Men Women and Science - Stewart-Williams and Halsey 2018

Men, Women, and Science 1
1 Running head: MEN, WOMEN, AND SCIENCE
5 Men, Women, and Science
6 Why the Differences and What Should Be Done?
8 Steve Stewart-Williams
9 School of Psychology, University of Nottingham Malaysia Campus, Semenyih
10
11 Lewis G. Halsey
12 Department of Life Sciences, University of Roehampton, London
13
14
15
16 The authors would like to thank Volker Behrends, Tom Reeve, and Martha
17 Villegas-Montes for their perceptive comments on an early draft of this paper.
18 Correspondence concerning this article should be addressed to Steve Stewart-
19 Williams at the School of Psychology, University of Nottingham Malaysia Campus.
20 Email: steve.stewart-williams@nottingham.edu.my
21
22
23
24
25 Word count: 20,500
26
27 Abstract
28 It is a well-known and widely lamented fact that men outnumber women in a number
29 of fields in STEM, including physics, mathematics, and computer science. The most
30 common explanations for the gender gaps are discrimination and social norms, and
31 the most common policy prescriptions are targeted at these ostensible causes.
32 However, a great deal of evidence in the behavioral sciences suggests that
33 discrimination and social norms are only part of the story. Other plausible contributors
34 include relatively large mean sex differences in career and lifestyle preferences, and
35 relatively small mean differences in cognitive aptitudes – some favoring males, others
36 favoring females – which are associated with progressively larger differences the
37 further above the mean one looks. A more complete picture of the causes of the
38 unequal sex ratios in STEM may productively inform policy debates, and is likely to
39 improve women’s situation across the STEM fields.
40
41 Public Significance Statements
42 A review of the literature suggests that gender gaps in STEM fields have many
43 causes, including average sex differences in preferences and cognitive specialties.
44 Anti-female discrimination, though it demonstrably exists, is probably not the
45 primary cause of STEM gender gaps in modern Western nations.
46 Policy prescriptions founded on the notion of pervasive anti-female bias are
47 unlikely to have positive effects.
48
49 KEYWORDS: Discrimination; Equality; Gender; Sex Differences; STEM.
50
51 Men, Women, and Science
52 Why the Differences and What Should Be Done?
53
54 Never has the issue of gender disparities been as widely discussed, or as
55 bitterly contested, as it has been in recent years. From the Oscars to the political
56 podium, from TV shows to the workplace, disparities are identified and debate
57 inevitably ensues. In the occupational realm, one of the primary focuses of this debate
58 has been the differential representation of men and women in STEM (science,
59 technology, engineering, and math).1 This was epitomized by the infamous “Google
60 memo,” in which then-Google employee James Damore (2017) questioned the extent
61 to which observed gender disparities in STEM are a product of discrimination. The
62 memo, and Damore’s subsequent dismissal from Google, provoked a great deal of
63 discussion and debate about the nature and nurture of human sex differences.
64 Unfortunately, much of this debate was decidedly inaccurate in its presentation of the
65 scientific research on the topic. This is the case even of our most prestigious scientific
66 journals; Nature, for instance, ran an article claiming that sex differences in
67 occupational outcomes were due entirely to discrimination (Chachra, 2017). No
68 doubt, discrimination is part of the explanation. However, a great deal of research
69 suggests that other factors contribute as well to the relative dearth of women in
70 STEM, and that these factors quite possibly play a larger role than discrimination
71 does. In this article, we survey the various lines of evidence bearing on this issue.
1 Strictly speaking, the issue is not STEM per se, but rather STEM fields that have a strong spatial or
mathematical component. According to Ceci, Ginther, Kahn, and Williams (2014), the STEM fields
should be divided into GEEMP fields (geoscience, engineering, economics, mathematics/computer
science, and the physical sciences) and LPS fields (life science, psychology, and social science). Men
outnumber women in GEEMP fields, but women are at parity with men, or even sometimes outnumber
them, in LPS fields.
72 We divide the task into six parts. First, we look at research suggesting that
73 men and women differ, on average, in their career and lifestyle preferences, and that
74 these differences are due in part to biological causes. Second, we consider the
75 possibility that men and women differ, on average, in certain cognitive aptitudes –
76 that men, for instance, score somewhat higher on tests of spatial ability whereas
77 women score somewhat higher on verbal tests. Third, we look at the controversial
78 suggestion that men are more variable than women in cognitive ability, such that there
79 are more men at the top of the ability distribution, and more men as well at the
80 bottom. Fourth, we look at the issue of gender discrimination in STEM, and the extent
81 to which this explains why more men than women end up in certain fields. Fifth, we
82 look some of the major policy prescriptions aimed at shrinking the STEM gender gap
83 and redressing the gender imbalances. And sixth, we consider whether the appropriate
84 aim of such interventions is a 50:50 sex ratio in every STEM field or simply a level
85 playing field. In other words, should the ultimate goal be equality of outcome or
86 equality of opportunity?
87
88 Sex Differences in Preferences and Priorities
89 To begin with, we examine arguably the most important non-bias explanation
90 for sex differences in STEM representation: the idea that men and women differ – on
91 average – in certain career-relevant preferences. Specifically, we look at sex
92 differences in occupational preferences and sex differences in life priorities. After
93 that, we consider the extent to which these differences are products of nature or
94 nurture, and we look at the evolutionary pressures that might have helped to shape
95 them.
96 Occupational Preferences
97 A large literature in psychology shows that men and women differ, on
98 average, in the kinds of occupations that interest them (Morris, 2016). One of the most
99 important recent papers on this topic was a comprehensive meta-analysis by Su,
100 Rounds, and Armstrong (2009). The paper focused on two main areas: preferences for
101 general types of occupations (e.g., those focused on people vs. things) and preferences
102 for specific careers (e.g., engineering vs. mathematics). In both cases, the authors
103 found substantial sex differences that plausibly help to explain observed STEM
104 gender gaps.
105 Starting with general occupational types, by far the largest sex difference was
106 that for interest in “people jobs” vs. “things jobs.” People jobs are jobs that center on
107 interacting with other human beings; things jobs are jobs that center on working with
108 objects, machines, or abstract rules. Members of both sexes can be found at every
109 point on the people vs. things continuum; however, more men than women exhibit a
110 preference for things jobs, whereas more women than men exhibit a preference for
111 people jobs. Averaging across studies, Su et al. (2009) found an effect size of d = .93
112 for the people vs. things sex difference. This is notably larger than most human sex
113 differences (Lippa, 2009; Stewart-Williams & Thomas, 2013), and indeed than most
114 effects in psychology (Eagly, 1995). To get an intuitive sense of the magnitude of the
115 difference, if one were to pick pairs of people at random, one man and one woman,
116 the man would be more things-oriented than the woman around 75% of the time.
117 The people vs. things sex difference immediately suggests an explanation – or
118 rather a partial explanation – for the fact that men outnumber women in fields such as
119 physics, engineering, and mathematics, whereas women are at parity with or even
120 outnumber men in fields such as psychology and the social sciences: The former
121 fields are of interest to more men than women, and the latter more women than men,
122 and people tend to gravitate to fields that are of greater interest to them. This is not
123 merely a plausible-sounding speculation; Yang and Barth (2015) have shown that
124 interest in things vs. people accounts for much (although not all) of the variance in
125 people’s occupational outcomes.
126 Research looking at preferences for specific occupations leads to a similar
127 conclusion. As Su et al. (2009) report, males on average express considerably more
128 interest than females in engineering (d = 1.11), and somewhat more interest in science
129 and mathematics (d = .36 and .34, respectively). These differences are present by
130 early adolescence and closely match the observed numbers of men and women
131 working in the fields in question. Su et al. (2009) point out that, if we assume for the
132 sake of argument that people working in a given field tend to come from the 25% of
133 people most interested in that field, sex differences in occupational interests would
134 account for the entirety of the engineering gender gap and much of the gap in science
135 and mathematics. In short, sex differences in occupational preferences are far from
136 trivial, and plausibly make a substantial contribution to observed occupational gender
137 gaps.
138 Life Priorities
139 Gender gaps in STEM – and especially the higher echelons of STEM fields –
140 may also be shaped in part by average sex differences in life priorities. As with
141 occupational preferences, people vary a lot in their life priorities, and the full range of
142 priorities can be found within each sex. Nonetheless, some priorities are more
143 common among men than women, and others among women than men (Hakim, 2005;
144 Schwartz & Rubel, 2005). One longitudinal study, for instance, found that, among
145 adults identified as mathematically gifted in early adolescence, the average man
146 reported placing more importance on career success and income, whereas the average
147 woman reported placing more importance on work-life balance and making time for
148 one’s family and friends (Benbow, Lubinski, Shea, & Eftekhari-Sanjani, 2000;
149 Lubinski, Benbow, & Kell, 2014). These differences were particularly pronounced
150 among people with children, largely because women’s priorities changed after they
151 became mothers (Ferriman, Lubinski, & Benbow, 2009). Moreover, sex differences in
152 self-reported priorities were evident in real-world behaviour. For example, as
153 Lubinski et al. (2014) observed, over the course of the last fifteen years, the men in
154 their sample had worked an average of 51 hours per week, whereas the women had
155 worked an average of 40.
156 Of course, sex differences in lifestyle preferences do not explain why the sex
157 ratio is so much more male-biased in math-intensive STEM fields than in most others.
158 Still, the differences do plausibly help to explain the fact that, in STEM and
159 elsewhere, men outnumber women among the minority in the higher echelons: Rising
160 to the top is a priority for fewer women than men, and thus fewer women than men
161 are willing to make the sacrifices required to achieve that goal. To be clear, some
162 women are willing to make those sacrifices, and the majority of men are not.
163 However, more men than women are willing, and this is plausibly part of the reason
164 that the sex ratio at the top is so often male-biased. Note that, according to one large
165 US study (N ≈ 4,000), the sex difference in career-mindedness is not a result of
166 women thinking that career advancement is impossible for them. The average woman
167 views career advancement as just as achievable as the average man, but as less
168 desirable (Gino, Wilmuth, & Brooks, 2015).

169 The Nature and Nurture of Sex Differences in Occupational Preferences
170 Sex differences in occupational preferences suggest one possible reason that
171 more men than women go into math-intensive STEM fields. The reason, put simply,
172 is that more men than women want to go into these fields. To the extent that this is so,
173 it reduces the need to posit workplace discrimination as the root cause of gender
174 disparities in STEM. Still, even if occupational preferences explained all of the STEM
175 gender gap (which we are certainly not suggesting), this would not imply that
176 discrimination played no role. After all, it could be argued that the preference
177 differences themselves are products of discriminatory socialization practices and
178 sexist cultural norms. No doubt, social factors help to determine which occupations
179 people choose or end up in. However, several lines of evidence suggest that sex
180 differences in this arena are shaped to an important degree by unlearned biological
181 factors.
182 First, sex differences in occupational preferences have remained remarkably
183 stable throughout the half-century or so that psychologists have measured them, even
184 in the face of significant shifts in women’s social roles and place in society (Su et al.,
185 2009). In particular, the sex difference in interest in things vs. people seems
186 stubbornly resistant to change. One analysis, for instance, found that whereas the
187 number of women pursuing high-status professions increased a great deal since the
188 1970s, the number pursuing things-related professions remained virtually static
189 (Lippa, Preston, & Penner, 2014). Notably, this was the case despite the fact that,
190 during that same period, a wide range of initiatives were established to try to entice
191 women into those very professions.
192 Second, the same sex differences in occupational preferences have been found
193 in every society where psychologists have looked for them. In one large study (N ≈
194 200,000), Lippa (2010) found the differences in 53 out of 53 nations: a level of cross-
195 cultural unanimity almost unheard of within psychology. Importantly, the gender gap
196 in occupational preferences was no larger in nations with higher levels of gender
197 inequality, suggesting that gender inequality is not a major determinant of the gap.
198 Meanwhile, other research suggests that the gender gap in STEM employment (as
199 opposed to STEM-related job preferences) is actually smaller in more gender unequal
200 nations, perhaps in part because economic hardship means that people in those nations
201 have less scope to act on their personal preferences, and a greater need to place
202 financial security above self-fulfillment in choosing a suitable career (Stoet & Geary,
203 2018).
204 Third, at least some of the relevant sex differences appear in nascent form
205 early in the developmental process. The first glimmer of the people vs. things
206 difference, for instance, may be evident in the first few days of life. Connellan, Baron-
207 Cohen, Wheelwright, Batki, and Ahluwalia (2000) famously presented 102 newborn
208 babies with two objects, one after the other: a human face and a mechanical mobile.
209 Many babies looked for equal amounts of time at both. However, among those who
210 looked for longer at one than the other, more boys than girls looked for longer at the
211 mobile (43% vs. 17%), whereas more girls than boys looked for longer at the face
212 (36% vs. 25%; although see Spelke, 2005). A similar pattern has been observed in a
213 number of nonhuman primates: Newborn female macaques are more attentive to faces
214 than are newborn males (Simpson et al., 2016), and female macaques and vervet
215 monkeys spend more time playing with toy dolls provided by experimenters than do
216 males of their respective species (Alexander & Hines, 2002; Hassett, Siebert, &
217 Wallen, 2008).

