Edited by:
Reviewed by:
*Correspondence:
This article was submitted to Developmental Psychology, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Women’s underrepresentation in science, technology, engineering, and mathematics (STEM) fields is a prominent concern in our society and many others. Closer inspection of this phenomenon reveals a more nuanced picture, however, with women achieving parity with men at the Ph.D. level in certain STEM fields, while also being underrepresented in some
A recent article in
Gender disparity in academia has been a generative topic of research for many years, with contemporary focus on this issue largely centering on men’s and women’s participation in (natural) sciences, technology, engineering, and mathematics (STEM). The general phenomenon is clear: on average, female representation in STEM fields (particularly those that are math-intensive) is lower than in the social sciences and humanities (SocSci/Hum). Though the magnitude of this gap has largely decreased across the last several decades, the difference is still reliable, prompting a number of efforts to explain it (for reviews, see
The low number of women in STEM is indeed of real concern. However, it is also important to observe that there is at least as much variation in female representation
Percent of American Ph.D.’s earned by women in 2011* in science, technology, engineering, and mathematics (STEM) and Social science/Humanities fields.
STEM Field | % of Ph.D.’s who are Female | Social science/Humanities field | % of Ph.D.’s who are Female |
---|---|---|---|
Physics | 18.0 | Music theory and composition | 15.8 |
Computer science | 18.6 | Philosophy | 31.4 |
Engineering | 22.2 | Economics | 34.4 |
Mathematics | 28.6 | Middle Eastern studies | 38.1 |
Astronomy | 29.2 | Classics | 41.8 |
Earth sciences | 36.2 | Political science | 43.1 |
Chemistry | 37.8 | History | 45.0 |
Statistics | 41.6 | Archeology | 52.3 |
Biochemistry | 45.4 | Linguistics | 59.2 |
Neuroscience | 49.4 | Anthropology | 59.6 |
Evolutionary biology | 49.8 | Spanish/Spanish literature | 59.9 |
Molecular biology | 54.4 | Comparative literature | 60.9 |
Sociology | 61.3 | ||
English literature | 62.4 | ||
Communication studies | 64.2 | ||
Education | 69.3 | ||
Psychology | 72.1 | ||
Art history | 76.8 |
In a recent study, we sought to test whether the FABs held by
Results from
We can also formulate a more detailed hypothesis here: people with more exposure to the fields in question (e.g., via college classes) will have FABs that predict female representation more precisely than the beliefs of people with less exposure. We expect this to be the case because the FABs of those with more exposure to a discipline will likely be more similar to those of the practitioners of that discipline, and are thus more likely to be similar to the kinds of beliefs that students will encounter and absorb as they start to consider higher education and careers in these fields (
A final aim of Study 1 was to address an alternative explanation for the hypothesized relationship between ability beliefs and female representation. As we have argued, underlying the main predictions described above is our claim that FABs influence women’s academic and career choices. However, might laypeople’s beliefs be simply inferred from their pre-existing knowledge about the proportion of women in the different fields? For instance, our participants—particularly those who have had college experience in the relevant fields and thus have had the opportunity to witness gender disparities firsthand—might rely on stereotypes against women’s intellectual abilities to arrive at the conclusion that fields with few women must require high levels of such abilities, whereas fields with many women must not. To address this possibility, we asked participants to estimate the proportion of women in the fields under investigation, and assessed whether participants’ ability beliefs still predicted female representation independent from these estimates. If so, this would undermine the possibility that participants simply infer their ability beliefs from their estimates of the field’s diversity.
