The Johnson study presented the results for all 42 tests, but calculated effect sizes only for those that met a stricter than normal standard of statistical significance (p < .01 instead of p < .05) because of the large number of tests involved. Results for the residual effects on 21 of the subtests that met that statistical standard are shown in the following table. I omit the p values. All but two of the p values for the residual effects were at the .001 level.80 The effect sizes stripped of g are ordered from the largest for females (positive) to the largest for males (negative).
COGNITIVE SEX DIFFERENCES IN THE MISTRA SAMPLE
Assessment activity: Coding (ID of symbol-number pairings)
Overall effect size: +0.56
Effect size stripped of g: +0.83
Assessment activity: Perceptual speed (evaluation of symbol pairs)
Overall effect size: +0.37
Effect size stripped of g: +0.68
Assessment activity: Spelling (multiple choice)
Overall effect size: ns
Effect size stripped of g: +0.66
Assessment activity: Word fluency (production of anagrams)
Overall effect size: ns
Effect size stripped of g: +0.64
Assessment activity: ID of familial relationships within a family tree
Overall effect size: ns
Effect size stripped of g: +0.63
Assessment activity: Rote memorization of meaningful pairings
Overall effect size: +0.33
Effect size stripped of g: +0.60
Assessment activity: Production of words beginning and ending with specified letters
Overall effect size: ns
Effect size stripped of g: +0.57
Assessment activity: Vocabulary (multiple choice)
Overall effect size: ns
Effect size stripped of g: +0.50
Assessment activity: Rote memorization of meaningless pairings
Overall effect size: ns
Effect size stripped of g: +0.42
Assessment activity: Chronological sequencing of pictures
Overall effect size: –0.28
Effect size stripped of g: –0.30
Assessment activity: Information (recall of factual knowledge)
Overall effect size: –0.29
Effect size stripped of g: –0.39
Assessment activity: Trace of a path through a grid of dots
Overall effect size: –0.42
Effect size stripped of g: –0.40
Assessment activity: Matching of rotated alternatives to probe
Overall effect size: ns
Effect size stripped of g: –0.45
Assessment activity: Reproduction of 2-D designs of 3-D blocks
Overall effect size: –0.34
Effect size stripped of g: –0.48
Assessment activity: Outline of cutting instructions to form the target figure
Overall effect size: –0.39
Effect size stripped of g: –0.48
Assessment activity: Arithmetic (mental calculation of problems presented verbally)
Overall effect size: –0.36
Effect size stripped of g: –0.53
Assessment activity: ID of unfolded version of a folded probe
Overall effect size: –0.44
Effect size stripped of g: –0.59
Assessment activity: ID of matched figures after rotation
Overall effect size: –0.55
Effect size stripped of g: –0.75
Assessment activity: ID of parts missing in pictures of common objects
Overall effect size: –0.60
Effect size stripped of g: –0.81
Assessment activity: ID of rotated versions of 2-D representation of 3-D objects
Overall effect size: –0.92
Effect size stripped of g: –1.04
Assessment activity: ID of mechanical principles and tools
Overall effect size: –1.18
Effect size stripped of g: –1.43
Source: Adapted from Johnson and Bouchard (2007): Table 4. “ns” signifies p > .01. Negative effect sizes indicate a higher male mean.
First, look at the column showing the overall effect size, calculated the same way as all the other effect sizes you have seen. Among the effect sizes that were statistically significant, four were “small” by the Cohen guidelines, eight were “medium,” one was “large,” and one was “very large.” We can safely assume that most of those that did not meet the p < .01 standard of statistical significance fell in the “small” range.
Now look at the right-hand column, showing the difference between males and females on these subtests when the role of g has been extracted. As Johnson and Bouchard anticipated, all of these effect sizes are larger than the overall effect size. Furthermore, only one qualifies as “small” while 13 are “medium,” five are “large,” and two are “very large.” (As you might predict, I think that if these conceptually related effect sizes were aggregated, the value of Mahalonobis D would be huge.) Johnson and Bouchard’s work tells us how much that apparent similarity in overall g is illusory: End points are similar, but ways of getting to them are different. Hence the title of their article: “Sex Differences in Mental Abilities: g Masks the Dimensions on Which They Lie.”
Linking Sex Differences in Neurocognitive Functioning with the People-Things Dimension
People generally enjoy the things they’re good at. They also like the experience of being good at what they do—a fundamental truth about the nature of human enjoyment that goes back to Aristotle. The sex differences in neurocognitive functioning point to a tendency for men and women to enjoy different kinds of activities. When I discussed visuospatial skills, I listed some of the vocations that, to attain excellence, require high visuospatial skills—math, programming, engineering, architecture, chemistry, the building trades. They’re all Things occupations. Excellence in verbal skills almost by definition requires one to be able to engage with other people. This is self-evidently true in occupations that require steady interaction with other people—teaching, patient-oriented medicine, and helping professions of all kinds. They’re all People occupations.
