Imagining oneself in a stereotyped role may stifle generalized tendencies to support social change

Abstract

Imagining oneself in a stereotyped role may not only increase women's endorsement of stereotypes about women and science, but also stifle broader concerns about social change. In the experiment, 81 women imagined themselves on a stereotypical or a counter-stereotypical career path (vs. a control condition). Participants in the stereotypical imagery condition endorsed to a higher extent the stereotypes about women and science, and crucially, were more resistant to social change in general. Stereotype endorsement mediated the relationship between exposure to stereotypes and resistance to social change. Results imply that tackling occupational gender stereotypes is crucial not only because they exclude women from male-dominated careers, but also because of a potentially pervasive negative impact on broader egalitarian concerns.

Keywords::

• stereotypes
• social change resistance
• STEM

The underrepresentation of women in STEM (Science, Technology, Engineering, and Maths) fields is a social issue of public relevance. In Europe, for example, women represent 55% of all University students, but they are still underrepresented in all STEM fields, as only 31% of first level students and 36% of PhD students in STEM disciplines are women (European Commission [EC], 2009). Moreover, on average across Europe only 32% of scientists and engineers are women (EC, 2009). This gender gap is being challenged by educational campaigns, such as those promoted by the European Platform of Women Scientists, or the US Committee on Women in Science, Engineering, and Medicine (CWSEM). While the underrepresentation of women in STEM is a key problem for gender equity and for society at large, the effects of this gender gap may be more generalized. In this article, we argue that exposure to gender role stereotypes has the potential to affect not only women's attitudes and beliefs in a related domain (e.g., women in STEM fields), but has a more generalized negative impact by reducing the recognition of inequality in broader areas of society—even those that are ostensibly unrelated to gender.

Women and stereotypes in science

A growing corpus of research suggests that continuous exposure to the unequal distribution of women and men across different educational and career fields (e.g., women are overrepresented in nursing studies and underrepresented in STEM fields) has the potential to affect women's attitudes and beliefs in a related domain (in this case, women in science). This idea is supported by the general theorizing that social stereotypes are inferred and generated based on information about the status that the social groups have in society (Eagly, 1987; Eagly & Steffen, 1984), and by the empirical evidence showing that exposure to gender stereotypes can impact women's cognition and behavior. For example, with respect to women and science it has been demonstrated that exposure to gender stereotypes impedes cognitive performance in school-aged girls (Huguet & Régner, 2007, 2009), encourages greater preference for feminine occupations in gender-identified women (Oswald, 2008), and affects women and men's educational and occupational choices (Eccles, Jacobs, & Harold, 1990).

Importantly, exposure to stereotypes can also have ideological consequences, and this is supported by evidence showing that priming women with social identities based on stereotypes reduces intentions of engaging in collective actions (Foster, 1999; Foster, Arnt, & Honkola, 2004). Similarly, exposing women to complementary gender stereotypes has been found to boost their support for the status quo (Jost & Kay, 2005), and to decrease their intentions of engaging in collective actions against gender inequalities (Becker & Wright, 2011). Engaging with stereotyped roles in education (e.g., women are supposed to be nurses, rather than scientists) can therefore have an impact on women's attitudes and beliefs concerning a proximal domain (in this case, STEM fields), by fostering greater endorsement of negative stereotypes about women in science. However, such engagement may also have a broader impact on women's attitudes and cognition in more distal domains by promoting a more generalized failure to recognize inequality in broader areas of society.

