual. Concerns the years, the number of women present in STEM fields has
about disconfirming the group-based stereotype were not gradually increased. This latest study, which used data col-
alleviated until it was shown that group-based change was lected from 2007 through 2008, revealed that women now
possible. appear in these careers in nearly equal numbers to men.
462 Sex Roles (2013) 68:454–463

Megan Gallo, head of the Mathematics department at The Normal School was designed to train female elementary
University of California, Berkeley, expressed her satisfac- school teachers and it wasn’t until 1935 that other fields of
tion with this recent finding. “Finally, the statistics are study were offered.
reflecting what we already know. Women are just as able Today, SDSU has over 30,000 students enrolled each
and successful in STEM as their male counterparts.” year in 85 different undergraduate majors. SDSU also offers
Gallo also noted that women have been recognized for 75 master’s programs and 14 joint doctoral programs.
their achievements in STEM fields. Notably, in 2003, Susan Among colleges with 14 or fewer PhD programs, SDSU
Morris received the Fields Medal, the Nobel Prize equiva- has been ranked number one as the most productive research
lent, for her continued work in Mathematics. university for the past four consecutive years. SDSU is also
Ultimately, this study reveals that women have gained among top ranked schools in fields of education, interna-
ground in historically male dominated fields. tional business, social work, and biology.
Due to the demands of the continuously expanding stu-
Unbalanced Condition dent body, the building of Arts and Letters and the Calpulli
Center were opened in the 2006–2007 school year. The
Gender Gap in STEM Closing, But Still Far To Go Imperial Valley branch campus was opened in 1959, and
just recently graduated its first 4-year class.
WASHINGTON, D.C. – Over the last 50 years, women San Diego State University has come a long way since its
have been clearly outnumbered in Science, Technology, modest beginnings as a small teacher’s college. Its success
Engineering, and Mathematics (STEM) fields. However, a has been made possible by continued support from students,
recent study by the Department of Education shows that this faculty, staff and alumni.
gender imbalance is getting smaller.
The Department of Education has been collecting statis-
tics on the number of men and women in STEM fields for
References
the last 25 years. The first study revealed that men out-
numbered women 4 to 1 in these fields and careers. Over
the years, the number of women present in STEM fields has Ben-Zeev, T., Fein, S., & Inzlicht, M. (2005). Arousal and stereotype
threat. Journal of Experimental Social Psychology, 41, 174–181.
gradually increased. This latest study, which used data col- doi:10.1016/j.jesp.2003.11.007.
lected from 2007 through 2008, revealed that women are Blanton, H., Crocker, J., & Miller, D. T. (2000). The effects of in-group
moving into these fields and beginning to rival men. versus out-group social comparison on self-esteem in the context
Megan Gallo, head of the Mathematics department at of a negative stereotype. Journal of Experimental Social
Psychology, 36, 519–530. doi:10.1006/jesp.2000.1425.
University of California, Berkeley, expressed her satisfac- Brewer, M. B., & Weber, J. G. (1994). Self-evaluation effects of
tion with this recent finding. “Finally, the statistics are interpersonal versus intergroup comparison. Journal of
reflecting what we already know. Women are just as able Personality and Social Psychology, 66, 268–275. doi:10.1037/
and successful in STEM as their male counterparts. 0022-3514.66.2.268.
Brown, R. P., & Josephs, R. A. (1999). A burden of proof: Stereotype
However, it is important to realize that women haven’t yet relevance and gender differences in math performance. Journal of
achieved equality to men. I hope that a future study reveals Personality and Social Psychology, 76, 246–257. doi:10.1037/
even more progress made by women in STEM.” 0022-3514.76.2.246.
Gallo also noted that women have been recognized for Fredricks, J. A., & Eccles, J. S. (2002). Children’s competence and
value beliefs from childhood to adolescence: Growth trajectories
their achievements in STEM fields. Notably, in 2003, Susan in two “male-typed” domains. Developmental Psychology, 38,
Morris received the Fields Medal, the Nobel Prize equiva- 519–533. doi:10.1037/0012-1649.38.4.519.
lent, for her continued work in Mathematics. Frome, P. M., & Eccles, J. S. (1998). Parents’ influence on children’s
Ultimately, this study reveals that women are gaining achievement-related perceptions. Journal of Personality and Social
Psychology, 74, 435–452. doi:10.1037/0022-3514.74.2.435.
ground in historically male dominated fields, but more Hill, C., Corbett, C., & St. Rose, A. (2010). Why so few? Women in
strides do need to be made to achieve equality. science, technology, engineering, and mathematics. Washington
DC: AAUW.
Control Condition Johns, M., Schmader, T., & Martens, A. (2005). Knowing is half the
battle: Teaching stereotype threat as a means of improving wom-
en’s math performance. Psychological Science, 16, 175–179.
Anniversary of SDSU Approaches doi:10.1111/j.0956-7976.2005.00799.
Martens, A., Johns, M., Greenberg, J., & Schimel, J. (2006).
SAN DIEGO – This upcoming school year marks the 113th Combating stereotype threat: The effect of self-affirmation on
women’s intellectual performance. Journal of Experimental
anniversary of San Diego State University. In 1897, the Social Psychology, 42, 236–243. doi:10.1016/j.jesp.2005.04.010.
University opened as the San Diego Normal School with Marx, D. M., Cole, C. E., Monroe, A. H., McDonald, M. E., Weisz, B.
only 91 students who attended classes above a pharmacy. M., & Shaffer, E. S. (2012a). When he helps her: The interplay of
Sex Roles (2013) 68:454–463 463