There are surely many other dimensions that vary among fields and influence the gender breakdown of the people who participate in them, and we do not claim that FAB is the only factor in determining academic gender gaps. Indeed, as we observed at the outset, other such factors have been evaluated extensively in prior studies (for reviews, see
Our first question addresses a potential alternative explanation for the predictive power of FABs. A critic might note that the extent to which mathematics is involved in a field appears to be particularly predictive of whether women are underrepresented or not: fields that are math-intensive attract and retain fewer women, with math-intensive STEM fields (e.g., engineering, math, or physics) characterized by the most extreme gender disparities (in comparison to STEM fields that are less math-intensive, like the life sciences, which often feature parity or even a predominance of women;
Results from Study 1 could bear on this question. If, as hypothesized we find that FABs are capable of predicting female representation across a variety of fields, including those unlikely to be thought of as drawing on mathematical skills (such as most social sciences and humanities disciplines), it is unlikely that these beliefs are
Finally, it is worth noting that, as in Study 1, college exposure may matter. Participants with college experience likely have more nuanced, differentiated beliefs both about which fields require mathematical skills and about which fields require intrinsic ability. We thus hypothesized that when looking specifically at individuals with college exposure, FABs would independently predict female representation over and above beliefs about math and verbal skills, supporting the idea that FAB can account for female representation across the academic spectrum.
Next, we turn to the issue of potential beliefs that may mediate the relationship between FABs and female representation. In particular, we explore the possibility that people’s beliefs about the importance of brilliance vs. effort for success in a field give rise to differentiated perceptions of the kind of atmosphere that field promotes. We focused our exploration on two important aspects of a field’s atmosphere that (1) could be plausibly inferred based on the field’s presumed emphasis on brilliance, and that (2) men and women have diverging attitudes toward: namely, the extent to which the field requires
Study 1 examined two main predictions, one broad and one more specific. Broadly speaking, we expected that there would be a relationship between laypeople’s FABs and female representation, such that fields believed to require brilliance would have fewer women. At a greater level of specificity, we expected that college exposure would differentiate the predictive power of FABs, such that the beliefs of those exposed to the fields during college would be particularly predictive. Finally, Study 1 also examined whether ability beliefs independently predicted female representation above and beyond people’s
Study 2 was designed to replicate the main findings of Study 1, and to extend the inquiry into additional beliefs that might relate to gender disparities. We made two predictions. First, we predicted that FABs would
Participants included 307 individuals recruited via Amazon’s Mechanical Turk (MTurk), an online crowd-sourcing platform
To avoid participant fatigue, we created three versions of the survey, each of which contained 10 of the 30 fields under investigation. (Fields were identical to those examined in our original study of academics (
Survey items for Study 1 and Study 2.
Being a top scholar of [field] requires a special aptitude that just can’t be taught. | |
If you want to succeed in [field], hard work alone just won’t cut it; you need to have an innate gift or talent. | |
With the right amount of effort and dedication, anyone can become a top scholar in [field]. (R) | |
When it comes to [field], the most important factors for success are motivation and sustained effort; raw ability is secondary. (R) | |
To succeed in [field] you have to be a special kind of person; not just anyone can be successful in it. (in Study 2 only.) | |
People who are successful in [field] are very different from ordinary people. (in Study 2 only.) | |
Please provide your best guess or estimate to this question: in the recent past, what percentage of doctoral (Ph.D.) degrees from American universities do you think have been earned by women in [field]? | |
Top-level success in [field] depends to a large extent on one’s verbal ability. | |
Top-level success in [field] depends to a large extent on one’s mathematical ability. | |
[Field] is a field in which you spend a lot of time working by yourself rather than being around other people. | |
[Field] is a field in which competition with others is much more common than collaboration. |
Next, a series of questions asked about participants’ academic exposure to the 10 fields, including whether they had had (1) a high school class, (2) a college class, and/or (3) a graduate-level class in each of them. Participants were also asked to estimate how many women had received American doctoral degrees in each field in the recent past, with 10 response options corresponding to 10% intervals ranging from 0 to 100%. A final set of questions asked about demographic information (gender, age, ethnicity, and race).