Ideological flexibility

We base the notion that there may be broader ideological impacts of exposure to gender stereotypes on the Categorization-Processing-Adaptation-Generalization model (CPAG; Crisp & Meleady, 2012; Crisp & Turner, 2011). Based on an integration of principles of multiple social categorization (Crisp & Hewstone, 2007) and bicultural identity integration (Benet-Martínez, Lee, & Leu, 2006) theories, the model provides a framework for understanding the impact of stereotypical and counter-stereotypical experiences on broader cognitive functioning. According to the model, when individuals are exposed to counter-stereotypes this not only changes stereotype content, but also elicits more general changes to cognitive, attitudinal, and ideological flexibility. Consistent with this model, studies have demonstrated that engaging with counter-stereotypes enhances cognitive flexibility (Gocłowska, Crisp, & Labuschagne, 2013), resilience to stereotype threat (Crisp, Bache, & Maitner, 2009), and also fosters greater intergroup tolerance (Vasiljevic & Crisp, 2013). However, the converse relation implied by the CPAG model is also relevant: exposure to stereotypes can cement rigid, stereotypic ways of thinking. Recent findings support this idea, particularly studies showing that priming a racially essentialist mindset can promote generalized close-mindedness (Tadmor, Chao, Hung, & Polzer, 2013), and that cognitive flexibility is negatively associated with generalized resistance to organizational change (Chung, Su, & Su, 2012).

With respect to women and STEM, uncovering evidence for this link has considerable implications for efforts to combat social inequality. The CPAG model suggests that exposure to stereotypes will stimulate rigid and conservative ways of thinking. If this is the case, then over-exposure to stereotypes in education may not only stifle women's willingness to challenge existing stereotypes relevant to one's own career, but more generally stifle their willingness to support broader social change. In this experiment, we asked women to imagine themselves on a stereotypical vs. counter-stereotypical career path. By doing so, participants' selves are salient and actively involved in the task. While self-relevance of the stereotype is not an essential precondition for stereotypes to affect cognition or behavior, stimuli that are self-relevant have easier access to chronic and automatic processing (Shih, Ambady, Richeson, Fujita, & Gray, 2002), and as such this might make the task more effective. Moreover, this manipulation is also high in ecological validity, in that imagining ourselves in various environments or career paths is a common mental simulation.

The first hypothesis is that asking women to imagine a stereotypical experience (i.e., imagining themselves in a female-dominated academic field) will activate rigid ways of thinking. As such, as compared to a control condition, this experimental task will elicit the following negative effects: (a) decreased group-based anger, (b) increased endorsement of stereotypes about women in STEM, (c) increased Opposition to Group Equality, and finally d) reduced willingness to support broader social change.

The second research question, instead, concerns whether asking women to imagine a counter-stereotypical experience (i.e., imagining themselves in a male-dominated academic field) will elicit beneficial effects. As mentioned above, there is evidence that exposure to counter-stereotypes can stimulate flexibility and lateral thinking (e.g., Gocłowska et al., 2013; Vasiljevic & Crisp, 2013), with potential spillover effects on other domains. Thus, there is ground to hypothesize that, as compared to a control condition, this experimental manipulation might elicit ideological flexibility effects as follows: (a) increased group-based anger, (b) decreased endorsement of stereotypes about women in STEM, (c) decreased Opposition to Group Equality, and finally (d) increased willingness to support broader social change. Alternatively though, taking this imagery task might not elicit any significant effects as compared to a neutral control condition, as the benefits that such experiences can elicit often depend upon the extent of prior exposure (Crisp & Turner, 2011), and can sometimes be ineffective and depleting when too challenging (Hutter & Crisp, 2006; Vasiljevic & Crisp, 2013).

Method

Participants and design

Eighty-one female students aged between 17 and 43 (M = 19.26, SD = 2.84) were recruited on an opportunity basis at an English university, and were compensated with course credit for their participation. A majority of participants were enrolled in a Psychology or in a dual Psychology degree (n = 73), whereas a minority of participants were enrolled in a social sciences or sociology degree (n = 4) or in an English or dual English degree (n = 3). One participant did not provide information about her subject of study.

The study was conducted in the laboratory with Qualtrics (Qualtrics Labs, Inc., Provo, UT), and it took approximately 10 min to be completed. Participants were randomly allocated to one of three conditions: stereotypical imagery prime, counter-stereotypical imagery prime, and control condition. For the manipulation we used an imagery prime, as mental imagery is known to foster similar responses to an actual experience, both on a behavioral and on an attitudinal level (for a review see Crisp, Birtel, & Meleady, 2011; Crisp & Turner, 2009, 2012).

