R—Realistic. Working with tools, instruments, and mechanical or electrical equipment. Activities include building, repairing machinery, and raising crops/animals.
I—Investigative. Investigating and attempting to understand phenomena in the natural sciences through reading, research, and discussion.
A—Artistic. Expressing oneself through activities such as painting, designing, singing, dancing, and writing; artistic appreciation of such activities (e.g., listening to music, reading literature).
S—Social. Helping, enlightening, or serving others through activities such as teaching, counseling, working in service-oriented organizations, and engaging in social/political studies.
E—Enterprising. Persuading, influencing, directing, or motivating others through activities such as sales, supervision, and aspects of business management.
C—Conventional. Developing and/or maintaining accurate and orderly files, records, accounts, etc.; following systematic procedures for performing business activities.40
The theory is often referred to as RIASEC, based on its combined initials. I should add that two of the labels were poorly chosen. As the descriptions make clear, “Enterprising” doesn’t refer to entrepreneurship or risk-taking. It’s about interacting with other people in a leadership or managerial role. “Conventional” doesn’t mean timid or boring or stuck with tradition. It refers to a preference for procedure, systematic practices, and orderliness. These more accurate understandings also make it clear why Enterprising is related to the People dimension and Conventional is related to what Baron-Cohen calls systemizing—and, by extension, the Things dimension.
Subsequently, psychometrician Dale Prediger developed a two-dimensional way of assessing the results of inventories testing Holland’s orientations. One dimension went from Ideas at one extreme to Data at the other. The second dimension went from Things at one extreme to People at the other. Prediger accompanied his analysis with formulas for computing a People-Things index and an Ideas-Data index.41
Schematically, Holland’s and Prediger’s ideas look like this.
Source: ACT (2009): Fig. 1.1.42
Holland’s and Prediger’s conceptualizations have remained at the center of vocational counseling because they work. Tests of where people stand on the six clusters have proved to be valid descriptors of people’s occupational interests and useful in giving people career guidance.43
Those tests have also revealed sex differences. In 2009, psychologists at the University of Illinois and Iowa State University conducted a meta-analysis. The first author was psychologist Rong Su. Their question: Taking the literature as a whole, where do men and women come out on the Holland orientations and the Prediger dimensions?
The authors assembled a database from 81 samples that amounted to 243,670 men and 259,518 women. On average, women’s vocational interests tilted toward occupations involving work with or understanding of other people; men’s vocational interests tilted toward working with things.
The biggest tilts involved the Realistic orientation—a male preference—with an effect size of –0.84, and the Social orientation—a female preference—with an effect size of +0.68. When the data were analyzed along Prediger’s two dimensions using his indexes, a striking contrast emerged. On the Data-Ideas dimension, there was virtually no sex difference.[44] On the People-Things dimension, the effect size was +0.93, meaning that women were on the People end and men were on the Things end of the dimension—a large effect size by any standard.
Sex Differences in Vocational Interests Are Replicated in the Jobs That Men and Women Occupy
What about the jobs that people actually hold? Women are overwhelmingly pointed toward People jobs and away from Things jobs, something I will document with regard to all jobs before zeroing in on women’s representation in STEM occupations.
All occupations. In 1938, the U.S. federal government started publishing the Dictionary of Occupational Titles. Since 1998, it has existed on the Internet as O*NET, a digital database that provides information about the skills, personal characteristics, cognitive requirements, experience requirements, and job outlook for each occupation in the list. As part of that information, O*NET now uses explicit criteria to score each occupation on the six RIASEC orientations on a continuous scale running from 1.0 through 7.0.45 We can use these data for jobs actually held to replicate the earlier findings from the Su data for test scores of vocational interests. In the following figure the black bars show the results on occupations by employed persons ages 25–54 in the combined Current Population Surveys for 2014–18. The gray bars show the results from the Su meta-analysis of data on vocational interests.
Source: Author’s analyses, CPS 2014–18 and Su, Rounds, and Armstrong (2009): Table 3.
It’s not often that two completely different databases produce such similar results. The big sex differences on both interests and jobs were for the Realistic and Social orientations—the conceptually clearest Things and People orientations—and on Prediger’s overall index combining all the RIASEC data.