These days, everyone who has been paying attention knows that the Things and People occupations I just listed are notorious for being disproportionately male and female respectively. You can guess what’s coming next.
4
Sex Differences in Educational and Vocational Choices
Proposition #3: On average, women worldwide are more attracted to vocations centered on people and men to vocations centered on things.
The third component of cognitive repertoires is social behavior, but there’s no point in cataloging all the ways in which men and women differ in social behavior. They go from the obvious and extreme (e.g., men commit the overwhelming majority of violent crimes) to the obvious and everyday (e.g., women perform the overwhelming majority of child-rearing tasks).1 I devote this chapter to an extended look at the People-Things thesis regarding education and vocation. More than a century after legal restrictions on women’s vocations were lifted and half a century since gender discrimination in hiring, promoting, and firing was outlawed, large disparities continue in the university educations that young men and women attain, the jobs they take, and how their careers unfold. What to do about this is a major policy debate. Here, I lay out some reasons for thinking that the persistence of these observable sex differences constitutes strong circumstantial evidence for underlying biological causes.
The Women of SMPY
From January 2012 to February 2013, a team of Vanderbilt psychologists surveyed 322 men and 157 women in their late 40s about their work preferences and life values. The men and women differed on many of their views. Limiting the list to ones with an absolute effect size of 0.35 or higher, here were the things that men valued more or agreed with more than women did.2 The effect sizes are shown in parentheses:
“The prospect of receiving criti
cism from others does not inhibit me from expressing my thoughts.” (–0.54)
A merit-based pay system (–0.53)
Having a full-time career (–0.51)
Inventing or creating something that will have an impact (–0.45)
A salary that is well above the average person’s (–0.43)
“I believe that society should invest in my ideas because they are more important than those of other people in my discipline.” (–0.42)
Being able to take risks on my job (–0.41)
Working with things (e.g., computers, tools, machines) as part of my job (–0.41)
“The possibility of discomforting others does not deter me from stating the facts.” (–0.40)
Having lots of money (–0.36)
Stereotypical men.
Meanwhile, here were the things that the women in the sample valued more than the men did, again limiting the list to ones where the absolute effect size was 0.35 or higher:3
Having a part-time career for a limited time period (+0.83)
Having a part-time career entirely (+0.78)
Working no more than 40 hours in a week (+0.72)4
Having strong friendships (+0.49)
Flexibility in my work schedule (+0.41)
Community service (+0.38)
Having time to socialize (+0.37)
Giving back to the community (+0.35)
Stereotypical women.
Why have I presented such predictable results? Because this chapter is about sex differences in educational and vocational choices, and this particular sample lets me put issues of sex differences in abilities aside. Every one of those middle-aged men and women had an IQ of about 140 or higher.5 They were part of the Study of Mathematically Precocious Youth—SMPY.
The Unique Advantages of the SMPY Samples
The results I just presented came from members of SMPY’s Cohort 2, born in 1964–67, who at age 13 had tested in the top 0.5 percent of overall intellectual ability: the top 1 in 200. All of them were also in the top percentile specifically in math skills.6 All of the respondents were intellectually qualified to have pursued any undergraduate major they preferred and any cognitively demanding career.7
The SMPY sample has other advantages. Every girl in the sample knew she was extremely talented in math by the time she entered her teens. Her mathematical talent was part of her self-image from an early age.
SMPY
Johns Hopkins psychologist Julian Stanley began SMPY in 1971. He recruited large numbers of 12-year-olds to take the SAT math test. The SAT is designed for high school juniors and seniors bound for college. By administering it to 12-and 13-year-olds who had not yet taken high school math courses, Stanley was able to identify students with exceptionally high aptitude for math. Over the years, SMPY established four cohorts of mathematically precocious youth who became part of a longitudinal study that continues as I write, jointly directed since 1991 by Camilla Benbow and David Lubinski.[8] I focus on the results from the 35-year and 40-year follow-ups for Cohort 2 with some supplemental findings regarding Cohort 3.
Virtually all of the parents of these girls were extremely supportive of their daughters’ talent. SMPY parents responded positively to their child’s invitation to seek admission to the program and then were willing to go through the time and effort to get their child to the testing site, which often meant a significant journey. Apart from these indicators, we also have the results of a study of the SMPY parents in that era. The study found that “(a) parents were treating their children differently based not on their child’s gender but apparently rather as a function of their child’s talent; (b) fathers did not appear to be more involved with the mathematically talented students than with the verbally talented; and (c) the majority of students, especially females, were not strongly sex typed.”9
Despite their parents’ support, it might be argued that the girls who entered SMPY’s Cohort 2 were still socialized to traditional pre-feminist norms. The modern feminist movement was in its first decade when they were born in the mid-1960s. It should be assumed that as little girls the SMPY women had gotten a full dose of socialization to female roles in a country that was still traditional on matters regarding gender roles.