In the stereotypical experience imagery condition, participants were presented with these instructions:

Please try to imagine that you are a Nursing student. Imagine what you think it would be like, in particular, to be a woman studying Nursing (i.e., what would be your everyday experiences interacting with other students) and describe it briefly.

In the counter-stereotypical imagery condition, participants were asked to imagine they were a Computer Science student, whereas in the control condition they were asked to imagine an outdoor scene. This control condition is commonly employed in mental simulation paradigms, such as the imagined contact literature (Crisp, Stathi, Turner, & Husnu, 2009). After the manipulation, participants were asked to complete a battery of questionnaires, and on completion of the study they were thanked and debriefed.

Measures

Participants completed a range of measures that focused on stereotypes about women and science, emotions about women in science, and more general tendencies to recognize group inequality and support social change. These included the following:

Group-based anger. We measured group-based anger with an adapted version of the four-item scale from van Zomeren, Spears, Fischer, and Leach (2004), measured on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). The following is a sample item: “When I think about the position of women in STEM fields I feel irritated.” The scale obtained Cronbach's α of .955.

Stereotype endorsement. This scale is based on the “mathematics as a male domain” subscale from the Modified Fennema–Sherman Mathematics Attitude Scales (Mulhern & Rae, 1998). Following Delisle, Guay, Senécal, and Larose (2009), the items were adapted to assess stereotypes in science by changing the word “mathematics” to the word “science.” The scale is composed of 12 items measured on a 7-points Likert scale (1 = strongly disagree, 7 = strongly agree). The following is a sample item: “It’s hard to believe a female could be a genius in science.” The scale obtained Cronbach's α of .724.

Opposition to equality. The OEQ scale is one of the two dimensions measured by the Social Dominance Orientation scale (Pratto, Sidanius, Stallworth, & Malle, 1994). The questionnaire measures disagreement with attitude statements that are supportive of efforts toward group equality. The scale is composed of seven items measured on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). The following is a sample item: “Group equality should be our ideal” (reversed). The scale obtained Cronbach's α of .922.

Resistance to change. Resistance to change was assessed, as in Jost et al. (2007), with two items measured on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). The scale measures the extent to which participants oppose social change. The following is a sample item: “I would be reluctant to make any large-scale changes to the social order.” The Spearman–Brown coefficient for the scale was .513.

Results

Means, standard deviations, and correlations are reported in Tables 1 and 2. There were a total of six missing values (distributed across four items), and missing values were random and not systematic. Indeed, Little's (1998) MCAR test was non-significant, χ²(91) = 96.16, p = .336. Data were analyzed with a set of two a-priori contrasts (Rosenthal & Rosnow, 1985), which reflect the hypotheses previously introduced. In the first planned contrast, the stereotypical imagery prime was compared to the control condition, while the second planned contrast compared the counter-stereotypical imagery prime to the control condition.

Table 1 Correlations between dependent variables.

	Stereotype endorsement	Resistance to change	Opposition to group equality
Resistance to change	.342^*^*
Opposition to group equality	.256^*	.223^*
Group-based anger	− .068	− .220^+	.001

Notes: ⁺p < .10, ^*p < .05, ^**p < .01.

Table 2 Means and standard deviations for each dependent variable across experimental conditions.