218 Finally, occupational preferences have been convincingly linked to prenatal
219 hormonal exposure (Levy & Kimura, 2009). Girls with congenital adrenal
220 hyperplasia (or CAH), which involves exposure to high levels of testosterone in the
221 womb, tend to have more male-typical career preferences than other girls (Beltz,
222 Swanson, & Berenbaum, 2011; Berenbaum & Beltz, 2011). Similarly, in non-clinical
223 populations, the average prenatal testosterone levels of workers in a field are
224 negatively correlated with the proportion of women working in that field (Manning,
225 Reimers, Baron-Cohen, Wheelwright, & Fink, 2010). The correlations between
226 prenatal hormones and occupational outcomes tend to be relatively modest; in the
227 Manning et al. study, for instance, they range from around .1 to .6. This is presumably
228 partly a product of measurement error and partly because prenatal hormones are only
229 one influence among many. Nonetheless, in light of the hormonal data, and the other
230 data surveyed in this section, it seems reasonable to conclude that sex differences in
231 occupational preferences are not purely a product of culture or socialization. Biology
232 plays a role as well.
233 Evolutionary Rationale
234 Although we have good reason to think that there is an innate contribution to
235 men and women’s occupational and lifestyle preferences, we have much less idea why
236 this might be the case. The most plausible evolutionary explanation is for sex
237 differences in lifestyle preference. In most parental species, females invest more than
238 males into offspring (Janicke, Häderer, Lajeunesse, & Anthes, 2016; Trivers, 1972).
239 Among mammals, for instance, females gestate and nurse the young, and females
240 usually provide the bulk of the direct parental care. In our species, sex differences in
241 parental investment are comparatively modest: Both sexes tend to invest substantially
242 in their young, rather than only the females. But there is still a difference: Men in all
243 known cultures invest less into offspring than women, and this has probably been the
244 case for most of our evolutionary history (Stewart-Williams & Thomas, 2013).
245 Importantly, the minimum biological investment is also notably smaller for men than
246 for women. Women’s minimum is a nine-month pregnancy and – until recently –
247 several years of breastfeeding. Men’s minimum is the time and effort required to
248 impregnate the woman.
249 As a result of these sex differences in parental investment, ancestral men could
250 potentially produce many more offspring than ancestral women, simply by mating
251 with multiple partners (Clutton-Brock & Vincent, 1991). Consequently, human males
252 evolved to be more interested than females in mating with multiple partners (Schmitt,
253 2005; Schmitt & Project, 2003), less choosy about their low-commitment sexual
254 partners (Buss & Schmitt, 1993; Kenrick, Groth, Trost, & Sadalla, 1993), and – of
255 particular relevance to the present topic – more inclined to compete and take risks to
256 obtain the status and resources that helped to make them attractive to women (Byrnes,
257 Miller, & Schafer, 1999; Daly & Wilson, 2001; Deaner, 2013; M. Wilson & Daly,
258 1985). In light of this theoretical framework – which is well-supported by research in
259 other species (Andersson, 1994; Janicke et al., 2016) – it is little surprise that more
260 men than women prioritize the pursuit of status over family, whereas more women
261 than men prioritize family and work-life balance.
262 Evolutionary explanations for sex differences in occupational preferences are
263 somewhat more of a stretch. For most of human evolution, there were no scientists, no
264 technologists, no engineers or mathematicians. As such, any innate contribution to sex
265 differences in interest in these vocations must be a byproduct of traits selected for
266 other reasons. One possibility is that the differences trace back to the sexual division
267 of labor among our hunter-gatherer forebears, and specifically the fact that women
268 specialized in caring for the young whereas men specialized in hunting and perhaps
269 waging war with other groups (a division of labor found also among our close
270 relatives, the chimpanzees; Muller, Wrangham, & Pilbeam, 2017). To fit them to
271 these roles, women may have evolved a stronger attentiveness to the needs of the
272 young, and by extension, to people in general, whereas men may have evolved a
273 stronger interest in the tools used for hunting and warfare (Geary, 2010; Hrdy, 2009).
274 Sex differences in interest in people-focused vs. things-focused jobs may just be an
275 adaptively neutral side effect of these ancient, more primal differences. It is worth
276 emphasizing that, although this hypothesis seems reasonable enough, it has yet to be
277 rigorously tested. Regardless of the explanation, however, the evidence for an innate
278 contribution to the relevant sex differences is strong.
279
280 Sex Differences in Aptitudes
281 Sex differences in occupational preferences are not the only reason we might
282 expect uneven sex ratios in certain STEM fields, even if discrimination were entirely
283 removed from the picture. A second, more controversial suggestion is that STEM
284 gender gaps are due in part to average sex differences in a number of STEM-relevant
285 cognitive capacities, which result in somewhat more men than women having a
286 suitable profile of aptitudes for working in some STEM fields. This statement could
287 easily be misconstrued, so it is important to be absolutely clear what is meant.
288 First, the claim is not that men perform better than women in every cognitive
289 domain. On the contrary, men perform better in some domains whereas women
290 perform better in others. The best-known examples are that men score higher than
291 women on most tests of spatial ability, whereas women score higher than men on
292 most tests of language ability, including verbal comprehension, reading, and writing
293 (Guiso, Monte, Sapienza, & Zingales, 2008; Halpern, 2012; Reynolds, Scheiber,
294 Hajovsky, Schwartz, & Kaufman, 2015).
295 Second, even in areas where men do perform better, the claim is not that all
296 men – or even most – perform better than all or most women. As with occupational
297 preferences, members of both sexes vary enormously in every cognitive aptitude, and
298 the distribution for men overlaps almost entirely with that for women (Hyde, 2005).
299 However, for some aptitudes, the distribution for one sex is shifted somewhat to the
300 right of that for the other, such that the average score for the former is somewhat
301 higher than that for the latter. In saying this, it is worth stressing that the average score
302 does not describe all members of the group, or even the typical member, but merely
303 represents the central tendency within a broad array of scores. Most people fall above
304 or below the average.
305 Third, the claim is not that these cognitive sex differences are especially large.
306 On the contrary, at the center of the distribution, they tend to be rather small (Hyde,
307 2005; Hyde, Lindberg, Linn, Ellis, & Williams, 2008; Stewart-Williams & Thomas,
308 2013). The only reason they matter at all is that even small differences at the mean are
309 associated with progressively larger differences the further from the mean one looks
310 (see Figure 1). For jobs requiring normal-range abilities – including many STEM jobs
311 – cognitive sex differences are likely to make little difference: The pool of potential
312 female candidates is similar in size to the pool of males. However, for jobs requiring
313 exceptional abilities, the sex ratio of potential candidates may be skewed in favor of
314 one sex or the other – even when the relevant sex differences in the general population
315 are small or even negligible (Baye & Monseur, 2016; Halpern, 2012; Halpern et al.,
316 2007; Pinker, 2002).
317 --------------------Insert Figure 1 about here--------------------