For each field, we calculated FAB scores by averaging scores across participants from the four ability belief questions. Higher scores indicated more emphasis on brilliance. Three separate FAB scores were calculated: (1) All Participants’ FAB (using data from all participants except those with graduate level experience in the field)
Study 1 tested two main predictions. First, we expected that participants’ FABs would be correlated with female representation regardless of participants’ level of direct prior exposure with the fields (via courses). Second, we predicted that beliefs held by individuals with college experience would nevertheless be predictive of female representation at a
To assess our first prediction, we examined the correlation between FABs and female representation. Any fields for which we received fewer than 10 participants in either the no-college-experience or college-experience samples were removed from the analysis; estimates based on so few participants would likely be unreliable. This resulted in 29 fields being retained for analysis. (The single removed field was neuroscience; only seven individuals reported college experience with this field.) As predicted, fields believed to require brilliance had lower female representation,
To address the second prediction, we separately examined beliefs held by people with college exposure and those held by people without college exposure. Beliefs of both groups were significantly negatively associated with female representation: College Exposure scores,
Finally, we added college-exposed and non-college-exposed participants’ estimates of female representation as predictors to the two regressions above. Consistent with our argument, the FABs of college-exposed participants remained a significant predictor of actual female representation even when adjusting for these participants’ estimates of female representation (β = -0.41, bootstrapped
Regressions predicting female representation using field-specific ability beliefs and estimates of female representation of participants with college experience (CE; Upper) and with no college experience (NCE; Lower), Study 1.
Predictor | |||||
---|---|---|---|---|---|
β | 0.686 | 15.87 | < 0.001 | ||
STEM indicator | 0.07 | 0.686 | |||
Estimate of female representation (CE) | 0.58 | 0.001 | |||
Field-specific ability beliefs (CE) | -0.41 | 0.043 | |||
β | 0.55 | 10.37 | <0.001 | ||
STEM indicator | -0.22 | 0.340 | |||
Estimate of female representation (NCE) | 0.44 | 0.023 | |||
Field-specific ability beliefs (NCE) | -0.30 | 0.257 |
We considered one final alternative interpretation, which applies particularly to the findings obtained with college-exposed individuals. Perhaps College Exposure FAB scores emphasize brilliance for fields where there are few women just because (1) men may be more likely than women to believe that brilliance is required for success, and (2) more men in the current sample may have taken college classes in disciplines where women are typically underrepresented. In other words, disciplines with lower female representation may have higher College Exposure FAB scores for the simple reason that male participants’ brilliance-focused ability beliefs are overrepresented in our sample for these disciplines. Consistent with this possibility, college-exposed men’s scores (
In sum, the results of Study 1 lend clear support to the predictions we derived from the FAB model: women are less likely to be represented in fields believed to require stable, innate ability. Furthermore, as predicted, the field-specific beliefs of people with college experience in our fields were predictive of female representation at a more detailed level than were the beliefs of those without college experience. To speculate, perhaps initially people hold a global belief that disciplines in the STEM family require innate skill; as a result, the predictive power of these initial, inchoate ability beliefs is mostly captured by the STEM vs. non-STEM distinction. It is only after exposure to the particularities of the fields and the beliefs of their practitioners that FABs take on independent predictive power in relation to female representation.
Study 2 provides an opportunity to replicate the above findings, and to further explore how gender breakdowns are related to field-differentiated beliefs about the types of skills and work that are required. Two predictions are central to Study 2. First, we expect that the FABs of participants with college experience will predict unique variance in female representation, above and beyond their beliefs about the role of mathematical or verbal skills. Second, we predict that participants’ assumptions about how much solitary and competitive work is required by individual fields will mediate the relationship between FABs and female representation.
Participants included 302 individuals recruited via Amazon’s MTurk, using the same inclusion criteria as in Study 1. Participants were compensated $0.95 for survey completion. Data were excluded from an additional 53 individuals who met one or more of the exclusion criteria used in Study 1: (1) failing to complete the survey, (2) answering an attention check question incorrectly, (3) having an IP address suggesting residence outside the U.S., and/or (4) having IP addresses indicating completion of similar studies (including our Study 1) in the past.
As in Study 1, three versions of the survey were created, each of which contained the same subsets of 10 of the 30 fields under investigation. Approximately equal numbers of subjects participated in the three versions, and assignment was random (Version 1,
As in Study 1, items were presented individually in random order with all 10 fields listed beneath each item. Participants again indicated their agreement with the statement as it applied to each of the 10 fields using a 7-point Likert scale (1 = strongly disagree to 7 = strongly agree, with eight as an option to indicate “don’t know”). Two attention-check questions were also included. The survey then ended with questions assessing high school, college, and graduate level exposure to each of the 10 fields, along with several demographic questions. (These questions were all identical to those in Study 1).