	Control	Stereotypical experience imagery condition	Counter-stereotypical imagery condition	Total
Stereotype endorsement	1.85 (0.44)	2.14 (0.60)	1.80 (0.50)	1.93 (0.53)
Opposition to group equality	2.89 (1.73)	2.39 (1.57)	2.50 (1.42)	2.60 (1.57)
Resistance to change	4.28 (1.75)	5.28 (1.71)	4.54 (2.24)	4.70 (1.94)
Group-based anger	2.64 (1.52)	2.24 (1.57)	3.01 (1.44)	2.64 (1.53)

Contrast 1 analysis revealed that exposure to the stereotypical imagery condition led to more stereotype endorsement compared to the control condition, t(77) = 1.98, p = .051, d = .551. Moreover, exposure to the stereotypical imagery condition led to greater resistance to change compared to the control condition, t(78) = 1.92, p = .059, d = .578, although the effect was only marginally significant. Finally, exposure to the stereotypical condition did not encourage greater opposition to equality, t(78) = 116, p = .249, d = − .303, nor lower group-based anger compared to the control condition, t(74) = 0.95, p = .345, d = − .259.

Contrast 2 analysis revealed that exposure to the counter-stereotypical imagery condition had no impact on stereotype endorsement (t(77) = 0.41, p = .685, d = − .106), resistance to change (t(78) = − .50, p = .620, d = .129), opposition to equality (t(77) = − 0.91, p = .365, d = − .246), or group-based anger (t(74) = 0.95, p = .345, d = .250), as compared to the control condition.1

Mediational analysis

A mediational analysis was performed to assess whether the effect of imagining a stereotypical experience on resistance to social change was mediated by variations in stereotype endorsement. Bootstrapping analysis was conducted using the SPSS macro “indirect” (Preacher & Hayes, 2008), and the IV was recoded in order to compare the stereotypical imagery condition to the control condition (i.e., the first planned contrast).Bootstrap estimates that follow are based on 5000 bootstrap samples. Results revealed that participants in the stereotypical imagery condition reported higher resistance to social change than participants in the control group, c = 1.00, SE = .4704, p = .038, whereas the direct effect was no longer significant, c′ = .6867, SE = .4559, p = .138. Bootstrap analysis revealed that the total indirect effect through the mediator was .3249, SE = .2398, 95% CI = +.0035+.9563, revealing a significant mediation effect (see Figure 1 for the full mediational model). Thus, stereotype endorsement mediated the association between imagery condition and resistance to change, as such that exposure to the stereotypical imagery condition increased resistance to change through increase in stereotype endorsement.2 Anger did not mediate the effects of stereotypical imagery condition on resistance to social change.

Figure 1 Stereotype endorsement as a mediator of the relationship between experimental condition and resistance to change.

Discussion

Results showed that asking women to imagine themselves on a stereotypical career path encouraged greater endorsement of the negative stereotype about women and science, and it also suggests that it stimulates greater reluctance to support general social change. Results also revealed that variations in stereotype endorsement explained the effect of stereotype imagery on resistance to change. Surprisingly, the manipulation had no effects on opposition to equality. This might be due to resistance to change and opposition to equality relying on different epistemic needs (Jost et al., 2007), where the former seems to be associated with the need for certainty, and the latter with the need to manage threat. This suggests that cognitive flexibility (or rigidity) is connected to the need for closure and certainty, rather than perceived threat. Finally, the manipulation had no effects on group-based anger about the situation of women in STEM. This might be due to the experimental procedure employed in the study. Indeed, the procedure did not provide participants with a description of the situation of women in STEM, and as such participants might have been oblivious about this social issue. A future replication of this study should consider providing participants with some background information about the issue of the underrepresentation of women in STEM.

Asking women to imagine themselves on a counter-stereotypical career path did not elicit any beneficial effects as compared to the control condition. As discussed in the introduction to the study, while there is evidence that imagining counter-stereotypical individuals can stimulate superior cognitive flexibility (Gocłowska et al., 2013; Vasiljevic & Crisp, 2013), this is not always the case. For example, when the imagery tasks are too demanding its beneficial effects are actually nullified (Vasiljevic & Crisp, 2013, Experiment 4). Also, our manipulation asked participants to imagine themselves on a male-dominated career path, and this might have activated concepts related to challenging stereotypes (e.g., barriers, obstacles, and potential threats to the self) that would otherwise not be activated with simple exposure to other counter-stereotypical individuals. Such self-relevant challenges might then mask the beneficial effects of exposure to counter-stereotypical experiences that have been detected by others. Thus, future replications should remove the active-self from the manipulation, and this could tackle other limitations, as will be discussed further in the limitations section.