318
319
320
321 Figure 1. For two groups, A and B, where the average score on a normally distributed
322 variable is higher for the former than the latter, the ratio of A-to-B gets progressively
323 larger the higher above the average one looks. For both groups, fewer and fewer
324 people occupy each segment above the mean, with the percentage decline getting
325 larger with each further step. For instance, the number of people falling between 1 and
326 2 standard deviations above the mean is 40% of that falling between the mean and 1
327 standard deviation, whereas the number of people falling between 2 and 3 standard
328 deviations above the mean is only 15% of that falling between 1 and 2 standard
329 deviations. This is the case for both groups. However, because the lower-scoring
330 group (B) starts this accelerating decline before the higher-scoring group (A), the
331 percentage decline for the lower-scoring group is always larger than that for the
332 higher-scoring one. The net result is that the ratio of A-to-B gets progressively larger
333 for scores above the mean. Meanwhile, the ratio of B-to-A gets progressively larger
334 for scores below the mean. The sex difference depicted in Figure 1 is large – a full
335 standard deviation between the average for each sex – and thus the skew quickly
336 becomes large as well. However, even small differences at the mean may be
337 associated with high levels of skew at sufficiently extreme levels of any given trait.
338
339 Fourth, the claim is not that women lack the cognitive talents to make it in
340 STEM. Most people lack the cognitive talents, and of those who do possess them,
341 some are men and some are women. The claim is simply that, because of small
342 average differences in a small subset of abilities, somewhat more men than women are
343 suited to work in some areas of STEM – and for the same reason, somewhat more
344 women than men are suited to work in others.
345 The Relevant Differences
346 With these important qualifications in mind, let us now consider some of the
347 sex differences in cognitive aptitudes that may help to explain the uneven sex ratios
348 found in some STEM areas. The first concerns spatial abilities. As mentioned, the
349 average score on most spatial tests is somewhat higher for men than for women. This
350 is a well-established finding (Voyer, Voyer, & Bryden, 1995), especially with respect
351 to mental rotation (Maeda & Yoon, 2013), and it seems reasonable to think it might be
352 part of the reason that somewhat more men than women gravitate to fields that require
353 above-average spatial abilities – fields such as physics and engineering. Consistent
354 with this suggestion, more than fifty years of research in educational psychology
355 indicates that spatial ability is indeed an important predictor of STEM success (Wai,
356 Lubinski, & Benbow, 2009).
357 A second, less widely appreciated cognitive sex difference is that, on average,
358 males as a group score higher than females as a group on tests of mechanical
359 reasoning – that is, tests of the ability to solve problems involving mechanical
360 principles and physical laws (Flores-Mendoza et al., 2013; Hedges & Nowell, 1995;
361 Lemos, Abad, Almeida, & Colom, 2013). Unlike most cognitive sex differences, this
362 one appears to be rather large, even at the mean, and is thus larger still at the right-
363 hand tail of the distribution (Hedges & Friedman, 1993). As with the spatial sex
364 difference, the difference in mechanical reasoning has clear implications for the
365 gender composition of fields such engineering and physics.
366 A third difference relates to mathematical ability; in this case, however, the
367 findings are much less straightforward. Sex differences in average mathematical
368 ability are generally small and highly variable: In some nations, boys do better on
369 tests of mathematical aptitude; in others, girls do (Baye & Monseur, 2016; Else-Quest,
370 Hyde, & Linn, 2010; Guiso, Monte, Sapienza, & Zingales, 2008; Hyde, Lindberg,
371 Linn, Ellis, & Williams, 2008; Stoet, Bailey, Moore, & Geary, 2016). At the same
372 time, though, there appear to be a number of math-related sex differences which may
373 be relevant to STEM outcomes. First, whereas females tend to do better in tests of
374 mathematical computation (Hyde, Fennema, & Lamon, 1990), males tend to do better
375 – at least from adolescence – in tests of mathematical reasoning (that is, tests of the
376 ability to solve problems using mathematical concepts; Benbow, 1988; Halpern, 2012;
377 Hyde et al., 1990). This is a potentially important finding because, as Hyde (1990)
378 notes, mathematical reasoning “is critical for success in many mathematics-related
379 fields, such as engineering and physics” (p. 151). Second, despite small and cross-
380 culturally variable differences in average scores, males consistently outnumber
381 females at the highest levels of mathematical performance (Baye & Monseur, 2016;
382 Makel, Wai, Peairs, & Putallaz, 2016; Reilly, Neumann, & Andrews, 2015; Wai,
383 Cacchio, Putallaz, & Makel, 2010). The upshot is that somewhat more men than
384 women are likely to have the mathematical aptitude to work in math-intensive STEM
385 fields – even though, once again, most men do not.
386 The gender gap in math-intensive fields may be amplified by the fact that,
387 among the minority of people who possess exceptional mathematical abilities, the
388 women are more likely than the men to possess exceptional language abilities as well
389 (Wang, Eccles, & Kenny, 2013). This means that mathematically gifted women have
390 more vocational options than their male counterparts, and consequently that fewer
391 mathematically gifted women end up pursuing a STEM career. To the extent that this
392 explains the gender gap in math-intensive fields, the gap results not from gifted
393 women having fewer options, but rather from them having more.
394 A final cognitive difference bearing on the question of STEM sex ratios relates
395 to Baron-Cohen’s (2003) distinction between systemizing and empathizing.
396 Systemizing refers to the desire and ability to understand or build “systems,”
397 including mechanical systems like cars, physical systems like galaxies, and abstract
398 systems like logic and mathematics. Empathizing, in contrast, refers to the desire and
399 ability to understand people: their thoughts, their desires, their feelings. Virtually
400 every human being possesses both abilities to a greater or lesser extent. On average,
401 though, men score somewhat higher than women on tests of systemizing, whereas
402 women score somewhat higher than men on tests of empathizing (Baron-Cohen,
403 Wheelwright, Hill, Raste, & Plumb, 2001). As usual, these differences are not large
404 among the majority in the normal range. However, among the minority of exceptional
405 empathizers, women outnumber men, and among the minority of exceptional
406 systemizers, men outnumber women (Baron-Cohen et al., 2014). It seems reasonable
407 to suppose that sex differences in systemizing and empathizing are part of the reason
408 that more men than women go into fields such as physics and engineering, whereas
409 more women than men go into fields such as psychology and education. Consistent
410 with this suggestion, Ruzich et al. (2015) observe that STEM workers score
411 consistently above workers in non-STEM fields on measures of systemizing ability.
412 Most commentators agree that sex differences in cognitive aptitudes are too
413 small to explain STEM gender gaps in their entirety, and that sex differences in
414 occupational preferences are a much more important contributor (Ceci, Williams, &
415 Barnett, 2009; Johnson, Carothers, & Deary, 2008). Still, the evidence for the
416 cognitive differences is robust, and it is perfectly plausible that they help to shape men
417 and women’s career choices and trajectories. Indeed, given the clear relevance of the
418 aptitudes in question to STEM, it would be surprising if this were not the case.
419 The Nature and Nurture of Sex Differences in Aptitudes
420 Where do these cognitive sex differences come from? As with sex differences
421 in occupational preferences, scientists do not yet have a complete or definitive
422 answer. Nonetheless, a range of evidence suggests that the differences are not due
423 solely to socialization, but rather have a substantial innate component. First, many of
424 the differences appear early in the developmental process. The sex difference in
425 mental rotation, for instance, can be detected by five months of age (Moore &
426 Johnson, 2008), and the sex difference in language ability can be detected by seven
427 months (Bando, López Bóo, & Li, 2016). These findings do not rule out purely
428 environmental explanations, but they do render them less plausible. At the very least,
429 they eliminate the possibility that social influences appearing after the age of one year
430 could provide a complete explanation for the differences. Certainly, sex differences in
431 cognition may sometimes get larger with time, which one might argue shows that the
432 differences are environmentally determined (see, e.g., Kersey, Braham, Csumitta,
433 Libertus, & Cantlon, 2018; Spelke, 2005). However, there are several responses to
434 this. One is that changes in the magnitude of sex differences are not necessarily a
435 product of social causes alone; to some extent, they may just reflect the natural
436 maturational process – especially when the changes coincide with the onset of
437 puberty. Another response is that, even if the changes were entirely a product of social
438 causes, this would not imply that the initial sex differences were a product of social
439 causes too. The initial differences could well still be innate.
440 Second, many of the traits under discussion have been linked to sex hormones.
441 Spatial ability, for instance, has been linked to both prenatal and pubertal testosterone:
442 Individuals exposed to higher levels of testosterone in the womb, and higher levels at
443 puberty, typically exhibit stronger spatial skills (Berenbaum & Beltz, 2011).
444 Similarly, testosterone supplements appear to boost spatial ability in female-to-male
445 transsexuals (although note that the studies demonstrating this effect tend to have
446 small samples; Slabbekoorn, van Goozen, Megens, Gooren, & Cohen-Kettenis, 1999).
447 In addition to spatial abilities, prenatal testosterone has been linked to higher
448 systemizing, and to lower social skills, empathizing, and verbal ability (Auyeung et
449 al., 2006; Chapman et al., 2006; Lutchmaya, Baron-Cohen, & Raggatt, 2001, 2002;
450 summarized in Baron-Cohen, 2003).
451 Third, sex differences in cognitive abilities appear to transcend cultural
452 boundaries. Across cultures, girls typically outperform boys on tests of linguistic
453 ability (Stoet & Geary, 2013), whereas boys typically outperform girls on most spatial
454 tasks (Cashdan, Marlowe, Crittenden, Porter, & Wood, 2012; Lippa, Collaer, &
455 Peters, 2010; although see Hoffman, Gneezy, & List, 2011). Certainly, there is
456 variation in the magnitude of these differences from nation to nation and from age to
457 age, suggesting a sizeable role for malleable social factors. Still, the direction of the
458 spatial/linguistic differences is essentially invariant. This is not what one would
459 expect based on the hypothesis of a purely environmental origin of these differences.
460 Fourth, efforts to eliminate the gaps seem to quickly reach the point of
461 diminishing returns. Wai et al. (2010) looked at the ratio of boys-to-girls in the top
462 .01% of US seventh graders taking a test of mathematical reasoning between 1981
463 and 2010. In the early 1980s, the ratio was 13 boys for every girl. By the early 1990s,
464 this had dropped to just four boys for every girl, perhaps as a result of increasing
465 access to a good math education for girls (see also Hyde et al., 1990). Since then,
466 however, the ratio of boys-to-girls among the top math performers has remained
467 largely the same, despite intensified efforts to eliminate the remaining gap.2
468 Meanwhile, the ratio of girls-to-boys among the top performers on tests of verbal and
469 reading ability has consistently favored girls (Makel et al., 2016; Wai et al., 2010).
470 Again, these findings are not what one would expect based on the assumption that the
471 gaps have a purely environmental origin.
472 Fifth, gay men tend to have spatial and linguistic abilities comparable to those
473 of straight women, whereas lesbians tend to have spatial abilities comparable to those
474 of straight men (but female-typical linguistic abilities; Xu, Norton, & Rahman, 2017).
475 This finding is difficult to explain on the assumption that social forces alone create the
2 Indeed, some research suggests that the ratio of males to females at the top levels of mathematical
reasoning has increased in recent years (Lakin, 2013).
476 usual pattern of sex differences. Gay men were presumably subject to essentially the
477 same gender-specific social forces as straight men, and lesbians the same gender-
478 specific social forces as straight women. As such, the near-reversal of the usual spatial
479 vs. language pattern is hard to reconcile with the claim that this pattern is due
480 primarily to social forces. Other variables, such as prenatal hormones, plausibly play a
481 larger role.
482 Finally, some of the most popular sociocultural explanations for STEM gender
483 gaps are flawed in a number of ways. This includes the claim that the gaps are a
484 product of a widespread stereotype that “girls can’t do math,” which persuades girls
485 and women that math is not for them, or which causes them to underperform in math
486 as a result of their anxiety about confirming the denigrating stereotype (a putative
487 example of a phenomenon known as stereotype threat; Spencer, Steele, & Quinn,
488 1999). Although such theories are very popular, they face a number of critical
489 challenges. To begin with, it is unclear whether the theories can account for the data,
490 even in principle. How, for example, would a blanket stereotype about female
491 mathematical inferiority explain the fact that girls get better grades than boys in math
492 class, or that they do better at mathematical computation but worse at mathematical
493 reasoning? It is also debatable to what extent the social influences point in the
494 direction the theory presupposes. According to one study, by four years of age, girls
495 tend to assume that boys are academically inferior, and by seven, boys assume the
496 same thing (Hartley & Sutton, 2013). In a similar vein, teachers tend to view their
497 female students as superior at math (and reading), even when aptitude tests indicate
498 that the boys are doing better (Robinson & Lubienski, 2011). Popular culture often
499 mirrors these trends, with girls depicted as academically superior to boys (consider,
500 for instance, Bart and Lisa Simpson). In short, there is no simple story about the social
501 influences to which boys and girls are exposed.
502 Alongside these theoretical concerns, recent research directly challenges the
503 notion that stereotype threat is a major contributor to STEM gender gaps (Stoet &
504 Geary, 2012). A meta-analysis of studies looking at the impact of stereotype threat on
505 girls’ performance in math, science, and spatial tests found a near zero effect, and
506 evidence of substantial publication bias (Flore & Wicherts, 2015). Soon afterwards, a
507 large, pre-registered study found no effect of stereotype threat on women’s math
508 performance (Finnigan & Corker, 2016). Finally, an analysis of thousands of real-life
509 chess games revealed that, in spite of stereotypes about male superiority at the
510 chessboard, women in fact played better when competing against men than when
511 competing against other women: the exact opposite of what stereotype threat would
512 predict (Stafford, 2017). At the very least, the effect of stereotype threat appears to be
513 more modest than originally assumed.
515 It seems reasonable to conclude that the cognitive sex differences considered
516 in this section are shaped to an important degree by unlearned biological factors. As
517 with sex differences in occupational preferences, however, it is not at all obvious why
518 this might be the case. Some evolutionary psychologists argue that sex differences in
519 spatial ability trace to the fact that Homo sapiens is an effectively polygynous species:
520 that is, a species in which males have somewhat greater reproductive variability than
521 females (Betzig, 2012; Labuda, Lefebvre, Nadeau, & Roy-Gagnon, 2010). In
522 polygynous species, the argument goes, males tend to have larger ranges than females,
523 and therefore tend to evolve stronger spatial and wayfaring skills (Gaulin, 1992). The
524 polygyny-related spatial sex difference may have been further amplified in our species
525 by the fact that human males are specialized for hunting and tracking animals, and
526 perhaps for engaging in coalitional warfare with neighboring groups (Silverman &
527 Phillips, 1998).
528 Many evolutionary psychologists are unpersuaded by these ideas, however. In
529 their view, cognitive sex differences were not specifically favored by natural
530 selection, but rather are byproducts of other differences that were. Clint, Sober,
531 Garland Jr, and Rhodes (2012) argue, for instance, that the male advantage in spatial
532 skills is simply an adaptively neutral side effect of hormonal sex differences that were
533 selected for other reasons. If this is right, then the spatial sex difference may have an
534 innate basis but not be a direct product of natural selection. At this stage, the ultimate
535 origins of human cognitive sex differences are uncertain. What does seem certain,
536 however, is that the differences are not solely a product of social forces. To some
537 extent – perhaps to an important extent – they are a part of human nature.
538
539 Sex Differences in Variability
540 We see, then, that small mean differences in certain STEM-relevant aptitudes
541 result in somewhat more males than females occupying the right-hand tail of the
542 distribution (Reilly et al., 2015). However, even if there were no differences at the
543 mean, males could still outnumber females among the minority at the right-hand tail.
544 This is because males and females differ in another way as well. In a wide variety of
545 traits, males as a group are more variable than females: The male distribution is
546 slightly flatter, and stretches out somewhat further on both sides of the mean (see
547 Figure 2). This is the case for a range of physical traits, including birth weight, adult
548 weight, adult height, and running speed (Lehre, Lehre, Laake, & Danbolt, 2009),
549 average heart rate during exercise (Hossack & Bruce, 1982), daily energy expenditure
550 (Westerterp & Speakman, 2008; Yamauchi et al., 2007), basal metabolic rate
551 (Westerterp & Goran, 1997; Yamauchi et al., 2007), and various aspects of brain
552 structure (Ritchie et al., 2018). It also appears to be the case for a range of
553 psychological traits, including creativity (Karwowski et al., 2016), general knowledge
554 (Feingold, 1992), physical aggression (Archer & Mehdikhani, 2003), and at least four
555 of the Big 5 personality traits: extraversion, openness, agreeableness, and
556 conscientiousness (Borkenau, McCrae, & Terracciano, 2013). Of particular relevance
557 to the present topic, males appear to be more variable than females in a number of
558 cognitive traits relevant to STEM (Baye & Monseur, 2016; Feingold, 1992). In this
559 section, we discuss these differences and where they might have come from.
561
562 Figure 2. A fundamental sex difference in humans and many other animals: For a
563 wide range of traits, males are more variable than females. As such, although only a
564 small percentage of people occupy either extreme of the distribution, more males do
565 than females, even when the mean scores for both sexes are identical.
566
567 Variability in STEM-Relevant Cognitive Traits
568 To begin with, many studies have found greater variance among males in
569 specific cognitive capacities, including mathematical aptitude, spatial ability, and
570 science knowledge. In one classic paper, Hedges and Novell (1995) analyzed the
571 cognitive test scores of six large, nationally representative US samples, together
572 covering a 32-year period. They found that, for 35 out of the 37 tests examined in the
573 study, male variability was greater than female. Importantly, this included all the tests
574 of mathematics, spatial ability, mechanical reasoning, and science knowledge. In most
575 cases, sex differences in average scores were small. Nevertheless, because males were
576 more variable, they tended to outnumber females among the minority with especially
577 high scores. (An exception was reading comprehension, for which males outnumbered
578 females at the bottom, but females outnumbered males at the top.)
579 Similar results have been found in other nations and using other tests. For
580 example, in a large sample of UK students (N ≈ 320,000), Strand, Deary, and Smith
581 (2006) found that, although sex differences were small at the mean, males
582 outnumbered females at the top and the bottom of the distribution for both
583 quantitative and nonverbal reasoning; for verbal reasoning, in contrast, males
584 outnumbered females only at the bottom (see Lohman & Lakin, 2009, for a US
585 replication of this pattern). Likewise, analyses of OECD and IEA data indicate that, in
586 most countries for which data are available, males are more variable than females in
587 math, reading, and science (Baye & Monseur, 2016; Machin & Pekkarinen, 2008).
588 As well as greater male variability in specific cognitive aptitudes, males may
589 be more variable in general cognitive ability or IQ (Deary, Irwing, Der, & Bates,
590 2007; Feingold, 1992; Strand et al., 2006). The gold-standard study on this topic is
591 Johnson et al. (2008). Unlike earlier studies, which used potentially unrepresentative
592 samples, Johnson et al. utilized IQ data from two population-wide surveys of 11-year-
593 old school children in Scotland. As expected, IQ variability was greater among boys
594 than girls, such that there were more boys than girls at both extremes of the IQ
595 distribution: more at the top, but also more at the bottom (although see Iliescu, Ilie,
596 Ispas, Dobrean, and Clinciu, 2016, for a recent failure to replicate this pattern in a
597 large, nationally representative Romanian sample).
598 To the extent that greater male variability results in more males than females
599 occupying the upper echelons of ability, whether for specific aptitudes or general
600 cognitive ability, this may help to explain the preponderance of males in certain
601 STEM fields (Levy & Kimura, 2009; Steven Pinker, 2002). As Johnson et al. (2008)
602 observe, sex differences in cognitive variability are not large enough to explain the
603 entirety of the STEM gender gap. They may, however, be one more piece of the
604 puzzle. See Box 1 for further discussion.
605 --------------------Insert Box 1 about here--------------------
606
Box 1: Exploring the Implications of Greater Male Variability
The claim that men are more variable than women in cognitive ability is
controversial. Here are some questions to ask about this claim:
1. Is it sexist? Is it sexist even if it turns out to be true?
2. If it is sexist against women to say that there are somewhat more men than women at the
highest levels of ability, is it sexist against men to say that there are also more men than
women at the lowest levels? If not, how might we explain this asymmetry?
3. Assume for a moment that males really are more variable in cognitive ability. Should we
suppress this information? Could we suppress it, even if we wanted to?
4. Might it be possible instead to emphasize the importance of avoiding exaggerating small
average differences, of keeping sight of the variation among individuals within each sex, and
of treating individuals as individuals, rather than as instantiations of the statistical properties
of the groups to which they belong?
607
608 The Nature and Nurture of Sex Differences in Variability
609 What might explain sex differences in within-sex variability? Given that the
610 magnitude of the differences fluctuates somewhat across cultures and times, it is
611 unlikely that they are a product solely of biological factors (Feingold, 1992; Hyde,
612 Mertz, & Schekman, 2009). However, as with average differences in preferences and
613 aptitudes, several lines of evidence suggest that biological factors play a crucial role.
614 First, greater male variability is found not only in psychological traits, but in
615 traits that are largely impervious to socialization and cultural norms, such as height,
616 birth weight, and BMI (Lehre et al., 2009). The sex differences in psychological
617 variability thus appear to be part of a broader pattern. Considerations of both
618 parsimony and plausibility suggest that this pattern probably has a single, common
619 cause, rather than distinct causes for its physical and psychological components.
620 Second, sex differences in psychological variability emerge in early childhood.
621 The sex difference in IQ variability, for instance, appears before children begin school
622 (Arden & Plomin, 2006). This does not definitively rule out a Nurture Only
623 explanation for the difference. However, it does add some weight to the scales on the
624 biological side of the argument, and it reduces the range of non-biological factors that
625 any Nurture Only explanation can invoke. Whatever the ultimate causes of greater
626 male variability, those causes are in place by the age of three.
627 Third, sex differences in variability are not unique to humans, but are found in
628 many other species. Among red deer, for instance, males are not only larger than
629 females but are also more variable in size (Clutton-Brock, Guinness, & Albon, 1982);
630 among primates, males are more variable in lifespan (Colchero et al., 2016); among
631 guenons, males are more variable in skull size (Cardini & Elton, 2017); and among
632 barn swallows, males are more variable in tail-length (Møller, 1991; Safran &
633 McGraw, 2004). Examples of greater variability in females are notably less common.
634 When we find this pattern in other species, the only realistic explanation is a
635 biological one. When we then find the same pattern in our own species, considerations
636 of parsimony and plausibility again suggest that a biological explanation is
637 appropriate for us too. Indeed, without a strong reason to think otherwise, the default
638 assumption should be that humans fit within the same explanatory framework that
639 applies to the rest of the animal kingdom, and thus that greater male variability in our
640 species has the same root cause as that in our nonhuman kin.
642 If sex differences in variability are due in large part to innate factors, how
643 might those factors have evolved? Once again, the answer is not yet certain, but
644 biologists have put forward a number of plausible suggestions. One popular
645 suggestion traces greater male variability in general to another, more fundamental sex
646 difference: greater male variability in reproductive success. As a result of sex
647 differences in parental investment, males in many species have greater variability than
648 females in the number of offspring they produce (Clutton-Brock & Vincent, 1991;
649 Trivers, 1972). At one extreme, some males have a very high number of offspring:
650 more than any female. At the other, because mating opportunities are finite, many
651 males have no offspring or very few. Most females, in contrast, fall somewhere in
652 between these extremes. In species where male reproductive variability is high,
653 selection favors any trait that increases a male’s chances of being among the few that
654 have many offspring, rather than the many that have few or none. One such trait
655 appears to be risk-proneness. In many species, selection has favored a greater
656 willingness among males to risk life and limb in the pursuit of status, resources, and
657 mating opportunities. Male risk-taking sometimes paid off and sometimes did not.
658 When it did pay off, however, it paid off so handsomely that, on average, risk-taking
659 males had more offspring than males who were more risk-averse. For females, in
660 contrast, risk-taking offered fewer reproductive advantages, because the ceiling
661 number of offspring for females is so much lower. Thus, males in many species
662 evolved a greater propensity to take risks than did females (Daly & Wilson, 2001).
663 According to one explanation of greater male variability, this calculus applies
664 not only to behaviour but to development: Male development is somewhat more “risk-
665 prone” than female development, such that males have a greater chance of developing
666 especially impressive traits but also a greater chance of developing less impressive
667 ones. The former males have a sufficiently high number of offspring that, on average,
668 males with the risky developmental program have more offspring than those with a
669 more conservative or risk-averse one. As such, the risky male developmental program
670 is selected – and with it, greater male variability (see, e.g., Pomiankowski & Møller,
671 1995; Rowe & Houle, 1996).
672 Might this apply to humans? Compared to most mammals, the human sex
673 difference in reproductive variability is rather modest (Stewart-Williams & Thomas,
674 2013). Nonetheless, genetic and anthropological data strongly suggest that there is
675 still a difference (Betzig, 2012; Labuda et al., 2010; summarized in M. L. Wilson,
676 Miller, & Crouse, 2017, Table 1). This plausibly resulted in the evolution of males
677 who were somewhat more risk-prone than their female counterparts, not just
678 behaviorally but developmentally. In other words, just as men evolved a riskier,
679 boom-or-bust behavioral strategy, so too they evolved a riskier, boom-or-bust
680 developmental strategy – one which sometimes pays off but sometimes backfires.
681 (For other evolutionary explanations of greater male variability, see, e.g., Archer &
682 Mehdikhani, 2003; Hill, 2018; Reinhold & Engqvist, 2013.)