We calculated FAB scores by averaging scores across the six items, and then averaging within fields to create field-level scores. Three separate FAB scores were calculated reflecting (1) All Participants’ FAB (using data from all participants except those with graduate level experience in the field), (2) College Exposure FAB (using data from participants who had taken college, but not graduate level, courses in the field), and (3) No College Exposure FAB (using data from participants who had taken neither college nor graduate courses in the field). Scores for the six ability beliefs questions had high internal reliability (for all participants, α = 0.89; for College Exposure, α = 0.93; for No College Exposure, α = 0.87). Deletion of the last two items added for Study 2 did not improve scale reliability, indicating it was appropriate to include them as part of the FAB scale.
To explore whether Study 2 replicated the key finding that FABs predict female representation, we again examined correlations between FAB and percentage of female Ph.D. recipients. As before, fields with fewer than 10 participants reporting either college or no college exposure were removed. This resulted in 27 fields being retained for analysis. (Middle Eastern studies, neuroscience, and archeology were removed because they had College Exposure
We next examined the relationship between female representation and the extent to which a field is perceived as demanding verbal and mathematical skills. Beliefs about the need for verbal skills were positively associated with female representation: beliefs of all participants,
We then tested our prediction that FABs of individuals with college exposure would predict female representation independently from beliefs about the role of mathematical and verbal skills. If so, this would strengthen the claim that FABs tap into something distinct from people’s beliefs about which fields require mathematical aptitude. To assess this prediction, we added perceptions of the need for verbal and mathematical skill as variables in the two regressions predicting female representation. For the regression testing beliefs of those
Regressions predicting female representation using field-specific ability beliefs and beliefs about the importance of verbal and mathematical skill of participants with college experience (CE; Upper) and with no college experience (NCE; Lower), Study 2.
Predictor | |||||
---|---|---|---|---|---|
β | 0.52 | 6.05 | 0.002 | ||
STEM indicator | -0.11 | 0.747 | |||
Field-specific ability beliefs (CE) | -0.39 | 0.085 | |||
Verbal skill beliefs (CE) | 0.26 | 0.489 | |||
Mathematical skill beliefs (CE) | -0.06 | 0.820 | |||
β | 0.49 | 5.29 | 0.004 | ||
STEM indicator | -0.21 | 0.524 | |||
Field-specific ability Beliefs (NCE) | -0.14 | 0.516 | |||
Verbal skill beliefs (NCE) | 0.17 | 0.614 | |||
Mathematical skill beliefs (NCE) | -0.26 | 0.454 |
As in Study 1, we also calculated a gender-balanced FAB score to examine the possibility that differences in male and female participants’ ability beliefs and college experience were driving the effects observed for college-exposed participants. (To reiterate, the possibility being tested here is that College Exposure FAB scores in fields with fewer women are inflated simply because men may have ability beliefs that are more brilliance-oriented and may also be overrepresented in the college-exposure sample for these fields.) Again, the proportion of college-exposed male participants within each field was negatively related to female representation at the Ph.D. level,
Finally, we tested the prediction that beliefs about solo work and competitiveness would mediate the relationship between FABs and female representation. Consistent with our argument, a bootstrapped (1,000 replications) product-of-coefficients mediation analysis performed with the PROCESS procedure in SPSS 22 (
Women are underrepresented in many STEM fields, but the pattern of gender distribution is complex, and a substantial amount of variation also exists in non-STEM fields. An important aim of the current studies was to provide an account for the wide variability in female representation across the entire academic spectrum. We maintain that the FAB hypothesis provides such an account. This hypothesis predicts that women will be underrepresented in fields believed to emphasize brilliance and inherent ability as the key to success; this is because women are often stereotyped as lacking the same sort of innate intelligence as men, and thus women will be discouraged from participating in fields to the extent that these fields are perceived as requiring this type of intelligence. Prior research has provided support for the FAB hypothesis within higher academia (
Several additional findings from the present studies are worth highlighting. The ability beliefs of individuals who had college-level exposure to the fields in question predicted female representation even when controlling for whether a field was in STEM or not, indicating that college may provide a unique context for refinement and elaboration of beliefs about what fields require for success. Results also suggested that the ability beliefs of participants with college experience are not simply a byproduct of participants’ inferring these beliefs based on their prior knowledge of female representation (Study 1). Further, college-exposed participants’ ability beliefs capture something beyond perceptions of specific types of skills required for success, as FABs of college-exposed individuals did not reduce to beliefs about which fields require mathematical and/or verbal skills (Study 2).