Theoretical implications

The present results offer initial support to the hypothesis that imagining oneself in a stereotyped role has negative effects on women's endorsement of gender stereotypes in that domain, as well as on general egalitarian concerns unrelated to the gender domain. Thus, promoting a stereotypical mindset has the potential to “spill out” of the relevant intergroup-relation domain. This is consistent with the evidence showing that priming racial essentialism can inhibit the creative process (Tadmor et al., 2013), and that lower cognitive flexibility is associated with greater resistance to organizational change (Chung et al., 2012). Altogether the evidence supports the prediction generated by the CPAG model (Crisp & Turner, 2011): Just like counter-stereotypes can promote a flexible mindset (Gocłowska et al., 2013; Vasiljevic & Crisp, 2013), stereotypes can elicit the opposite effect, that is, a rigid and categorical mindset.

This research extends the work of Jost and Kay (2005) by demonstrating that imagining oneself in a gender stereotyped role has the potential to affect various ideological outcomes, and thus these stereotypes do not need to be complementary in nature in order to stimulate support for the status quo. Moreover, the results support the idea that very specific gender educational stereotypes (i.e., women are supposed to be Nurses rather than Computer scientists) are enough to elicit the negative ideological effects hereby explored, and as such they are particularly relevant to the discourse on education and women in STEM. The underrepresentation of women in STEM fields then, not only encourages and supports women's endorsement of stereotypes about themselves and science (Eagly & Steffen, 1984), but may also have broader impacts on women's general ability to recognize unequal relationships in broader areas of society. This is consistent with the idea that stereotyping, and also self-stereotyping, maintain and justify the system, which explains why members from low status groups (in this case, women) contribute as well to the system of status differences by endorsing the current state of affairs and failing to challenge social change (Jost & Banaji, 1994).

Given the chronic underrepresentation of women in STEM fields, and other male-dominated fields such as business, it is vital to explore the proximal and more distal impact of gender roles stereotypes in education on women, as they represent the disadvantaged group in this equation. However, the decision to include only women in this investigation comes with some limitations, as it widely accepted that the status quo is determined and endorses by both groups (Jost & Banaji, 1994). As our task required participants' self-involvement, the manipulation would probably elicit at least partially different effects on a male sample. This is because the effects would be confounded by a status valence. Indeed, for men imagining being a nurse student would not only be counter-stereotypical, but it would also be associated with lower status. This might then actually elicit greater resistance to change, or possibly mask the findings. Future investigations should adapt the manipulation for a male population, in order to explore the effect in both groups.

If substantiated, the implications of this phenomenon are numerous. We know that the negative effects of stereotype priming can occur regardless of the recipient awareness of the stereotype activation (Devine, 1989; Wheeler & Petty, 2001), and regardless of the extent of stereotype endorsement (Huguet & Régner, 2009), thus implying that all stereotypes need in order to be harmful is for the recipient to be merely aware of the existence of the stereotype. If exposure to stereotypes facilitates greater acceptance of the status quo and greater non-response to social inequalities, we might expect these general ideological beliefs to have an impact on more distal—but related—equality concerns, such as environmentally related concerns or even attitudes towards non-human animals. Indeed, pro-environmental attitudes are believed to be part of general egalitarian social relationships (Winter, 2000), and the endorsement of social hierarchy and inequality is an ideological predictor of both higher intergroup bias and conflict (e.g., Duckitt, 2006; Esses, Jackson, & Armstrong, 1998), and of stronger beliefs in the human-animal divide (Costello & Hodson, 2010). Future research should explore the relationship between stereotype priming and these broader egalitarian concerns.