683 Certainly, environmental forces may exert some influence on the size of the
684 variability gap (Hyde et al., 2009), perhaps enlarging it and perhaps sometimes
685 making it smaller. The basic pattern itself, however, is plausibly a part of our
686 evolutionary heritage – one that helps to shape the modern occupational landscape.
687
688 Discrimination
689 We have now discussed three reasons why men and women might not be
690 equally represented in STEM, even if there were no discrimination. This does not
691 imply, of course, that there is no discrimination, or that discrimination is not one of
692 the factors shaping STEM gender gaps. Even if fewer women than men are interested
693 in working in math-intensive STEM fields, it could still be the case that those who are
694 interested face a hostile environment in the classroom and the workplace, are subject
695 to disparaging stereotypes and low expectations, and are less likely to be hired,
696 published, cited, or awarded grants than their male colleagues.
697 This is certainly a plausible suggestion; after all, no one denies that there was
698 considerable sexism in science prior to the second wave of the feminist movement,
699 and it might be unduly optimistic to think that this would evaporate entirely in little
700 more than half-a-century. At the same time, though, the discrimination hypothesis
701 represents a rather serious accusation against people working in STEM, and thus it is
702 only fair to look carefully at the evidence for and against the hypothesis. For the
703 reasons given already, gender disparities are not in themselves direct evidence of
704 discrimination; unless the sexes were psychologically identical, equality of
705 opportunity would almost certainly not translate into equality of outcomes (Steven
706 Pinker, 2002; Radcliffe Richards, 1980). Nonetheless, various lines of evidence do
707 bear on the question of how much discrimination remains in the world of STEM.3
708 Discrimination against Women in STEM
709 The most abundant source of objective evidence for sexism against women in
710 STEM comes from experimental studies looking at people’s reactions to hypothetical
711 applicants for STEM jobs. Otherwise identical job applications are given either a
712 female or a male name, and then evaluated by participants naïve to the purpose of the
713 study. In one widely cited paper in this genre, Moss-Racusin, Dovidio, Brescoll,
714 Graham, and Handelsman (2012) had science faculty from six major universities rate
715 applications for a laboratory manager position. They found that the raters – female
716 and male alike – gave higher ratings to supposedly male applicants than they did to
717 supposedly female ones. Specifically, participants rated the males as more competent
718 and hirable, and as deserving a higher salary. In another, earlier study, Steinpreis,
719 Anders, and Ritzke (1999) found that, for middling job applications, academic
720 psychologists – again, female and male alike – expressed greater willingness to hire a
721 male job candidate than an identical female one. For outstanding applications, in
722 contrast, there was no effect of gender.
723 Admittedly, both of these studies had a number of weaknesses, including the
724 fact that the samples included only 127 and 238 participants, respectively. However,
725 the findings are broadly consistent with a large body of research in this area. A meta-
726 analysis of studies looking at simulated employment decisions (N = 22,348) found a
727 weak pro-male bias in male-dominated fields (d = .13), although no gender bias in
3 Note that our discussion applies largely to the Western world, and especially the Anglophone world,
as this is where most of the relevant research and commentary has focused. Note also that we limit the
discussion to measures of bias other than self-report surveys (e.g., Funk & Parker, 2018), as the latter
are vulnerable to complaints of subjectivity and expectation effects.
728 balanced or female-dominated fields (Koch, D'Mello, & Sackett, 2015). The meta-
729 analysis did not focus specifically on STEM-related decisions; however, given that
730 many STEM fields are male-dominated, the results nonetheless increase the
731 plausibility of the STEM-specific findings. Of course, even if those findings are valid,
732 it is not clear whether they generalize to real-world hiring decisions, where decision
733 makers have more experience and are more motivated to make the best choice.
734 Indeed, the Koch et al. meta-analysis showed that pro-male biases were attenuated in
735 just those circumstances, and fake-résumé experiments (in which fictitious job
736 applications are sent out in response to genuine job advertisements, and subsequent
737 call-backs counted) have yielded mixed results with respect to gender bias (Baert,
738 2018). Still, it is possible that bias plays a role in STEM hiring decisions, at least
739 sometimes, and this is something that every academic and STEM worker who cares
740 about gender equality and fairness should take extremely seriously.
741 Moreover, discrimination in hiring is not the only form of discrimination that
742 ought to concern us. Other studies have uncovered other possible examples of
743 discrimination, in STEM and in academia more generally. For example, professors are
744 less likely to respond to informal inquiries about a PhD program when the inquirer is
745 a woman or non-white (Milkman, Akinola, & Chugh, 2015); papers authored by
746 female economists need to be better written to be accepted into top-tier journals
747 (Hengel, 2017); male researchers are somewhat more likely to share their data and
748 published research with other males than with females (Massen, Bauer, Spurny,
749 Bugnyar, & Kret, 2017); and female academics less often give talks at prestigious US
750 universities, even controlling for the rank of the available speakers and even though
751 that women are no more likely to turn down an invitation (Nittrouer et al., 2017).
752 More egregious still, a Japanese medical school was recently accused of lowering
753 women’s scores on an entrance exam in order to reduce the number of women on their
754 program (Cyranoski, 2018). One might point to weaknesses in any particular study.
755 Nonetheless, the overall pattern of findings is disquieting, to say the least.
756 Challenges to the Discrimination Hypothesis
757 At the same time, a number of cautions and qualifications are necessary. First,
758 it is important to emphasize that discrimination is almost certainly not the entire story
759 when it comes to gender gaps in STEM. The gender bias demonstrated in
760 experimental hiring studies, for instance, tends to be relatively subtle, whereas the
761 gender gap in math-intensive STEM fields is relatively large. This implies that the
762 former is unlikely to be a complete explanation of the latter.
763 Furthermore, discrimination alone cannot readily explain why women are less
764 well represented in some fields than in others. Why would discrimination stop women
765 from going into math-intensive STEM fields such as physics and engineering, but not
766 into other prestigious, high-paying fields such as law, medicine, or veterinary science?
767 One might argue that math-intensive fields are particularly inhospitable to women, as
768 a result of the supposedly widespread stereotype that women are intellectually inferior
769 in mathematics. The problem with this idea, however, is that, until recently, many
770 people assumed that women were intellectually inferior in every area, but this did not
771 stop women from approaching parity with men in virtually every other desirable
772 occupation. Why would discrimination only hold women back in math-intensive
773 STEM fields? And why would it hold them back in math-intensive fields everywhere
774 in the world, rather than, say, math-intensive fields in the United States, psychology in
775 South Africa, and law in Scandinavia? Bias and discrimination fail to explain why
776 women are consistently underrepresented in some fields but not in others. In contrast,
777 sex differences in interests and cognitive specializations provide a straightforward
778 explanation of the pattern.
779 Not only does discrimination fail to explain all the data, but the evidence for
780 discrimination in STEM is actually weaker and more mixed than is often assumed.
781 Certainly, as we saw in the previous section, some studies find evidence of anti-
782 female discrimination in the sciences. At the same time, however, other studies fail to
783 find such discrimination, or find discrimination in favor of women (see, e.g., Baert,
784 2018; Bhattacharya, Kanaya, & Stevens, 2017; Blank, 1991; Breda & Hillion, 2016;
785 Lloyd, 1990; Lutter & Schröder, 2014; Miller & Wai, 2015; Riegle-Crumb, King, &
786 Moore, 2016; Veldkamp, Hartgerink, van Assen, & Wicherts, 2017). This raises the
787 possibility that our picture of the level and nature of discrimination in STEM is at
788 least somewhat distorted.
789 The most important voices on this issue are Stephen Ceci and Wendy
790 Williams. In their view, the idea that women are routinely discriminated against in
791 STEM, while true in earlier generations, is no longer true. The culture of science has
792 changed a great deal over the last half century, but people’s beliefs about that culture
793 have not kept pace with the change. Consider hiring practices. Real-world data
794 suggest that, although fewer women apply for math-intensive STEM jobs, those who
795 do are no less likely to be interviewed and no less likely to be offered the job. On the
796 contrary, they are often more likely to be (Ceci & Williams, 2011). This is the
797 opposite of what we would expect if there were pervasive anti-female bias in STEM.
798 If anything, it looks like there may be a pro-female bias, at least in the West.4
4The situation in non-Western nations is less clear. In some cases, overt anti-female discrimination
may still be prevalent (see, e.g., Cyranoski, 2018).
799 Of course, an alternative explanation would be that the female candidates tend
800 to be better than the males, perhaps because those few females who manage to survive
801 and thrive in male-dominated fields tend to be especially gifted. To determine whether
802 there really is a pro-female bias in STEM hiring, Williams and Ceci (2015) conducted
803 a large-scale hiring-decision study: the largest such study to date. The pair sent
804 hypothetical job applications to tenure-track professors in biology, economics,
805 engineering, and psychology, and asked them to assess the applicants’ suitability for a
806 tenure-track position. The final sample included nearly 900 professors from 371 US
807 universities. Averaging across conditions, Williams and Ceci found a 2:1 bias in favor
808 of female applicants. The pro-female bias was found in all four fields and among both
809 male and female faculty. (The only exception was male economics professors, who
810 showed no significant bias in either direction.) Thus, rather than being biased against
811 women, this study suggested that, when it comes to employment decisions, STEM
812 faculty appear to be biased in their favor.5
813 As well as finding little evidence of anti-female bias in hiring, Ceci and
814 Williams find little evidence of bias in promotion, publication, citation, or the
815 awarding of grants (Ceci et al., 2014; Ceci & Williams, 2011; see also Hechtman et
816 al., 2018). Studies that claim to find such bias are often widely discussed and cited
817 (see, e.g., Budden et al., 2008, on gender bias in acceptance rates for papers first-
818 authored by females,6 and van der Lee & Ellemers, 2015; Wennerås & Wold, 1997,
819 on gender bias in grant success). However, according to Ceci and Williams (2011), a
5 A comparable pro-female preference has also been documented outside STEM, in the Australian
Public Services (Hiscox et al., 2017). Such findings are consistent with the idea that concerns about
anti-female discrimination might sometimes overshoot and inadvertently produce anti-male
discrimination instead.
6 Note that a reanalysis of the data by Webb, O’Hara, and Freckleton (2008) found that there was in
fact no evidence of gender bias in the review process.

820 systematic review of all the available evidence suggests that deviations from gender
821 fairness are rare, and that they just as often favor women as men. Again, such results
822 are not what we would expect if anti-female bias were endemic.
823 To be clear, Ceci and Williams do not deny that women still face unique
824 challenges and hurdles in STEM. This includes, for example, the poor mesh between
825 the demands of a STEM career and the demands of motherhood – a point we come
826 back to later. But discrimination is no longer a major barrier. As Ceci and Williams
827 (2011) sum up the situation, “the ongoing focus on sex discrimination in reviewing,
828 interviewing and hiring represents costly, misplaced effort. Society is engaged in the
829 present in solving problems of the past” (p. 3157).
830 A Mixed Picture
831 It seems fair to say that the evidence for pervasive discrimination in STEM is
832 equivocal, with some studies finding pro-male bias (Cyranoski, 2018; Moss-Racusin
833 et al., 2012), some finding the reverse (Williams & Ceci, 2015), and some finding a
834 mixture of both (Breda & Hillion, 2016). What should we conclude? In our view,
835 there are two main interpretations. The first is that the apparently mixed findings are
836 not in fact inconsistent. Rather than there being uniform bias against women, or
837 uniform bias against men, there are pockets of bias against both sexes (and quite
838 possibly no gender bias at all at some institutions and in some cases). The second
839 interpretation is that, at this stage, the findings are simply inconclusive: The jury is
840 still out. But this in itself suggests that sex-based discrimination could not be hugely
841 prevalent in STEM; if it were, it would be easier to detect a clear signal and the
842 research would paint a more consistent picture of the situation. This, in turn, suggests
843 that factors other than discrimination are probably the primary explanation for the
844 persistence of gender gaps in STEM.

845 A Hidden Barrier to the Progress of Women in STEM
846 Before shifting topics, we should briefly consider another potential barrier to
847 the progress of women in STEM – one that often is overlooked: stereotypes of the
848 sexist academy. In the quest to promote women in STEM, academics and activists
849 may sometimes inadvertently overstate the ubiquity of bias and discrimination against
850 women in this sector. An unintended consequence may be to scare away some women
851 who would otherwise be interested in a STEM career (Sesardic & De Clercq, 2014;
852 Williams & Ceci, 2015). If women are given the impression that the STEM workplace
853 is a hotbed of sexism, and an unwelcome place for women, many might quite
854 understandably decide to look for other fields in which to make their mark. Ironically,
855 the consequent dearth of women in STEM might then itself be taken as further
856 evidence that STEM is a hotbed of sexism, creating a self-reinforcing, vicious cycle.
857 Needless to say, if the STEM workplace really were a hotbed of sexism, this
858 would be something we would need to confront, even if doing so put off some
859 budding female scientists. However, given that we have reasonable evidence that, for
860 the most part, science is fair for women, and that discrimination is the exception
861 rather than the rule, conveying such a dark image of the STEM workplace might do
862 more harm than good. See Figure 3 for a summary of the many factors contributing to
863 the gender gaps in STEM.

865
866 Figure 3: Occupational outcomes are a product of many different factors; workplace
867 discrimination is one among many.
868
869 Policy Options
870 Sex differences in STEM representation are not just an academic matter. The
871 question of what should be done about these differences – or indeed whether anything
872 should be done – is one of the most widely discussed political issues related to the
873 modern academy. In this section, we look at some of the most important policy
874 prescriptions that have been mooted or implemented over recent years. The discussion
875 is not intended to be exhaustive, but rather aims to highlight broad trends and
876 commonly overlooked consequences of the policies in question. Table 1 provides an
877 overview.
878 --------------------Insert Table 1 about here--------------------
879
880 Table 1
881 Potential Unintended Consequences of Policy Interventions Aimed at Reducing
882 Gender Gaps in STEM
Policy Potential Unintended Consequences
Outreach If handled poorly, may inadvertently put undue pressure
on girls to pursue a STEM career or may send a
discouraging message to boys.
Incentives for women to Costly.
go into STEM May push women to make choices less in line with their
preferences and long-term interests.
Arguably discriminatory against men.
Blinded evaluation of job May result in fewer women being hired in certain STEM
applications fields.
Diversity training Tacit accusation of bias.
Dubious efficacy, as quite possibly not targeting the
primary causes of current STEM gaps.
May sometimes backfire (i.e., increase bias).
Affirmative action and Abandons the moral principle that we should avoid
quotas discriminating on the basis of sex.
Casts doubt on women’s genuine achievements.
May strengthen the stereotype that women cannot
succeed in science.
Family-Friendly Policies If benefits offered to mothers only, may force some
couples to avoid gender-atypical parenting arrangements,

even if they would prefer those arrangements.
883
884 Outreach
885 An initial, relatively uncontroversial intervention is outreach: educating
886 children and young people about science-related careers, and emphasizing that these
887 are careers that girls as well as boys should consider. Interventions of this kind would
888 not need to deny that there are average differences between the sexes in STEM-
889 relevant traits. On the contrary, average sex differences provide a strong argument in
890 favor of the interventions. After all, even if gender gaps in STEM representation are
891 primarily a result of sex differences in preferences, aptitudes, and variability, the mere
892 existence of those gaps could still help to sway the career choices of individuals
893 whose interests and talents buck the usual trend. In fields where the gender gap is
894 large, some girls and women who would otherwise pursue a science career might be
895 put off by the fact that many more men than women take that path (Dasgupta & Stout,
896 2014). That being the case, it may always be necessary to encourage and support these
897 individuals, and to encourage everyone else to accept atypical career choices and be
898 tolerant of individual differences. (Notice, incidentally, that the same argument would
899 presumably weigh just as strongly toward encouraging boys and men with atypical
900 career preferences to follow their interests as well.)
901 Of course, as with any intervention, there is a danger of creating harm as well
902 as good. One potential harm could come from outreach programs that focus on girls
903 alone: girls-only science workshops, for instance, or advertising campaigns that depict
904 girls but not boys engaged in science-related activities. Such interventions could
905 inadvertently convey the message to boys that they are no longer welcome in science,
906 and that if they choose to pursue a career in that area, they may face an uphill battle
907 due to institutional favoritism toward girls and women. Girls-only programs could
908 also risk losing the support of people who would otherwise be allies, but who worry
909 that the issue has been captured by a strain of gender politics more concerned about
910 eliminating sex differences than about opening the doors for all.
911 Another potential harm is that well-meaning efforts to encourage girls to
912 pursue careers in STEM could sometimes tip over into excessive pressure to take that
913 road. Susan Pinker (2008) interviewed women who had left successful STEM careers
914 to pursue careers in other areas. Many reported that, as girls and young adults, they
915 were so strongly encouraged to go into STEM that they ended up in jobs they did not
916 especially enjoy. In light of this potential pitfall, we suggest that the aim of outreach
917 efforts should not be to get women into science per se, but rather to give everyone
918 accurate information about STEM career options so that they can make an informed
919 choice about what would suit them best.
920 Incentives for Women to Go into STEM
921 A second type of intervention involves offering incentives to women to go into
922 male-dominated STEM fields. Examples include fee waivers, scholarships, and
923 monetary incentives for completing one’s training in a targeted area. As with
924 outreach, this is already a common practice, and it seems probable that the incentives
925 on offer would encourage more women to make gender-atypical choices (Navarra-
926 Madsen, Bales, & Hynds, 2010).
927 But although the incentives probably work, a number of arguments can be
928 leveled against this practice. One is that it discriminates on the basis of sex: It offers
929 advantages to some individuals but not others, purely on the basis of a fixed biological
930 attribute. Even leaving this aside, however, it is worth considering the wisdom of
931 devoting large amounts of resources to encouraging women to do something they