Notably, these findings have important consequences for potential interventions to improve diversity, both in terms of timing and in terms of content. College may be a pivotal experience during which people’s FABs become entrenched, and start to conform to those of their instructors. This highlights the crucial role that college educators play in communicating these maladaptive beliefs—but also suggests that they may be able to play an active role in changing the relevant messages. In particular, our data suggest that instructors who want to promote diversity might aim to minimize discussion of innate talent, regardless of the domain of skills with which it is associated, and instead highlight the importance of effort, practice, and persistence to success in a field. Prior work on individuals’ achievement beliefs suggests that such growth-oriented messages can be relayed in a range of ways: by choice of adjectives (in particular by avoiding words like “brilliant,” “genius,” etc.; ,
The current studies also suggest that beliefs about solo and competitive work may mediate the relationship between ability beliefs and female representation. It is possible that this result reflects a process by which ability beliefs influence perceptions of what it is like to work in certain fields, which in turn may influence the participation of women in these fields. Of course, we acknowledge this is not the only possible pathway here; our mediation analyses were designed to test an a priori hypothesis regarding how ability beliefs relate to representation, but they cannot determine directionality. Similarly, causality regarding ability beliefs and female representation cannot be claimed from the current studies due to the correlational nature of the data. However, our theoretical model posits that ability beliefs do drive women’s career and educational choices, and recent experimental manipulations in our lab have provided evidence consistent with this causal claim (e.g.,
Further investigating the precise pathways by which non-academics’ ability beliefs influence participation is one important topic for future research. To begin, it is worth noting that young men and women often decide whether or not to pursue a field long before interacting with professors, graduate students, or any other active practitioners of that field (e.g.,
In summary, we have provided support for the FAB hypothesis, demonstrating that women tend to be underrepresented in fields believed to require innate intellectual talent for success. Our data also open up possibilities for future research on the pathways by which ability beliefs influence women’s participation. Finally, these studies point to possibilities for effective interventions. If the practitioners of fields with gender gaps made a concerted effort to highlight the role of sustained, long-term effort in achievement, the gender gaps in these fields may correspondingly be diminished.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We acknowledge that beliefs about solo/competitive work mediating FAB’s relationship with women’s representation represents only one possible causal pathway; it is also possible that people could perceive the solo/competitive nature of a field and then conclude that it requires raw ability. More generally, we also note that there are likely many more factors involved in the pathways that ultimately result in the observed field-by-field variation in women’s representation. More comprehensive exploration of these factors, as well as experimental work, will be needed to definitively establish how FABs influence the observed gender gaps.
All human subjects research reported in this paper was approved by the Institutional Review Board of the first author’s home institution.
Mechanical Turk offers a convenience sample rather than a fully nationally representative sample. Analyses of American MTurk workers have demonstrated that women are overrepresented, that workers are typically younger and more educated than average, and that Blacks and Hispanics are underrepresented (
We excluded data regarding individual fields if they were provided by people reporting graduate-level experience in that field. We did so because we wanted to exclude beliefs held by people with extensive familiarity with the field gained through graduate-level exposure, allowing the focus of the current study to be restricted only to individuals with no college experience vs. college experience with the field.
The indirect paths were again significant even when adjusting for beliefs about mathematical skills, both for college-exposed participants, ab = -9.21 (-22.11, -1.73) and for non-college-exposed participants, ab = -9.54 (-19.77, -2.89).