Limitations

An important limitation concerns the bidirectional mediation model explored in the study. Indeed, analyses revealed that imagining oneself in a stereotypical career path stimulated women's resistance to change via endorsement of stereotypes about women in STEM, and it also stimulated their endorsement of such stereotypes, via resistance to change. Our hypothesis derives from a system-justification perspective, where stereotypes serve ideological functions (Jost & Banaji, 1994), and they encourage the acceptance of the current state of affairs, thus potentially restraining general support for change. As such, the model reported in Figure 1 best fits the theoretical perspective hereby presented. Importantly, this mutual relationship might have been caused by the simultaneous measurement of the variables involved, or by the similarity of measurement methodology. It might be advisable for future investigations to use multidimensional measures, for example employing an implicit association test to measure stereotype endorsement, rather than an explicit endorsement scale. Lastly, it is also methodologically plausible that the bidirectional relationship is caused by an unmeasured external variable. As such, future replications should consider measuring also some relevant cognitive antecedents, e.g., the previously mentioned need for certainty (Jost et al., 2007) and cognitive flexibility (Crisp & Turner, 2011).

Another limitation previously mentioned lies with the active-self quality of the manipulation, where participants were asked to imagine a personal experience. As such, this quality makes our manipulation different from previously employed stereotype or counter-stereotype primes. As discussed earlier, this quality might have activated other relevant concepts, especially in the counter-stereotypical imagery prime condition, such as potential challenges, barriers, and threats to the self. Furthermore, the active-self quality of the manipulation might carry another potential limitation, where the experimental prime might have not activated the same constructs for all participants. Indeed, according to the Active-Self Account of Prime-to-Behavior effects, while most primes will activate the same concepts and traits in most people, primes that rely on the self will necessarily access also individuated self-relevant constructs (Wheeler, DeMarree, & Petty, 2007). Our sample was relatively homogenous, given that all participants were women studying in non-STEM fields, which is a relevant group to test our hypotheses on. However, there might be interactions with other self-relevant activated concepts that were unexplored in this study. For example, in this case, accessing gender occupation-related stereotypes might interact with participants' gender identification, where low gender identified women could be less prone to access stereotypical content concerning their gender occupational opportunities.

Conclusions

The differential representation of women and men in STEM fields has the potential to reinforce gender roles and more general gender stereotypes. Specifically, the unequal representation of men and women across different education and employment fields (e.g., women being overrepresented in nursing careers and underrepresented in STEM fields) excludes women from male-dominated careers, and supports and encourages endorsement of gender role stereotypes. In current societies, science is an agent of power and it is connected with the most influential institutions, that is, the state and education (Fox, 2006). Thus, gender stratifications in science ultimately have the potential to legitimize unequal relationships between genders within society at large (Fox, 2006). Our findings suggest that challenging gender stratification in STEM fields is crucial both because of gender equity concerns, but also because such stratification might promote generalized tendencies to accept the status quo, with potential negative consequences for efforts to promote and encourage equality and fairness in society.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

^1. One-way ANOVAs on the outcome measures produced the following results: stereotype endorsement F(2, 79) = 3.23, p = .045, resistance to social change F(2, 80) = 1.99, p = .144, opposition to group equality F(2, 79) = 0.75, p = .476, and group-based anger F(2, 76) = 1.64, p = .201.

^2. We also tested the opposite causal path with resistance to change as the mediator, and stereotype endorsement as the dependent variable. Results revealed that participants in the stereotypical imagery condition reported higher stereotype endorsement than participants in the control group, c = .2813, SE = .1449, p = .058, although this effect was only marginally significant. The direct effect was not significant, c′ = .1661, SE = .1436, p = .253. Bootstrap analysis revealed that the total indirect effect through the mediator was .1152, SE = .0759, 95% CI = +.0101+.3205, revealing a significant mediation effect. Resistance to change mediated the association between imagery prime and stereotype endorsement, thus the mediational relationship was bidirectional.