932 would not otherwise do. Although rarely described that way, this is clearly what the
933 practice amounts to – after all, if the targeted women did want to do it anyway, the
934 incentives would not be necessary. By interfering with women’s choices, it is possible
935 that immediate incentives could nudge some women away from options that might
936 suit them better in the longer term and which might ultimately make them happier.7
937 Certainly, people are not always right about what will make them happy. The
938 question, is though, whether those who call for incentives and other measures to help
939 shrink STEM gender gaps are more likely to be right – or whether women’s happiness
940 is the main motivation behind their efforts.
941 Blinded Evaluation
942 A third intervention is blinded evaluation of job applications, journal article
943 submissions, grant applications, and the like – in other words, removing any evidence
944 of the applicant or author’s sex before beginning the evaluation process. Where this
945 can be done, it is a relatively easy way to neutralize the potentially distorting effects
946 of demographic stereotypes. A possible criticism of the practice, at least as applied to
947 hiring decisions, is that it could only be implemented during the earliest stages of the
948 process: Gender can be concealed in a CV, but not in an interview or job talk.
949 Fortunately, however, a great deal of research suggests that stereotypes exert most of
950 their influence on person perception during those earliest stages, when perceivers
951 have little individuating information about the person being perceived (Koch et al.,
952 2015; Rubinstein, Jussim, & Stevens, 2018). As such, blinded evaluation could well
953 eliminate most of the biasing effects of demographic stereotypes. Another point in
7 To be clear, this is not because STEM does not suit women; clearly, it suits some but not others. The
point is that the incentives may encourage some women who might be better suited to other areas to
take the STEM option instead.
954 favor of the practice is that blinded evaluation automatically eliminates all forms of
955 bias, including not only anti-female bias but anti-male bias as well. Moreover, if there
956 is little bias in either direction, the policy would cost little and do little damage –
957 unlike some of the more heavy-handed interventions on offer.
958 Despite its many merits, however, blinded evaluation may prove to be a
959 politically unpopular option. If Ceci and Williams (2014; 2015) are right that women
960 are often favored rather than disfavored in science hiring, blinded evaluation of job
961 applications would presumably result in somewhat fewer women being hired than is
962 presently the case. Given the strong push toward increasing the numbers of women in
963 STEM, such an outcome is likely to rule against the policy. This is not just a
964 theoretical possibility; the Australian Public Services recently suspended a blinded
965 evaluation trial when they discovered that the practice slightly increased men’s
966 chances of getting hired, and slightly decreased women’s (Hiscox et al., 2017). Notice
967 that, in abandoning the trial, the policy makers effectively revealed that their goal is
968 equality of outcome rather than equality of opportunity.
969 Diversity Training
970 Another intervention aimed at tackling discrimination in STEM is diversity
971 training, also known as anti-bias training. Diversity training takes many forms, but the
972 common thread is the aim of increasing awareness and tolerance of diversity in the
973 workplace, and of helping people from different backgrounds to avoid bias and work
974 together harmoniously. The practice has become increasingly popular over the last
975 few decades, and is now a billion-dollar industry (Anand & Winters, 2008).
976 In spite of its popularity, though, a number of criticisms and concerns have
977 been raised regarding diversity training. To begin with, in its application to the STEM
978 gender gap, the enterprise is premised on the assumption that bias is the primary cause
979 – or at least a major cause – of the differential representation of men and women in
980 STEM. As we saw earlier, however, the evidence for endemic anti-female bias is
981 inconclusive, and the main cause of STEM gender gaps appears to be average sex
982 differences in people’s vocational preferences. This raises several issues with respect
983 to diversity training.
984 The first is ethical. To mandate diversity training is tacitly to accuse people of
985 bias – probabilistically if not in each individual case – which is a rather serious
986 accusation to level at people on the basis of conflicting and contested evidence. A
987 second issue is practical. If bias is no longer the main driver of the gender gaps in
988 STEM, then interventions targeting bias are unlikely to have a positive impact. And if
989 that is the case, then anti-bias training represents a considerable waste of resources –
990 resources that could otherwise be used more productively. Consistent with this
991 assessment, research on the efficacy of anti-bias training paints a mixed picture at
992 best. Several studies have concluded that the most popular anti-bias programs have
993 little impact on diversity outcomes (Dobbin, Kalev, & Kelly, 2007; Kalev, Dobbin, &
994 Kelly, 2006). Furthermore, in some cases, the programs may actually backfire,
995 increasing bias rather than reducing it (Duguid & Thomas-Hunt, 2015; Moss-Racusin
996 et al., 2014). Of course, it is possible that bias is still pervasive but that we have not
997 yet found effective interventions to tackle it. It is also possible, however, that the
998 programs are not targeting the real root causes of women’s lower representation in
999 politically progressive nations (Ceci & Williams, 2011).
1000 In discussing diversity training, it is worth paying special attention to the
1001 concept of implicit or unconscious bias. This concept has become central to diversity
1002 training over recent decades, based primarily on research using the Implicit
1003 Association Test or IAT, a putative measure of implicit bias. In the last few years,
1004 however, the IAT has come under increasing fire, and with it interventions aimed at
1005 thwarting implicit bias. The main criticisms are as follows: (1) The IAT produces
1006 inconsistent results for individuals. Its test-retest reliability is low (Gawronski,
1007 Morrison, Phills, & Galdi, 2017). (2) The IAT does a poor job of predicting
1008 discriminatory behavior (Oswald, Mitchell, Blanton, Jaccard, & Tetlock, 2015).
1009 Explicit bias is a better predictor. (3) Although it seems to be possible to change
1010 people’s implicit biases, the effects of such changes on behavior are trivially small or
1011 non-existent, even in the immediate wake of the intervention (Forscher et al., 2017).
1012 For all three reasons, it seems unlikely that diversity interventions targeting implicit
1013 bias represent a wise allocation of resources. Indeed, even one of the creators of the
1014 IAT, Anthony Greenwald, recommends against the use of interventions promising to
1015 eliminate implicit bias (cited in Lopez, 2017).
1016 Affirmative Action and Quotas
1017 Another, more heavy-handed strategy would be to implement a policy of
1018 affirmative action for women in science, or even a strict quota system. As discussed
1019 earlier, there is some reason to believe that this is already happening informally:
1020 Women appear to be preferentially hired in at least certain STEM fields (Ceci et al.,
1021 2014; Ceci & Williams, 2011). Some suggest, however, that this process should be
1022 formalized, and that more stringent rules about numbers of women ought to be
1023 established (see, e.g., Wallon, Bendiscioli, & Garfinkel, 2015).8 Among the most
1024 common arguments for affirmative action and quotas are that these policies would
1025 provide a counterweight to existing biases, compensate for past injustices, break the
1026 cultural “habit” of male-dominance in certain areas, increase the pool of same-sex role
8Indeed, in some countries, this is happening already; the Max Plank institute in Germany, for
example, offers jobs that only women can apply for (Max Plank Society, 2017).
1027 models available to girls and women, and increase the diversity of available
1028 perspectives.
1029 Some of these goals seem self-evidently desirable; others are topics of debate.
1030 However, even if we were to accept the desirability of all the goals, there are reasons
1031 to be wary of affirmative action as a means of achieving them, especially if this
1032 involves quotas that depart markedly from existing sex ratios. To begin with, there is
1033 an ethical question. Pro-female favoritism represents an explicit rejection of the
1034 principle of equality of opportunity in favor of discrimination on the basis of sex:
1035 precisely what feminism originally set out to overcome. As the philosopher Janet
1036 Radcliffe-Richards (1980) observed, the moral foundation of the women’s liberation
1037 movement – and indeed of all liberation movements – is the idea that people should
1038 be treated fairly and equitably, and that unjust barriers should be dismantled. A policy
1039 that advantages one demographic group over another necessarily abandons those
1040 ideals. In doing so, it risks leaving the women’s movement without its moral
1041 foundation, thereby reducing the issue to a mere power struggle between competing
1042 groups, rather than a matter of principle.
1043 Certainly, throughout history, men were often advantaged over women in a
1044 similarly unprincipled way. It is not clear, however, why any individual woman today
1045 should be advantaged over any individual man, simply because other men were
1046 advantaged over other women in the past. Reversing historical injustices does not
1047 erase them; it merely adds to the total number of injustices in the world. The question
1048 we face today, therefore, is this: Is the appropriate response to injustice to eliminate it,
1049 or simply to turn it on its head?
1050 Of course, one might argue that affirmative action for women in STEM would
1051 not in fact be unjust; instead, the policy would help to equalize men and women’s
1052 chances of advancing in STEM, which are presently unequal due to anti-female
1053 discrimination. As mentioned, however, the evidence for pervasive discrimination in
1054 STEM is equivocal, with some studies suggesting that, at least in some ways, women
1055 are favored over men. Given that advantaging one demographic group over another is
1056 not an ethically trivial act, we should insist on much stronger evidence before
1057 enacting such a policy.
1058 Moreover, it is not only men who might be harmed by affirmative action. In a
1059 number of ways, women themselves could be harmed as well. To begin with,
1060 affirmative action policies may cast a shadow of doubt over women’s genuine
1061 accomplishments. If such policies become widespread, then whenever women win
1062 jobs, grants, or awards, people may find themselves wondering – secretly and despite
1063 their best intentions – whether the women in question were judged by a lower
1064 standard, simply because of their sex (a rather sexist practice in itself, one might
1065 argue). This is not just a pitfall for onlookers; successful women themselves might
1066 end up harboring doubts about their own achievements. As the physicist Stephanie
1067 Meyer (2015) once put it: “How am I supposed to build confidence if I never know
1068 whether I earned my success? Women scientists already suffer from imposter
1069 syndrome; practices that make us think we may be given unearned advantage aren’t
1070 helpful.”
1071 As well as casting doubt on the success of individual women, affirmative
1072 action could harm the image of women in STEM more generally. One of the primary
1073 goals of the women-in-STEM movement has been to eliminate the pernicious and
1074 demonstrably false stereotype that women cannot succeed in science.9 Affirmative
1075 action policies are unlikely to contribute to that project. On the contrary, the policies
1076 may bolster the stereotype. Aside from the fact that they seem to imply that women in
1077 science need the extra help, strong preferences could lower the average level of
1078 performance of women working in STEM. This is not because any individual woman
1079 would perform any worse, but rather is a simple statistical consequence of the fact that
1080 the pool of female STEM candidates is smaller than that of the males. As Figure 4
1081 shows, if equal numbers of top performers are drawn from two samples, but one of
1082 those samples is smaller than the other, the mean level of ability of those from the
1083 smaller sample will necessarily be lower than that from the larger, even if the means
1084 and variances of the two samples are identical. In effect, equalizing the number of
1085 individuals taken from each group would mean lowering the minimum standard for
1086 the smaller group. For less stringent quotas, the minimum standard would not need to
1087 be lowered as much. Nonetheless, it would still need to be lowered.

1089
1090
9Note that we appear to be making slow but steady progress toward this goal (Miller, Nolla, Eagly, &
Uttal, 2018).
1091 Figure 4. If equal or similar numbers of top performers are drawn from samples of
1092 different sizes, the average level of ability of those drawn from the smaller sample
1093 will be lower than that of those drawn from the larger. This is the case even if the
1094 means and standard deviations of the two groups are identical.
1095
1096 This could have damaging consequences for women. In the absence of
1097 affirmative action, a person’s sex tells you little about their probable STEM abilities:
1098 Any woman who has been accepted to a given university, or secured a job at a given
1099 institution, is likely to be just as talented as any man at the same university or in the
1100 same institution. However, if strong affirmative action policies are enacted, sex
1101 suddenly does tell you something about women’s probable STEM abilities: It tells
1102 you that they might not necessarily be as good (cf. Haidt & Jussim, 2016). Again, this
1103 is not because women cannot succeed in STEM – some can and some cannot, just like
1104 men. Instead, it is an inevitable consequence of the fact that enacting strong
1105 preferences for members of a smaller group necessarily means lowering the minimum
1106 standard by which members of that group are judged.
1107 Family-Friendly Policies
1108 A final proposal is that STEM career paths could be reconfigured in ways that
1109 would make them more family-friendly. This is a view that Ceci and Williams (2010,
1110 2011) have championed. In their estimation, one of the main remaining barriers to
1111 career success for women in STEM is the incompatibility of jobs in this area with the
1112 demands of motherhood. Not only do women alone get pregnant and nurse their
1113 young, but women are more likely than men to take time out from their careers to care
1114 for their young children. Nowhere is the clash between motherhood and STEM more
1115 apparent than with regard to the academic tenure system in the United States. As Ceci
1116 and Williams (2010) put it:
1117
1118 The tenure structure in academe demands that women having children make their greatest
1119 intellectual contributions contemporaneously with their greatest physical and emotional
1120 achievements, a feat not expected of men. When women opt out of full-time careers to have
1121 and rear children, this is a choice – constrained by biology – that men are not required to
1122 make. (p. 278)
1123
1124 It is worth pointing out that the family-friendliness of STEM jobs varies a
1125 great deal from nation to nation, and that the lack of family-friendly policies would
1126 not explain why women are less well represented in math-intensive fields than in
1127 others. Nonetheless, finding ways to make STEM occupations more compatible with
1128 parenthood could help to level the playing field in math-intensive and non-math-
1129 intensive fields alike. Suitable policies might include increasing the flexibility of the
1130 window in which academics are able to complete the requirements of tenure, and
1131 increasing the provision of subsidized childcare.
1132 Of course, some might take issue with the “assumption” that women are the
1133 primary caregivers for their young. But this is not an assumption in any normative
1134 sense; it is simply an observation about what tends to happen. And given that it is
1135 what tends to happen - and that it is not something we know how to change, even if it
1136 were ethically acceptable to do so – family-friendly policies might help to equalize
1137 men and women’s opportunities by removing a barrier that faces more women than
1138 men. Furthermore, if enacted in a gender neutral way, such that mothers or fathers
1139 could avail themselves of any parental benefits, the policies would not exert any
1140 special pressure on women to take the primary caregiver role. Either sex could take it,
1141 if they so desired (although see Antecol, Bedard, & Stearns, in press, on possible
1142 unintended consequences of such policies).
1143
1144 Levelling the Playing Field vs. Equalizing Sex Ratios
1145 Having considered a range of possible policy interventions, we should now
1146 step back and ask another, more fundamental question: What should the ultimate goal
1147 of these interventions be? Should we strive for a 50:50 sex ratio in every profession
1148 where one sex currently dominates? Or should we strive instead simply to eliminate
1149 bias and equalize people’s opportunities, then let the cards fall where they may?
1150 If men and women were identical in their aptitudes and aspirations, these
1151 would quite possibly amount to the same thing: Levelling the playing field would
1152 automatically result in a 50:50 sex ratio. Given, however, that men and women are not
1153 identical in their aptitudes and aspirations, we have no reason to expect gender parity,
1154 even under conditions of perfect fairness. On the contrary, the natural expectation
1155 would be that men and women would not be at parity, but rather that men would be
1156 more common in some fields, and women in others, as a result of their freely made
1157 choices. To the extent that this is the case, it becomes much more difficult to justify
1158 pursuing a 50:50 sex ratio in every field. Most women do not want a career in science
1159 and nor do most men. Why should the small fraction of women who do want such a
1160 career be the same size as the small fraction of men? To put it another way, as long as
1161 everyone has the opportunity to pursue a STEM career, and as long as the selection
1162 process is fair, why would it be important to get as many women as men into jobs that
1163 fewer women want?