role model gender and comparison relevance on performance Rydell, B. J., Rydell, M. T., & Boucher, K. L. (2010). The effect of
under threat. Manuscript submitted for publication. negative performance stereotypes on learning. Journal of
Marx, D. M., & Goff, P. A. (2005). Clearing the air: The effect of Personality and Social Psychology, 99, 883–896. doi:10.1037/
experimenter race on target’s test performance and subjective a0021139.
experience. British Journal of Social Psychology, 44, 645–657. Schmader, T., & Johns, M. (2003). Converging evidence that stereo-
doi:10.1348/014466604X17948. type threat reduces working memory capacity. Journal of
Marx, D. M., Monroe, A. H., Cole, C. E., & Gilbert, P. N. (2012b). No Personality and Social Psychology, 85, 440–452. doi:10.1037/
doubt about it: When doubtful role models undermine perfor- 0022-3514.85.3.440.
mance under threat. Manuscript submitted for publication. Shapiro, J. R. (2011). Different groups, different threats: A multi-threat
Marx, D. M., & Roman, J. S. (2002). Female role models: Protecting approach to the experience of stereotype threats. Personality and
women’s math test performance. Personality and Social Psychology Social Psychology Bulletin, 37, 464–480. doi:10.1177/
Bulletin, 28, 1183–1193. doi:10.1177/01461672022812004. 0146167211398140.
Marx, D. M., Stapel, D. A., & Muller, D. (2005). We can do it: The Shapiro, J. R., & Neuberg, S. L. (2007). From stereotype threat to
interplay of a collective self-construal orientation and social com- stereotype threats: Implications of a multi-threat framework for
parisons under threat. Journal of Personality and Social causes, moderators, mediators, consequences, and interventions.
Psychology, 88, 432–446. doi:10.1037/0022-3514.88.3.432. Personality and Social Psychology Review, 11, 107–130.
McIntyre, R. B., Paulson, R., & Lord, C. (2003). Alleviating women’s doi:10.1177/1088868306294790.
mathematics stereotype threat through salience of group achieve- Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat
ments. Journal of Experimental Social Psychology, 39, 83–90. and women’s math performance. Journal of Experimental Social
doi:10.1016/S0022-1031(02)00513-9. Psychology, 35, 4–28. doi:10.1006/jesp.1998.1373.
Murphy, M. C., Steele, C. M., & Gross, J. J. (2007). Signaling threat: Steele, C. M. (1997). A threat in the air: How stereotypes shape
How situational cues affect women in math, science, and engi- intellectual identity and performance. American Psychologist,
neering settings. Psychological Science, 18, 879–885. 52, 613–629. doi:10.1037/0003-066X.52.6.613.
doi:10.1111/j.1467-9280.2007.01995. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intel-
Prislin, R. (2009). Dynamics of change: Minority influence makes the lectual test performance of African Americans. Journal of
world go around. In M. Hewstone & R. Martin (Eds.), Minority Personality and Social Psychology, 69, 797–811. doi:10.1037/
influence and innovation: Antecedents, processes, and conse- 0022-3514.69.5.797.
quences (pp. 285–312). New York, NY: Psychology Press. Steele, C. M., Spencer, S. J., & Aronson, J. (2002). Contending with
Rabinovich, A., Morton, T. A., Postmes, T., & Verplanken, B. (2011). group image: The psychology of stereotype and social identity
The collective self and individual choice: The effects of inter- threat. In M. Zanna (Ed.), Advances in Experimental Social
group comparative context on environmental values and behavior. Psychology (pp. 379–440). San Diego, CA: Academic.
British Journal of Social Psychology. doi:10.1111/j.2044- The College Board. (2011). College bound seniors national profile
8309.2011.02022. reports of SAT program test takers.
Rosenthal, H. E. S., & Crisp, R. J. (2006). Reducing stereotype Walton, G. M., & Cohen, G. L. (2007). A question of belonging: Race,
threat by blurring intergroup boundaries. Personality and social fit, and achievement. Journal of Personality and Social
Social Psychology Bulletin, 32, 501–511. doi:10.1177/ Psychology, 92, 82–96.
0146167205281009. Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of
Rydell, B. J., McConnell, A. R., & Beilock, S. L. (2009). Multiple stereotypic beliefs. Journal of Personality and Social Psychology,
social identities and stereotype threat: Imbalance, accessibility, 45, 961–977. doi:10.1037/0022-3514.45.5.961.
and working memory. Journal of Personality and Social Willingham, W. W., & Cole, N. S. (1997). Gender and fair assessment.
Psychology, 96, 949–966. doi:10.1037/a0014846. Mahwah: Lawrence Erlbaum.

View publication stats

You might also like