If the girls who entered SMPY had typically come from small towns or from middle-class suburban neighborhoods in the Midwest or South, that argument would have merit. But instead they came from highly educated, upper-middle-class families located in the Washington, Baltimore, and Philadelphia areas. By the early 1970s, these families and neighborhoods were probably more explicitly and emphatically feminist than comparable neighborhoods today. It may be hard for readers not old enough to remember for themselves how long—50 years now—the upper-middle-class milieu has been overwhelmingly feminist, so perhaps a few reminders are in order.
First, consider the timeline of legal reforms:
1963: John Kennedy’s Presidential Commission on the Status of Women released its strongly pro-feminist report, and the Equal Pay Act of 1963 mandated equal pay for equal work.
1964: Title VII of the Civil Rights Act of 1964 forbade employer discrimination on the basis of sex. Griswold v. Connecticut invalidated legal restrictions on access to birth control.
1967: A presidential executive order extended affirmative action in employment and education to include women.
1968: Sexual harassment was added to federal antidiscrimination law as a basis for bringing actions against employers.
1972: Title IX of the Education Amendments mandated nondiscrimination in any school receiving government aid (effectively all of them) and included broad enforcement powers.
The girls of Cohort 2 were born into a world in which legal equality had been established, but in some respects that was the least of it. Even by the time they were in elementary school, the list of legal victories had been accompanied by a cultural sea change that began in the 1960s and was at its height during the 1970s and early 1980s. In the upper-middle-class schools and neighborhoods where most of the SMPY girls grew up, courses in elementary school were filled with inspirational stories about women scientists, political leaders, artists, and authors. High schools were putting boys and girls in the same gym classes, and high school counselors were urging female students to go into male-dominated careers. On campuses, young women were hearing faculty and their fellow students urging them to forgo marriage and childbearing in favor of a career. Gloria Steinem’s famous slogan “A woman needs a man like a fish needs a bicycle” comes from the early 1970s, epitomizing a celebration of women that found cultural expression in films, popular music, and television.10 When the girls of Cohort 2 reached college age in 1982–85, they all knew that the most famous universities in the nation were eager to add them to their student bodies and even more eager for them to populate their majors in science, technology, engineering, and math, familiarly known as STEM.
That’s the female sample we are able to follow through almost 50 years of their lives: extraordinarily talented women who knew they were talented from an early age, were urged to enter STEM fields, and were often urged not to let children and family dictate their lives. Yet they reached their late 40s with a profile of stereotypical sex differences in career and life priorities. What had they been doing in the meantime?
Even Among the Gifted, People Tend to Like Doing What They Do Best
Educationally, males and females in Cohort 2 were in a dead heat, with nearly the same high proportions getting bachelor’s, master’s, and doctoral degrees.11 Yet the traditional gender gap in STEM majors persisted. The SMPY women were about twice as likely to take STEM majors as the general population of female undergraduates,12 but this was true of the men also, and so the male-female ratio in STEM degrees among the SMPY sample (1.6) was fractionally higher than the ratio in the general undergraduate population (1.5).13 Meanwhile, twice as many of these gifted young women were getting degrees in the social sciences, business, and the humanities as were the gifted young men.14
&n
bsp; Why the persistence of the tilt of men toward STEM and women toward the social sciences and the humanities? The special nature of the SMPY sample enables a test of this proposition: It doesn’t make any difference how extremely talented you are in one field; you tend to gravitate toward the field in which you are the most talented.15 Even though everyone in Cohort 2 was gifted in math skills, those whose verbal skills were even higher than their math skills tended to end up in the social sciences, humanities, business, and law, while those whose math skills were greater than their verbal skills tended to end up in STEM fields. This pattern was true for both males and females.16
Much of the theoretical literature assumes that this tendency is driven by students’ and their parents’ knowledge of grades and test scores—that the numbers tell them they are better at math than verbal or vice versa.17 But there is good reason to think that the impulse runs deeper than conscious knowledge. This finding emerged first from Project Talent, a 1960 study based on a nationally representative sample (total sample size was about 400,000) of American high school students whose participants were given tests of visuospatial skill to accompany their scores on verbal and mathematics tests. Visuospatial skill is not ordinarily something that schools test or that students think about. And yet it turned out that such skills played an important independent role in shaping the students’ academic and vocational choices, strongly reinforcing a tendency for students to go into STEM majors if those skills were high and depressing the likelihood of going into STEM majors if they were low.18
Human Diversity Page 9