1164 The Pursuit of Happiness
1165 One way to start tackling this question is to observe that a 50:50 sex ratio in all
1166 STEM fields is not actually a good in itself, but is only a good to the extent that it
1167 increases aggregate happiness and social wellbeing. Importantly, though, to the degree
1168 that occupational disparities are a product of men and women acting on their own
1169 preferences and pursuing their own best interests, it is doubtful that forcing a 50:50
1170 sex ratio would actually achieve these ends.
1171 To begin with, men and women could have different life outcomes, but still be
1172 happy with their lives. One longitudinal study found that, among two cohorts of
1173 individuals identified as academically gifted as children, men and women had
1174 somewhat different aspirations and took somewhat different paths, but ended up
1175 similarly happy with their careers, their relationships, and their lives overall (Lubinski
1176 et al., 2014). Thus, even among those best positioned to achieve their life ambitions,
1177 occupational gender parity appears not to be necessary for happiness.
1178 Not only might it not be necessary, but policies that artificially engineer
1179 gender parity – financial incentives and quotas, for instance – could potentially lower
1180 aggregate happiness. To the extent that these policies work, they necessarily mean
1181 that some people will be funneled into occupations that are less in line with their
1182 tastes and talents. To get more women into university physics programs, for instance,
1183 would require persuading at least some women to choose that option when they
1184 otherwise would not have done so. (At the same time, unless enrolment numbers were
1185 increased, it would also mean turning away some men who otherwise would have.)
1186 The women in question presumably would not come from the ranks of housewives or
1187 secretaries; more than likely they would be women who would otherwise have gone
1188 into other, equally prestigious fields, such as law or medicine. Is there any reason to
1189 think that these women would be happier doing physics? Given that people tend to
1190 make the decisions they think will suit them best and be most satisfying for them, it
1191 seems plausible to think that, on average, they might be somewhat less happy. Again,
1192 people are not always right about what will make them happy. What reason, though,
1193 do we have to think that STEM policymakers are more likely to be right? What reason
1194 do we have to think that third-parties – academics, activists, and politicians – have a
1195 better idea than women themselves about what choices women should make?10
1196 Admittedly, this whole line of argument is premised on the assumption that
1197 our aim as a society should be to maximize happiness, and some might reject that
1198 assumption. Anyone who does, though, should be expected to make a strong case for
1199 their position. Why should we put a statistical, collective goal – i.e., more equal sex
1200 ratios in STEM – above the happiness and autonomy of the flesh-and-blood
1201 individuals who constitute those collectives? Why should policymakers’ preference
1202 for gender parity take precedence over individual men and women’s preferences
1203 regarding their own careers and lives?
1204 Sex Differences as a Sign of Social Health
1205 A recurring theme of discussions of occupational gender disparities is the
1206 often-unspoken assumption that sex differences are inherently problematic, or that
1207 they constitute direct evidence of sexism and the curbing of women’s opportunities.
1208 Some research, however, points to the opposite conclusion. A growing body of
1209 evidence suggests that, in nations with greater wealth and higher levels of gender
1210 equality, sex differences are often larger than they are in less wealthy, less equal
1211 nations. This is true for a wide range of variables, including attachment styles, the Big
10 One might also argue that, even if policymakers did have a better idea, they would still have no right
in a free society to interfere with women’s choices (or with men’s).
1212 Five personality traits, choice of academic speciality and occupation, crying,
1213 depression, enjoyment of casual sex, intimate partner violence, mathematical ability,
1214 mental rotation, self-esteem, subjective wellbeing, and values (summarized in
1215 Schmitt, 2015). Importantly, it is also true of objectively measurable traits such as
1216 height, BMI, and blood pressure, which gives us some reason to think that the pattern
1217 is not simply a product of cross-cultural differences in the ways that people answer
1218 questionnaires or take tests.
1219 What, then, is the cause of the pattern? One possibility is that when people are
1220 free to develop relatively unrestrainedly, nascent differences between individuals –
1221 and average differences between the sexes – have more opportunity to emerge and
1222 grow. In the case of psychological traits, the suggestion would be that men and
1223 women in wealthier, more developed nations have greater freedom to pursue what
1224 interests them and to nurture their own individuality. This freedom may, in turn, result
1225 in larger psychological sex differences (Schmitt, Realo, Voracek, & Allik, 2008).
1226 Regardless of the explanation, though, if certain sex differences are larger in
1227 societies with better social indicators, then rather than being products of a sexist or
1228 oppressive society, these differences may be indicators of the opposite: a
1229 comparatively free and fair one (Sommers, 2013). If so, this casts society’s efforts to
1230 eradicate the sex differences in an entirely new light. Rather than furthering gender
1231 equality, such efforts may involve attacking a positive symptom of gender equality.
1232 By mistaking the fruits of our freedom for evidence of oppression, we may institute
1233 policies that, at best, burn up time and resources in a futile effort to cure a
1234 misdiagnosed disease, and at worst actively limit people’s freedom to pursue their
1235 own interests and ambitions on a fair and level playing field.
1236 The Sexist Assumption Underlying the Demand for Parity
1237 Finally, the strong emphasis on increasing the numbers of women in male-
1238 dominated fields is arguably somewhat sexist. As Susan Pinker (2008) points out, it
1239 tacitly assumes that women do not know what they want, or that they want the wrong
1240 things and thus that wiser third-parties need to “fix” their existing preferences.
1241 Furthermore, it tacitly assumes that the areas in which men dominate are superior. The
1242 psychologist Denise Cummins (2015) put the point well when she observed that, “The
1243 hidden assumption underlying the push to eliminate gender gaps in traditionally male-
1244 dominated fields is that such fields are intrinsically more important and more valuable
1245 to society than fields that traditionally attract more women.” Given that traditionally
1246 female-dominated fields include education, healthcare, and social care, this
1247 assumption is not only sexist; it is also clearly false. As Judith Kleinfeld observed:
1248
1249 We should not be sending [gifted] women the message that they are less worthy human
1250 beings, less valuable to our civilization, lazy or low in status, if they choose to be teachers
1251 rather than mathematicians, journalists rather than physicists, lawyers rather than engineers.
1252 (cited in Steven Pinker, 2002, p. 359)
1253
1254 Certainly, many female-dominated fields pay less, on average, than male-
1255 dominated STEM fields.11 There is a great deal of debate about the reasons for this,
1256 and the extent to which it is a product of sexism vs. the fact that, on average, women
1257 view pay as a less important consideration in choosing a career than do men, and see
1258 things such as job security and flexible work hours as more important (see, e.g.,
11Note that this is not always the case. Various non-STEM professions, including law, medicine, and
pharmacy, pay considerably more than STEM, and now attract more women than men (Susan Pinker,
2010).
1259 Lubinski et al., 2014; Redmond & McGuinness, 2017). Such matters are beyond the
1260 scope of this article. We would point out, however, that even if sexism were the entire
1261 explanation for current pay disparities, the most appropriate solution would
1262 presumably be to strive for fair pay in female-dominated fields, not to try to get more
1263 women into fields that pay more but which, on average, they find less appealing. And
1264 to the extent that the explanation is that women place less weight on a high income in
1265 choosing a career, efforts to get women to prioritize income tacitly assume, once
1266 again, that women’s existing priorities are misguided, and that they ought to adopt
1267 more male-typical priorities instead.
1268 To be clear, we entirely agree that we should try to root out sexism wherever it
1269 still lurks, and tear down any lingering barriers to the progress of women in STEM (as
1270 well as any barriers to the progress of men). These are eminently good goals.
1271 However, for the reasons discussed, a 50:50 sex ratio is not a good goal.
1272
1273 Conclusion: Many Factors at Play
1274 In summary, any exhaustive discussion of the relative dearth of women in
1275 certain STEM fields must take into account the burgeoning science of human sex
1276 differences. If we assume that men and women are psychologically indistinguishable,
1277 then any disparities between the sexes in STEM will be seen as evidence of
1278 discrimination, leading to the perception that STEM is highly discriminatory. Such a
1279 perception, however, is almost certainly false. A large body of research points to the
1280 following three conclusions:
1281
1282 (1) that men and women differ, on average, in their occupational preferences, aptitudes, and
1283 levels of within-sex variability;