STEM occupations. The same thing happens within different kinds of STEM occupations: In 2015, two of the same authors, Rong Su and James Rounds, conducted another meta-analysis focusing on distinctions within scientific and technical occupations. Again, their database of vocational preferences and a database of actual jobs held by the U.S. working-age population produced almost interchangeable results. In both cases, the biggest sex differences favoring men involved the most Things-oriented jobs; the biggest sex differences favoring women involved the most People-oriented jobs.[46] The degree of consistency of the sex differences in vocational interests and occupations is quite remarkable. To draw the discussion together, consider the table below.
SEX DIFFERENCES (D) IN VOCATIONAL INTERESTS AND OCCUPATIONS ACROSS DIFFERENT MEASURES AND SAMPLES
RIASEC dimension: Realistic
Meta-analysis of 503,188 scores on interest inventories: –0.84
Adult scores of SMPY cohorts 1, 2, 3, and 4: –0.92
Ratings of occupations held by Americans ages 25–54: –0.77
RIASEC dimension: Investigative
Meta-analysis of 503,188 scores on interest inventories: –0.26
Adult scores of SMPY cohorts 1, 2, 3, and 4: –0.28
Ratings of occupations held by Americans ages 25–54: –0.08
RIASEC dimension: Conventional
Meta-analysis of 503,188 scores on interest inventories: +0.33
Adult scores of SMPY cohorts 1, 2, 3, and 4: –0.47
Ratings of occupations held by Americans ages 25–54: +0.27
RIASEC dimension: Enterprising
Meta-analysis of 503,188 scores on interest inventories: –0.04
Adult scores of SMPY cohorts 1, 2, 3, and 4: –0.50
Ratings of occupations held by Americans ages 25–54: +0.09
RIASEC dimension: Artistic
Meta-analysis of 503,188 scores on interest inventories: +0.35
Adult scores of SMPY cohorts 1, 2, 3, and 4: +1.06
Ratings of occupations held by Americans ages 25–54: +0.22
RIASEC dimension: Social
Meta-analysis of 503,188 scores on interest inventories: +0.68
Adult scores of SMPY cohorts 1, 2, 3, and 4: +0.88
Ratings of occupations held by Americans ages 25–54: +0.84
Source: Su, Rounds, and Armstrong (2009); Author’s analysis, combined ACS, 2011–15; Lubinski and Benbow (2006): Table 5. A negative score indicates a higher male mean.
The three columns draw on databases that are quite different in both measures and samples, yet they tell a clear and consistent story.[47] Effect sizes on Realistic favoring men were –0.84, –0.92, and –0.77 for the three sources; effect sizes on Social favoring women were +0.68, +0.88, and +0.84 for the three sources.
Trends in Vocational Interests and Choices Since 1970
Everything I’ve given you so far is evidence for the existence of a phenomenon—sex differen
ces on the People-Things dimension—but no evidence for its cause. All these effect sizes could simply mean that women and men alike continue to be socialized into certain gender-typical interests as well as gender-typical jobs.
Let’s turn to evidence that I believe makes this interpretation difficult to defend. A look back at what has happened to educational and job choices over the last 50 years suggests that vocational doors really did open up for women during the 1970s, that women took advantage of those new opportunities to the extent that they wanted to, and that we fairly quickly reached a new equilibrium.
Women’s Undergraduate Majors Since 1970: A Brief Surge in Things-Oriented STEM and Enduring Increases in People-Oriented STEM
Women’s undergraduate majors offer a case in point. In 1971, 38 percent of women’s bachelor’s degrees were in education. That proportion had fallen by half by the early 1980s. Meanwhile, degrees in business grew from 3 percent in 1971 to 20 percent by 1982. Women were no longer limited to K–12 teaching as their major professional option (besides nursing), and they quickly made other professional interests known.