1284 (2) that these differences are not due solely to sociocultural causes but have a substantial
1285 innate component; and
1286 (3) that these differences, coupled with the demands of bearing and rearing children, are the
1287 main source of the gender disparities we find in STEM today. Discrimination appears to play a
1288 smaller role, and in some cases may favor women, rather than disfavoring them.
1289
1290 These conclusions have important implications for the way we handle STEM
1291 sex differences. Based on the foregoing discussion, we suggest that the approach that
1292 would be most conducive to maximizing individual happiness and autonomy would
1293 be to strive for equality of opportunity, but then to respect men and women’s
1294 decisions regarding their own lives and careers, even if this does not result in gender
1295 parity across all fields. Approaches that focus instead on equality of outcomes –
1296 including quotas and affirmative action – may exact a toll in terms of individual
1297 happiness. To the extent that these policies override people’s preferences, they
1298 effectively place the goal of equalizing the statistical properties of groups above the
1299 happiness and autonomy of the individuals within those groups. Moreover, policies
1300 aimed at achieving equality of outcomes necessarily involve abandoning one of the
1301 principles that made the women’s movement possible in the first place: the principle
1302 that we should treat all people fairly and avoid discriminating on the basis of sex.
1303 Some might derive different conclusions from the emerging understanding of human
1304 sex differences. Either way, though, it seems hard to deny that this understanding
1305 should be factored into the discussion.
1306
1307 References
1308 Alexander, G. M., & Hines, M. (2002). Sex differences in response to children's toys
1309 in nonhuman primates (Cercopithecus aethiops sabaeus). Evolution and
1310 Human Behavior, 23, 467-479.
1311 Anand, R., & Winters, M.-F. (2008). A retrospective view of corporate diversity
1312 training from 1964 to the present. Academy of Management Learning and
1313 Education, 7, 356-372.
1314 Andersson, M. (1994). Sexual selection. Princeton, NJ: Princeton University Press.
1315 Antecol, H., Bedard, K., & Stearns, J. (in press). Equal but inequitable: Who benefits
1316 from gender-neutral tenure clock stopping policies? American Economic
1317 Review.
1318 Archer, J., & Mehdikhani, M. (2003). Variability among males in sexually selected
1319 attributes. Review of General Psychology, 7, 219-236.
1320 Arden, R., & Plomin, R. (2006). Sex differences in variance of intelligence across
1321 childhood. Personality and Individual Differences, 41, 39-48.
1322 Auyeung, B., Baron-Cohen, S., Chapman, E., Knickmeyer, R., Taylor, K., & Hackett,
1323 G. (2006). Foetal testosterone and the child systemizing quotient. European
1324 Journal of Endocrinology, 155, S123-S130.
1325 Baert, S. (2018). Hiring discrimination: An overview of (almost) all correspondence
1326 experiments since 2005. In S. M. Gaddis (Ed.), Audit studies: Behind the
1327 scenes with theory, method, and nuance (pp. 63-77). Cham, Switzerland:
1328 Springer.
1329 Bando, R., López Bóo, F., & Li, X. (2016). Sex differences in language and socio-
1330 emotional skills in early childhood. IDB Working Paper Series IDB-WP-714.
1331 Baron-Cohen, S. (2003). The essential difference: Male and female brains and the
1332 truth about autism. New York: Basic Books.
1333 Baron-Cohen, S., Cassidy, S., Auyeung, B., Allison, C., Achoukhi, M., Robertson, S.,
1334 . . . Lai, M.-C. (2014). Attenuation of typical sex differences in 800 adults with
1335 autism vs. 3,900 controls. PLOS ONE, 9, e102251.
1336 Baron-Cohen, S., Wheelwright, S., Hill, J., Raste, Y., & Plumb, I. (2001). The
1337 "Reading the Mind in the Eyes" Test revised version: A study with normal
1338 adults, and adults with Asperger syndrome or high-functioning autism.
1339 Journal of Child Psychology and Psychiatry, 42, 241-251.
1340 Baye, A., & Monseur, C. (2016). Gender differences in variability and extreme scores
1341 in an international context. Large-scale Assessments in Education, 4.
1342 Beltz, A., Swanson, J., & Berenbaum, S. (2011). Gendered occupational interests:
1343 Prenatal androgen effects on psychological orientation to Things versus
1344 People. Hormones and Behavior, 60, 313-317.
1345 Benbow, C. P. (1988). Sex differences in mathematical reasoning ability in
1346 intellectually talented preadolescents: Their nature, effects, and possible
1347 causes. Behavioral and Brain Sciences, 11, 169-183.
1348 Benbow, C. P., Lubinski, D., Shea, D. L., & Eftekhari-Sanjani, H. (2000). Sex
1349 differences in mathematical reasoning ability at age 13: Their status 20 years
1350 later. Psychological Science, 11, 474-480.
1351 Berenbaum, S. A., & Beltz, A. M. (2011). Sexual differentiation of human behavior:
1352 Effects of prenatal and pubertal organizational hormones. Frontiers in
1353 Neuroendocrinology, 32, 183-200.
1354 Betzig, L. (2012). Means, variances, and ranges in reproductive success: Comparative
1355 evidence. Evolution and Human Behavior, 33, 309-317.
1356 Bhattacharya, D., Kanaya, S., & Stevens, M. (2017). Are university admissions
1357 academically fair? Review of Economics and Statistics, 99, 449-464.
1358 Blank, R. M. (1991). The effects of double-blind versus single-blind reviewing:
1359 Experimental evidence from the American Economic Review. American
1360 Economic Review, 81, 1041-1067.
1361 Borkenau, P., McCrae, R. R., & Terracciano, A. (2013). Do men vary more than
1362 women in personality? A study in 51 cultures. Journal of Research in
1363 Personality, 47, 135-144.
1364 Breda, T., & Hillion, M. (2016). Teaching accreditation exams reveal grading biases
1365 favor women in male-dominated disciplines in France. Science, 353, 474-478.
1366 Budden, A. E., Tregenza, T., Aarssen, L. W., Koricheva, J., Leimu, R., & Lortie, C. J.
1367 (2008). Double-blind review favours increased representation of female
1368 authors. Trends in Ecology and Evolution, 23, 4-6.
1369 Buss, D. M., & Schmitt, D. P. (1993). Sexual strategies theory: An evolutionary
1370 perspective on human mating. Psychological Review, 100, 204-232.
1371 Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk
1372 taking: A meta-analysis. Psychological Bulletin, 125, 367-383.
1373 Cardini, A., & Elton, S. (2017). Is there a “Wainer’s rule”? Testing which sex varies
1374 most as an example analysis using GueSDat, the free Guenon Skull Database.
1375 Hystrix, the Italian Journal of Mammalogy, 28, 147-156.
1376 Cashdan, E., Marlowe, F. W., Crittenden, A., Porter, C., & Wood, B. M. (2012). Sex
1377 differences in spatial cognition among Hadza foragers. Evolution and Human
1378 Behavior, 33, 274-284.
1379 Ceci, S. J., Ginther, D. K., Kahn, S., & Williams, W. M. (2014). Women in academic
1380 science: A changing landscape. Psychological Science in the Public Interest,
1381 15, 75-141.
1382 Ceci, S. J., & Williams, W. M. (2010). Sex differences in math-intensive fields.
1383 Current Directions in Psychological Science, 19, 275-279.
1384 Ceci, S. J., & Williams, W. M. (2011). Understanding current causes of women’s
1385 underrepresentation in science. Proceedings of the National Academy of
1386 Sciences, 108, 3157-3162.
1387 Ceci, S. J., Williams, W. M., & Barnett, S. M. (2009). Women's underrepresentation
1388 in science: Sociocultural and biological considerations. Psychological Bulletin,
1389 135, 218-261.
1390 Chachra, D. (2017). To reduce gender biases, acknowledge them. Nature, 548, 373.
1391 Chapman, E., Baron-Cohen, S., Auyeung, B., Knickmeyer, R., Taylor, K., & Hackett,
1392 G. (2006). Fetal testosterone and empathy: Evidence from the Empathy
1393 Quotient (EQ) and the "Reading the Mind in the Eyes" Test. Social
1394 Neuroscience, 1, 135-148.
1395 Clint, E. K., Sober, E., Garland Jr, T., & Rhodes, J. S. (2012). Male superiority in
1396 spatial navigation: Adaptation or side effect? Quarterly Review of Biology, 87,
1397 289-313.
1398 Clutton-Brock, T. H., Guinness, F. E., & Albon, S. D. (Eds.). (1982). Red deer:
1399 Behaviour and ecology of two sexes. Cambridge, UK: Cambridge University
1400 Press.
1401 Clutton-Brock, T. H., & Vincent, A. C. J. (1991). Sexual selection and the potential
1402 reproductive rates of males and females. Nature, 351, 58-60.
1403 Colchero, F., Rau, R., Jones, O. R., Barthold, J. A., Conde, D. A., Lenart, A., . . .
1404 Vaupel, J. W. (2016). The emergence of longevous populations. Proceedings
1405 of the National Academy of Sciences.
1406 Connellan, J., Baron-Cohen, S., Wheelwright, S., Batki, A., & Ahluwalia, J. (2000).
1407 Sex differences in human neonatal social perception. Infant Behavior and
1408 Development, 23, 113-118.
1409 Cummins, D. (2015). Why the STEM gender gap is overblown. PBS.org. Retrieved
1410 22 March, 2018, from https://www.pbs.org/newshour/nation/truth-women-
1411 stem-careers
1412 Cyranoski, D. (2018). Japanese university investigates claim that it downgraded
1413 women’s test scores. Nature.
1414 Daly, M., & Wilson, M. (2001). Risk-taking, intrasexual competition, and homicide.
1415 Nebraska Symposium on Motivation, 47, 1-36.
1416 Damore, J. (2017). Google's ideological echo chamber: How bias clouds our thinking
1417 about diversity and inclusion. Internet Archive. Retrieved 1 March, 2018, from
1418 https://web.archive.org/web/20170809220001/https://diversitymemo-static.s3-
1419 us-west-2.amazonaws.com/Googles-Ideological-Echo-Chamber.pdf
1420 Dasgupta, N., & Stout, J. G. (2014). Girls and women in science, technology,
1421 engineering, and mathematics: STEMing the tide and broadening participation
1422 in STEM careers. Policy Insights from the Behavioral and Brain Sciences, 1,
1423 21-29.
1424 Deaner, R. O. (2013). Distance running as an ideal domain for showing a sex
1425 difference in competitiveness. Archives of Sexual Behavior, 42, 413-428.
1426 Deary, I. J., Irwing, P., Der, G., & Bates, T. C. (2007). Brother-sister differences in
1427 the g factor in intelligence: Analysis of full, opposite-sex siblings from the
1428 NLSY1979. Intelligence, 35, 451-456.
1429 Dobbin, F., Kalev, A., & Kelly, E. (2007). Diversity management in corporate
1430 America. Contexts, 6, 21-27.
1431 Duguid, M. M., & Thomas-Hunt, M. C. (2015). Condoning stereotyping? How
1432 awareness of stereotyping prevalence impacts expression of stereotypes.
1433 Journal of Applied Psychology, 100, 343-359.
1434 Eagly, A. H. (1995). The science and politics of comparing men and women.
1435 American Psychologist, 50, 145-158.
1436 Else-Quest, N. M., Hyde, J. S., & Linn, M. C. (2010). Cross-national patterns of
1437 gender differences in mathematics: A meta-analysis. Psychological Bulletin,
1438 136, 103-127.
1439 Feingold, A. (1992). Sex differences in variability in intellectual abilities: A new look
1440 at an old controversy. Review of Educational Research, 61, 61-84.
1441 Ferriman, K., Lubinski, D., & Benbow, C. P. (2009). Work preferences, life values,
1442 and personal views of top math/science graduate students and the profoundly
1443 gifted: Developmental changes and gender differences during emerging
1444 adulthood and parenthood. Journal of Personality and Social Psychology, 97,
1445 517-532.
1446 Finnigan, K. M., & Corker, K. S. (2016). Do performance avoidance goals moderate
1447 the effect of different types of stereotype threat on women’s math
1448 performance? Journal of Research in Personality, 63, 36-43.
1449 Flore, P. C., & Wicherts, J. M. (2015). Does stereotype threat influence performance
1450 of girls in stereotyped domains? A meta-analysis. Journal of School
1451 Psychology, 53, 25-44.
1452 Flores-Mendoza, C., Widaman, K. F., Rindermann, H., Primi, R., Mansur-Alves, M.,
1453 & Pena, C. C. (2013). Cognitive sex differences in reasoning tasks: Evidence
1454 from Brazilian samples of educational settings. Intelligence, 41, 70-84.
1455 Forscher, P. S., Lai, C. K., Axt, J. R., Ebersole, C. R., Herman, M., Devine, P. G., &
1456 Nosek, B. A. (2017). A meta-analysis of change in implicit bias. PsyArXiv.
1457 Funk, C., & Parker, K. (2018). Women and men in STEM often at odds over
1458 workplace equity. Washington, DC: Pew Research Center.
1459 Gaulin, S. J. C. (1992). Evolution of sex-differences in spatial ability. Yearbook of
1460 Physical Anthropology, 35, 125-151.
1461 Gawronski, B., Morrison, M., Phills, C. E., & Galdi, S. (2017). Temporal stability of
1462 implicit and explicit measures: A longitudinal analysis. Personality and Social
1463 Psychology Bulletin, 43, 300-312.
1464 Geary, D. C. (2010). Male, female: The evolution of human sex differences (2nd ed.).
1465 Washington, DC: American Psychological Association.
1466 Gino, F., Wilmuth, C. A., & Brooks, A. W. (2015). Compared to men, women view
1467 professional advancement as equally attainable, but less desirable.
1468 Proceedings of the National Academy of Sciences, 112, 12354-12359.
1469 Guiso, L., Monte, F., Sapienza, P., & Zingales, L. (2008). Culture, gender, and math.
1470 Science, 320, 1164-1165.
1471 Haidt, J., & Jussim, L. (2016). Hard truths about race on campus. Wall Street Journal.
1472 Retrieved 22 March, 2018, from https://www.wsj.com/articles/hard-truths-
1473 about-race-on-campus-1462544543
1474 Hakim, C. (2005). Sex differences in work-life balance goals. In D. M. Houston (Ed.),
1475 Work-Life Balance in the 21st Century (pp. 55-79). London, UK: Palgrave
1476 Macmillan.
1477 Halpern, D. F. (2012). Sex differences in cognitive abilities (4th ed.). New York, NY:
1478 Psychology Press.
1479 Halpern, D. F., Benbow, C. P., Geary, D. C., Gur, R. C., Hyde, J. S., & Gernsbacher,
1480 M. A. (2007). The science of sex differences in science and mathematics.
1481 Psychological Science in the Public Interest, 8, 1-51.
1482 Hartley, B. L., & Sutton, R. M. (2013). A stereotype threat account of boys' academic
1483 underachievement. Child Development, 84, 1716-1733.
1484 Hassett, J. M., Siebert, E. R., & Wallen, K. (2008). Sex differences in rhesus monkey
1485 toy preferences parallel those of children. Hormones and Behavior, 54, 359-
1486 364.
1487 Hechtman, L. A., Moore, N. P., Schulkey, C. E., Miklos, A. C., Calcagno, A. M.,
1488 Aragon, R., & Greenberg, J. H. (2018). NIH funding longevity by gender.
1489 Proceedings of the National Academy of Sciences.
1490 Hedges, L. V., & Friedman, L. (1993). Gender differences in variability in intellectual
1491 abilities: A reanalysis of Feingold’s results. Review of Educational Research,
1492 63, 94-105.
1493 Hedges, L. V., & Nowell, A. (1995). Sex differences in mental test scores, variability
1494 and numbers of high-scoring individuals. Science, 269, 41–45.
1495 Hengel, E. (2017). Publishing while female. Are women held to higher standards?
1496 Evidence from peer review. Cambridge Working Paper Economics: 1753.
1497 Hill, T. P. (2018). An evolutionary theory for the variability hypothesis.
1498 arXiv:1703.04184v6 [q-bio.PE].
1499 Hiscox, M. J., Oliver, T., Ridgway, M., Arcos-Holzinger, L., Warren, A., & Willis, A.
1500 (2017). Going blind to see more clearly: Unconscious bias in Australian Public
1501 Service (APS) shortlisting processes. Behavioural Economics Team of the
1502 Australian Government. Retrieved 20 March, 2018, from
1503 https://www.pmc.gov.au/domestic-policy/behavioural-economics/going-blind-
1504 see-more-clearly-unconscious-bias-australian-public-service-aps-shortlisting-
1505 processes
1506 Hoffman, M., Gneezy, U., & List, J. A. (2011). Nurture affects gender differences in
1507 spatial abilities. Proceedings of the National Academy of Sciences, 108,
1508 14786-14788.
1509 Hossack, K. F., & Bruce, R. A. (1982). Maximal cardiac function in sedentary normal
1510 men and women: Comparison of age-related changes. Journal of Applied
1511 Physiology: Respiratory, Environmental and Exercise Physiology, 53, 799-
1512 804.
1513 Hrdy, S. B. (2009). Mothers and others: The evolutionary origin of mutual
1514 understanding. Cambridge, MA: Harvard University Press.
1515 Hyde, J. S. (2005). The gender similarities hypothesis. American Psychologist, 60,
1516 581-592.
1517 Hyde, J. S., Fennema, E., & Lamon, S. J. (1990). Gender differences in mathematics
1518 performance: A meta-analysis. Psychological Bulletin, 107
1519 Hyde, J. S., Lindberg, S. M., Linn, M. C., Ellis, A. B., & Williams, C. C. (2008).
1520 Gender similarities characterize math performance. Science, 321, 494-495.
1521 Hyde, J. S., Mertz, J. E., & Schekman, R. (2009). Gender, culture, and mathematics
1522 performance. Proceedings of the National Academy of Sciences of the United
1523 States of America, 106, 8801-8807.
1524 Iliescu, D., Ilie, A., Ispas, D., Dobrean, A., & Clinciu, A. I. (2016). Sex differences in
1525 intelligence: A multi-measure approach using nationally representative
1526 samples from Romania. Intelligence, 58, 54-61.
1527 Janicke, T., Häderer, I. K., Lajeunesse, M. J., & Anthes, N. (2016). Darwinian sex
1528 roles confirmed across the animal kingdom. Science Advances, 2, e1500983.
1529 Johnson, W., Carothers, A., & Deary, I. J. (2008). Sex differences in variability in
1530 general intelligence: A new look at the old question. Perspectives on
1531 Psychological Science, 3, 518-531.
1532 Kalev, A., Dobbin, F., & Kelly, E. (2006). Best practices or best guesses? Assessing
1533 the efficacy of corporate affirmative action and diversity policies. American
1534 Sociological Review, 71, 589-617.
1535 Karwowski, M., Jankowska, D. M., Gralewski, J., Gajda, A., Wiśniewska, E., &
1536 Lebuda, I. (2016). Greater male variability in creativity: A latent variables
1537 approach. Thinking Skills and Creativity, 22, 159-166.
1538 Kenrick, D. T., Groth, G., Trost, M. R., & Sadalla, E. K. (1993). Integrating
1539 evolutionary and social exchange perspectives on relationships: Effects of
1540 gender, self-appraisal, and involvement level on mate selection criteria.
1541 Journal of Personality and Social Psychology, 64, 951-969.
1542 Kersey, A. J., Braham, E. J., Csumitta, K. D., Libertus, M. E., & Cantlon, J. F. (2018).
1543 No intrinsic gender differences in children’s earliest numerical abilities. NPJ
1544 Science of Learning, 3.
1545 Koch, A. J., D'Mello, S. D., & Sackett, P. R. (2015). A meta-analysis of gender
1546 stereotypes and bias in experimental simulations of employment decision
1547 making. Journal of Applied Psychology, 100, 128-161.
1548 Labuda, D., Lefebvre, J.-F., Nadeau, P., & Roy-Gagnon, M.-H. (2010). Female-to-
1549 male breeding ratio in modern humans: An analysis based on historical
1550 recombinations. American Journal of Human Genetics, 86, 353-363.
1551 Lakin, J. M. (2013). Sex differences in reasoning abilities: Surprising evidence that
1552 male–female ratios in the tails of the quantitative reasoning distribution have
1553 increased. Intelligence, 41, 263-274.
1554 Lehre, A.-C., Lehre, K. P., Laake, P., & Danbolt, N. C. (2009). Greater intrasex
1555 phenotype variability in males than in females is a fundamental aspect of the
1556 gender differences in humans. Developmental Psychobiology, 51, 198-206.
1557 Lemos, G. C., Abad, F. J., Almeida, L. S., & Colom, R. (2013). Sex differences on g
1558 and non-g intellectual performance reveal potential sources of STEM
1559 discrepancies. Intelligence, 41, 11-18.
1560 Levy, J., & Kimura, D. (2009). Women, men, and the sciences. In C. H. Sommers
1561 (Ed.), The science on women and science (pp. 202-284). Washington, DC:
1562 AEI Press.
1563 Lippa, R. A. (2009). Sex differences in personality traits and gender-related
1564 occupational preferences across 53 nations: Testing evolutionary and social-
1565 environmental theories. Archives of Sexual Behavior, 39, 619-636.
1566 Lippa, R. A. (2010). Sex differences in personality traits and gender-related
1567 occupational preferences across 53 nations: Testing evolutionary and social-
1568 environmental theories. Archives of Sexual Behavior, 39, 619-636.
1569 Lippa, R. A., Collaer, M. L., & Peters, M. (2010). Sex differences in mental rotation
1570 and line angle judgments are positively associated with gender equality and
1571 economic development across 53 nations. Archives of Sexual Behavior, 39,
1572 990-997.
1573 Lippa, R. A., Preston, K., & Penner, J. (2014). Women's representation in 60
1574 occupations from 1972 to 2010: More women in high-status jobs, few women
1575 in things-oriented jobs. PLOS ONE, 9, e95960.
1576 Lloyd, M. E. (1990). Gender factors in reviewer recommendations for manuscript
1577 publication. Journal of Applied Behavior Analysis, 23, 539-543.
1578 Lohman, D. F., & Lakin, J. M. (2009). Consistencies in sex differences on the
1579 Cognitive Abilities Test across countries, grades, test forms, and cohorts.
1580 British Journal of Educational Psychology, 79, 389-407.
1581 Lopez, G. (2017). For years, this popular test measured anyone’s racial bias. But it
1582 might not work after all. Vox. Retrieved 22 March, 2018, from
1583 https://www.vox.com/identities/2017/3/7/14637626/implicit-association-test-
1584 racism
1585 Lubinski, D., Benbow, C. P., & Kell, H. J. (2014). Life paths and accomplishments of
1586 mathematically precocious males and females four decades later.
1587 Psychological Science, 25, 2217-2232.
1588 Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2001). Foetal testosterone and
1589 vocabulary size in 18- and 24-month-old infants. Infant Behavior and
1590 Development, 24, 418-424.
1591 Lutchmaya, S., Baron-Cohen, S., & Raggatt, P. (2002). Foetal testosterone and eye
1592 contact in 12-month-old human infants. Infant Behavior and Development, 25,
1593 327-335.
1594 Lutter, M., & Schröder, M. (2014). Who becomes a tenured professor, and why?
1595 Panel data evidence from German sociology, 1980–2013. MPIfG Discussion
1596 Paper 14/19
1597 Machin, S., & Pekkarinen, T. (2008). Global sex differences in test score variability.
1598 Science, 322, 1331-1332.
1599 Maeda, Y., & Yoon, S. Y. (2013). A meta-analysis on gender differences in mental
1600 rotation ability measured by the Purdue Spatial Visualization Tests:
1601 Visualization of Rotations (PSVT:R). Educational Psychology Review, 25, 69-
1602 94.
1603 Makel, M. C., Wai, J., Peairs, K., & Putallaz, M. (2016). Sex differences in the right
1604 tail of cognitive abilities: An update and cross cultural extension. Intelligence,
1605 59, 8-15.
1606 Manning, J. T., Reimers, S., Baron-Cohen, S., Wheelwright, S., & Fink, B. (2010).
1607 Sexually dimorphic traits (digit ratio, body height, systemizing–empathizing
1608 scores) and gender segregation between occupations: Evidence from the BBC
1609 internet study. Personality and Individual Differences, 49, 511-515.
1610 Massen, J. J. M., Bauer, L., Spurny, B., Bugnyar, T., & Kret, M. E. (2017). Sharing of
1611 science is most likely among male scientists. Scientific Reports, 7, 12927.
1612 Max Planck Society. (2017). More opportunities for female scientists. Max-Planck-
1613 Gesellschaft. Retrieved 16 May, 2018, from
1614 https://www.mpg.de/11767653/lise-meitner-programme
1615 Meyer, S. (2015). Women in science: What explains gaps? Part I. Psychology Today.
1616 Retrieved March 20, 2018, from
1617 https://www.psychologytoday.com/us/blog/rabble-rouser/201510/women-in-
1618 science-what-explains-gaps-part-i
1619 Milkman, K. L., Akinola, M., & Chugh, D. (2015). What happens before? A field
1620 experiment exploring how pay and representation differentially shape bias on
1621 the pathway into organizations. Journal of Applied Psychology, 100, 1678-
1622 1712.
1623 Miller, D. I., Nolla, K. M., Eagly, A. H., & Uttal, D. H. (2018). The development of
1624 children's gender‐science stereotypes: A meta‐analysis of 5 decades of U.S.
1625 draw‐a‐scientist studies. Child Development.
1626 Miller, D. I., & Wai, J. (2015). The bachelor’s to Ph.D. STEM pipeline no longer
1627 leaks more women than men: A 30-year analysis. Frontiers in Psychology, 6.
1628 Møller, A. P. (1991). Sexual selection in the monogamous barn swallow (Hirundo
1629 rustica). I. Determinants of tail ornament size. Evolution, 45, 1823-1836.
1630 Moore, D. S., & Johnson, S. P. (2008). Mental rotation in human infants: A sex
1631 difference. Psychological science, 19, 1063-1066.
1632 Morris, M. L. (2016). Vocational interests in the United States: Sex, age, ethnicity,
1633 and year effects. Journal of Counseling Psychology, 63, 604-615.
1634 Moss-Racusin, C. A., Dovidio, J. F., Brescoll, V. L., Graham, M. J., & Handelsman,
1635 J. (2012). Science faculty’s subtle gender biases favor male students.
1636 Proceedings of the National Academy of Sciences, 109, 16474-16479.
1637 Moss-Racusin, C. A., van der Toorn, J., Dovidio, J. F., Brescoll, V. L., Graham, M. J.,
1638 & Handelsman, J. (2014). Scientific diversity interventions. Science, 343, 615-
1639 616.
1640 Muller, M. N., Wrangham, R. W., & Pilbeam, D. R. (Eds.). (2017). Chimpanzees and
1641 human evolution. Cambridge, MA: Belknap Press.
1642 Navarra-Madsen, J., Bales, R. A., & Hynds, D. L. (2010). Role of scholarships in
1643 improving success rates of undergraduate Science, Technology, Engineering
1644 and Mathematics (STEM) majors. Procedia - Social and Behavioral Sciences,
1645 8, 458-464.
1646 Nittrouer, C. L., Hebl, M. R., Ashburn-Nardo, L., Trump-Steele, R. C. E., Lane, D.
1647 M., & Valian, V. (2017). Gender disparities in colloquium speakers at top
1648 universities. Proceedings of the National Academy of Sciences.
1649 Oswald, F. L., Mitchell, G., Blanton, H., Jaccard, J., & Tetlock, P. E. (2015). Using
1650 the IAT to predict ethnic and racial discrimination: small effect sizes of
1651 unknown societal significance. Journal of Personality and Social Psychology,
1652 108, 562-571.
1653 Pinker, Steven. (2002). The blank slate: The modern denial of human nature. New
1654 York, NY: Viking.
1655 Pinker, Steven. (2010). On women, STEM and hidden bias. Minding the Campus.
1656 Retrieved 17 May, 2018, from https://www.susanpinker.com/wp-
1657 content/uploads/2014/08/On-Women-STEM-and-Hidden-Bias.pdf
1658 Pinker, Susan. (2008). The sexual paradox: Troubled boys, gifted girls, and the real
1659 difference between the sexes. New York, NY: Scribner.
1660 Pomiankowski, A., & Møller, A. P. (1995). A resolution of the lek paradox.
1661 Proceedings of the Royal Society of London. Series B: Biological Sciences,
1662 260, 21-29.
1663 Radcliffe Richards, J. (1980). The sceptical feminist: A philosophical enquiry. Oxon,
1664 UK: Routledge.
1665 Redmond, P., & McGuinness, S. (2017). The gender wage gap in Europe: Job
1666 preferences, gender convergence and distributional effects. IZA DP No. 10933
1667 Reilly, D., Neumann, D. L., & Andrews, G. (2015). Sex differences in mathematics
1668 and science achievement: A meta-analysis of National Assessment of
1669 Educational Progress assessments. Journal of Educational Psychology, 107,
1670 645-662.
1671 Reinhold, K., & Engqvist, L. (2013). The variability is in the sex chromosomes.
1672 Evolution, 67, 3662-3668.
1673 Reynolds, M. R., Scheiber, C., Hajovsky, D. B., Schwartz, B., & Kaufman, A. S.
1674 (2015). Gender differences in academic achievement: Is writing an exception
1675 to the gender similarities hypothesis? Journal of Genetic Psychology, 176,
1676 211-234.
1677 Riegle-Crumb, C., King, B., & Moore, C. (2016). Do they stay or do they go? The
1678 switching decisions of individuals who enter gender atypical college majors.
1679 Sex Roles, 74, 436-449.
1680 Ritchie, S. J., Cox, S. R., Shen, X., Lombardo, M. V., Reus, L. M., Alloza, C., . . .
1681 Deary, I. J. (2018). Sex differences in the adult human brain: Evidence from
1682 5216 UK Biobank participants. Cerebral Cortex, bhy109-bhy109.
1683 Robinson, J. P., & Lubienski, S. T. (2011). The development of gender achievement
1684 gaps in mathematics and reading during elementary and middle school:
1685 Examining direct cognitive assessments and teacher ratings. American
1686 Educational Research Journal, 48, 268-302.
1687 Rowe, L., & Houle, D. (1996). The lek paradox and the capture of genetic variance by
1688 condition dependent traits. Proceedings of the Royal Society of London, Series
1689 B: Biological Sciences, 263, 1415-1421.
1690 Rubinstein, R. S., Jussim, L., & Stevens, S. T. (2018). Reliance on individuating
1691 information and stereotypes in implicit and explicit person perception. Journal
1692 of Experimental Social Psychology, 75, 54-70.
1693 Ruzich, E., Allison, C., Chakrabarti, B., Smith, P., Musto, H., Ring, H., & Baron-
1694 Cohen, S. (2015). Sex and STEM occupation predict Autism-Spectrum
1695 Quotient (AQ) scores in half a million people. PLOS ONE, 10, e0141229.
1696 Safran, R. J., & McGraw, K. J. (2004). Plumage coloration, not length or symmetry of
1697 tail-streamers, is a sexually selected trait in North American barn swallows.
1698 Behavioral Ecology, 15, 455-461.
1699 Schmitt, D. P. (2005). Sociosexuality from Argentina to Zimbabwe: A 48-nation
1700 study of sex, culture, and strategies of human mating. Behavioral and Brain
1701 Sciences, 28, 247-275.
1702 Schmitt, D. P. (2015). The evolution of culturally-variable sex differences: Men and
1703 women are not always different, but when they are… it appears not to result
1704 from patriarchy or sex role socialization. In V. A. Weekes-Shackelford & T.
1705 K. Shackelford (Eds.), The evolution of sexuality (pp. 221-256). New York,
1706 NY: Springer.
1707 Schmitt, D. P., & Project, M. o. t. I. S. D. (2003). Universal sex differences in the
1708 desire for sexual variety: Tests from 52 nations, 6 continents, and 13 islands.
1710 Schmitt, D. P., Realo, A., Voracek, M., & Allik, J. (2008). Why can’t a man be more
1711 like a woman? Sex differences in Big Five personality traits across 55 cultures.
1713 Schwartz, S. H., & Rubel, T. (2005). Sex differences in value priorities: Cross-cultural
1714 and multimethod studies. Journal of Personality and Social Psychology, 89,
1715 1010-1028.
1716 Sesardic, N., & De Clercq, R. (2014). Women in philosophy: Problems with the
1717 discrimination hypothesis. Academic Questions, 27, 461-473.
1718 Silverman, I., & Phillips, K. (1998). The evolutionary psychology of spatial sex
1719 differences: Evolutionary theory and data. In C. B. Crawford & D. Krebs, L.
1720 (Eds.), Handbook of evolutionary psychology: Ideas, issues, and applications
1721 (pp. 595-612). Mahwah, NJ: Erlbaum.
1722 Simpson, E. A., Nicolini, Y., Shetler, M., Suomi, S. J., Ferrari, P. F., & Paukner, A.
1723 (2016). Experience-independent sex differences in newborn macaques:
1724 Females are more social than males. Scientific Reports, 6, 19669.
1725 Slabbekoorn, D., van Goozen, S. H. M., Megens, J., Gooren, L. J. G., & Cohen-
1726 Kettenis, P. T. (1999). Activating effects of cross-sex hormones on cognitive
1727 functioning: A study of short-term and long-term hormone effects in
1728 transsexuals. Psychoneuroendocrinology, 24, 423-447.
1729 Sommers, C. H. (2013). What 'lean in' misunderstands about gender differences. The
1730 Atlantic. Retrieved 12 January, 2017, from
1731 https://www.theatlantic.com/sexes/archive/2013/03/what-lean-in-
1732 misunderstands-about-gender-differences/274138/
1733 Spelke, E. S. (2005). Sex differences in intrinsic aptitude for mathematics and
1734 science? A critical review. American Psychologist, 60, 950-958.
1735 Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women's
1736 math performance. Journal of Experimental Social Psychology, 35, 4-28.
1737 Stafford, T. (2017). Female chess players outperform expectations when playing men.
1738 psyarxiv.com/bpy3t.
1739 Steinpreis, R. E., Anders, K. A., & Ritzke, D. (1999). The impact of gender on the
1740 review of the Curricula Vitae of job applicants and tenure candidates: A
1741 national empirical study. Sex Roles, 41, 509-528.
1742 Stewart-Williams, S., & Thomas, A. G. (2013). The ape that thought it was a peacock:
1743 Does evolutionary psychology exaggerate human sex differences?
1744 Psychological Inquiry, 24, 137-168.
1745 Stoet, G., Bailey, D. H., Moore, A. M., & Geary, D. C. (2016). Countries with higher
1746 levels of gender equality show larger national sex differences in mathematics
1747 anxiety and relatively lower parental mathematics valuation for girls. PLoS
1748 ONE, 11, e0153857.
1749 Stoet, G., & Geary, D. C. (2012). Can stereotype threat explain the gender gap in
1750 mathematics performance and achievement? Review of General Psychology,
1751 16, 93-102.
1752 Stoet, G., & Geary, D. C. (2013). Sex differences in mathematics and reading
1753 achievement are inversely related: Within- and across-nation assessment of 10
1754 years of PISA data. PLoS ONE, 8, e57988.
1755 Stoet, G., & Geary, D. C. (2018). The gender-equality paradox in science, technology,
1756 engineering, and mathematics education. Psychological Science.
1757 Strand, S., Deary, I. J., & Smith, P. (2006). Sex differences in Cognitive Abilities Test
1758 scores: A UK national picture. British Journal of Educational Psychology, 76,
1759 463-480.
1760 Su, R., Rounds, J., & Armstrong, P. I. (2009). Men and things, women and people: A
1761 meta-analysis of sex differences in interests. Psychological Bulletin, 135, 859-
1762 884.
1763 Trivers, R. L. (1972). Parental investment and sexual selection. In B. Campbell (Ed.),
1764 Sexual selection and the descent of man: 1871-1971 (pp. 136-179). Chicago,
1765 IL: Aldine Press.
1766 van der Lee, R., & Ellemers, N. (2015). Gender contributes to personal research
1767 funding success in The Netherlands. Proceedings of the National Academy of
1768 Sciences, 112, 12349-12353.
1769 Veldkamp, C. L. S., Hartgerink, C. H. J., van Assen, M. A. L. M., & Wicherts, J. M.
1770 (2017). Who believes in the storybook image of the scientist? Accountability
1771 in Research, 24, 127-151.
1772 Voyer, D., Voyer, S., & Bryden, M. P. (1995). Magnitude of sex differences in spatial
1773 abilities: A meta-analysis and consideration of critical variables. Psychological
1774 Bulletin, 117, 250-270.
1775 Wai, J., Cacchio, M., Putallaz, M., & Makel, M. C. (2010). Sex differences in the
1776 right tail of cognitive abilities: A 30-year examination. Intelligence, 38, 412-
1777 423.
1778 Wai, J., Lubinski, D., & Benbow, C. P. (2009). Spatial ability for STEM domains:
1779 Aligning over 50 years of cumulative psychological knowledge solidifies its
1780 importance. Journal of Educational Psychology, 101, 817-835.
1781 Wallon, G., Bendiscioli, S., & Garfinkel, M. S. (2015). Exploring quotas in academia.
1782 Robert Bosch Stiftung project number: 12.5.8050.0239.0. Retrieved 22 March,
1783 2018, from
1784 http://www.embo.org/documents/science_policy/exploring_quotas.pdf
1785 Wang, M.-T., Eccles, J. S., & Kenny, S. (2013). Not lack of ability but more choice:
1786 Individual and gender differences in choice of careers in science, technology,
1787 engineering, and mathematics. Psychological Science, 24, 770-775.
1788 Webb, T. J., O’Hara, B., & Freckleton, R. P. (2008). Does double-blind review
1789 benefit female authors? Trends in Ecology and Evolution, 23, 351-353.
1790 Wennerås, C., & Wold, A. (1997). Nepotism and sexism in peer-review. Nature, 387,
1791 341-343.
1792 Westerterp, K. R., & Goran, M. I. (1997). Relationship between physical activity
1793 related energy expenditure and body composition: A gender difference.
1794 International Journal of Obesity, 21, 184-188.
1795 Westerterp, K. R., & Speakman, J. R. (2008). Physical activity energy expenditure has
1796 not declined since the 1980s and matches energy expenditures of wild
1797 mammals. International Journal of Obesity, 32, 1256-1263.
1798 Williams, W. M., & Ceci, S. J. (2015). National hiring experiments reveal 2:1 faculty
1799 preference for women on STEM tenure track. Proceedings of the National
1800 Academy of Sciences, 112, 5360-5365.
1801 Wilson, M., & Daly, M. (1985). Competitiveness, risk taking, and violence: The
1802 young male syndrome. Ethology and Sociobiology, 6, 59-73.
1803 Wilson, M. L., Miller, C. M., & Crouse, K. N. (2017). Humans as a model species for
1804 sexual selection research. Proceedings of the Royal Society B: Biological
1805 Sciences, 284
1806 Xu, Y., Norton, S., & Rahman, Q. (2017). Sexual orientation and neurocognitive
1807 ability: A meta-analysis in men and women. Neuroscience and Biobehavioral
1808 Reviews.
1809 Yamauchi, T., Kim, S.-N., Lu, Z., Ichimaru, N., Maekawa, R., Natsuhara, K., . . .
1810 Ishii, M. (2007). Age and gender differences in the physical activity patterns
1811 of urban schoolchildren in Korea and China. Journal of Physiological
1812 Anthropology, 26, 101-107.
1813 Yang, Y., & Barth, J. M. (2015). Gender differences in STEM undergraduates'
1814 vocational interests: People-thing orientation and goal affordances. Journal of
1815 Vocational Behavior, 91, 65-75.

Men Women and Science - Stewart-Williams and Halsey 2018

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Men Women and Science - Stewart-Williams and Halsey 2018

Caricato da

Copyright:

Formati disponibili

Men, Women, and Science 1

1 Running head: MEN, WOMEN, AND SCIENCE

5 Men, Women, and Science

6 Why the Differences and What Should Be Done?

9 School of Psychology, University of Nottingham Malaysia Campus, Semenyih

12 Department of Life Sciences, University of Roehampton, London

17 Villegas-Montes for their perceptive comments on an early draft of this paper.

18 Correspondence concerning this article should be addressed to Steve Stewart-

19 Williams at the School of Psychology, University of Nottingham Malaysia Campus.

25 Word count: 20,500

32 However, a great deal of evidence in the behavioral sciences suggests that

39 improve women’s situation across the STEM fields.

41 Public Significance Statements

43 causes, including average sex differences in preferences and cognitive specialties.

44 Anti-female discrimination, though it demonstrably exists, is probably not the

45 primary cause of STEM gender gaps in modern Western nations.

46 Policy prescriptions founded on the notion of pervasive anti-female bias are

47 unlikely to have positive effects.

49 KEYWORDS: Discrimination; Equality; Gender; Sex Differences; STEM.

51 Men, Women, and Science

52 Why the Differences and What Should Be Done?

54 Never has the issue of gender disparities been as widely discussed, or as

61 to which observed gender disparities in STEM are a product of discrimination. The

67 occupational outcomes were due entirely to discrimination (Chachra, 2017). No

68 doubt, discrimination is part of the explanation. However, a great deal of research

88 Sex Differences in Preferences and Priorities

89 To begin with, we examine arguably the most important non-bias explanation

91 average – in certain career-relevant preferences. Specifically, we look at sex

92 differences in occupational preferences and sex differences in life priorities. After

97 A large literature in psychology shows that men and women differ, on

99 important recent papers on this topic was a comprehensive meta-analysis by Su,

104 gender gaps.

125 people’s occupational outcomes.

126 Research looking at preferences for specific occupations leads to a similar

138 Life Priorities

152 self-reported priorities were evident in real-world behaviour. For example, as

155 worked an average of 40.

165 US study (N ≈ 4,000), the sex difference in career-mindedness is not a result of

168 desirable (Gino, Wilmuth, & Brooks, 2015).

169 The Nature and Nurture of Sex Differences in Occupational Preferences

177 differences themselves are products of discriminatory socialization practices and

182 First, sex differences in occupational preferences have remained remarkably

191 women into those very professions.

217 Wallen, 2008).

218 Finally, occupational preferences have been convincingly linked to prenatal

232 plays a role as well.

233 Evolutionary Rationale

248 impregnate the woman.

258 1985). In light of this theoretical framework – which is well-supported by research in

261 than men prioritize family and work-life balance.

262 Evolutionary explanations for sex differences in occupational preferences are

264 technologists, no engineers or mathematicians. As such, any innate contribution to sex

278 contribution to the relevant sex differences is strong.

280 Sex Differences in Aptitudes

287 easily be misconstrued, so it is important to be absolutely clear what is meant.

294 Hajovsky, Schwartz, & Kaufman, 2015).

304 or below the average.

316 2007; Pinker, 2002).

317 --------------------Insert Figure 1 about here--------------------

344 women than men are suited to work in others.

345 The Relevant Differences

356 Lubinski, & Benbow, 2009).

365 gender composition of fields such engineering and physics.