There were also big changes in women’s bachelor of science degrees—but of a particular kind. Consider first the most Things-oriented STEM careers—physics, chemistry, earth sciences, computer science, mathematics, and engineering. The percentage of women’s degrees obtained in those majors more than doubled from 1971 to 1986—but “more than doubled” meant going from 4 percent to 10 percent.48 And 1986 was the high point. By 1992, that number had dropped to 6 percent, where it has remained, give or take a percentage point, ever since. It’s not that women were unwilling to undertake majors that require courses in math and science. Women’s degrees in People-oriented STEM—biology and health majors—doubled in just the eight years from 1971 (9 percent) to 1979 (18 percent), remained at roughly that level through the turn of the century, then surged again, standing at 27 percent of degrees in 2017. Rather, they wanted to use their math and science so that they could study topics that dealt with living things, especially people, rather than topics restricted to inanimate things and abstract concepts.
Occupations of College-Educated Women: Change in the 1970s and Early 1980s, Stability Since the Late 1980s
I turn from trends over time in women’s vocational interests and choices of undergraduate majors to trends over time in the jobs they hold as employed adults.[49]
To analyze the trends in the jobs that women hold, I took advantage of the Department of Labor’s O*NET database that assigns RIASEC scores to every occupation.50 I then used Prediger’s work to identify jobs that tilt toward the People and Things ends of the spectrum. The specifics are given in the note.[51]
Analyzing women’s choices over the years since 1970 is complicated by three contemporaneous trends: Women were rapidly entering the labor force through the mid-1990s, concentrated among married women; the percentage of People jobs as a proportion of all jobs was increasing as the service sector grew and manufacturing declined; and fewer people were getting married. Furthermore, all of these trends played out differently for college-educated women and women with no more than a high school education, so I will present the results separately for those two categories. All numbers refer to employed persons ages 25–54. I begin with the story for college-educated women.
The pair of figures below shows how occupations for college-educated women changed when categorized as People-oriented or Things-oriented.52 The figure on the left shows percentages of employed women in People jobs versus Things job. The figure on the right shows the sex ratios for the two types of jobs. The dotted lines represent the average for 2014–18, giving you a way to see how long the current situation has lasted.
Source: Author’s analysis, CPS. Sample limited to women ages 25–54.
Based on either perspective, observers in the late 1980s could be excused for thinking that men and women would converge within a few decades. From 1970 through the mid-1980s, the percentage of women in Things jobs had risen and the male-female ratio had plunged. If those slopes had been sustained, the percentages of men and women in Things jobs would have intersected around 2001. The percentages of men and women in People jobs would have intersected in the mid-1990s.
But convergence was already slowing by the late 1980s and had effectively stalled by 1990. The percentage of women in Things jobs hit 30 percent in 1990 and never subsequently surpassed it. Rounded to one decimal point, the male-female ratio in Things jobs reached 1.4 in 1982 and remained in the narrow range of 1.4–1.6 through 2018.53
For college-educated women, the distribution of vocational choices along the People-Things dimension changed substantially from the 1970s into the mid-1980s, but little has changed since then. It looks as if women were indeed artificially constrained from moving into a variety of Things occupations as of 1970, that those constraints were largely removed, and that equilibrium was reached around 30 years ago.
For Women with No More Than a High School Education, Not Much Changed
Overall, the story for high-school-educated women is like the Sherlock Holmes story with the dog that didn’t bark. In 1971, 30 percent of women were employed in People jobs. By 2018, that figure had risen to 39 percent. During the same period, the proportion of women employed in Things jobs moved in a narrow range from a high of 53 percent in 1971 to a low of 46 percent in 2004. As of 2018, the figure stood at 51 percent. Most of the rise in People jobs can be associated with the increased role of People jobs in the economy, but not all.54 The effects of the feminist revolution in the 1970s and early 1980s that were so evident for college-educated women were missing for women with no more than a high school education.
This finding surprised me. From the 1970s onward, well-paying jobs for women without college educations were opening in manufacturing and a wide variety of technical specialties that were formerly closed to them—not only opened up but opened wide. From the 1970s through the mid-1990s, males with no more than a high school education were dropping out of the labor force, working shorter hours, and were more often unemployed, while women with no more than a high school education were joining the labor force and working longer hours.55 A variety of evidence indicates that during the same period working-class men were becoming more unreliable employees.56 It appears that the time was right for women to enter these jobs if only for economic reasons, independently of vocational preferences.
Some women took advantage of those new opportunities. For example, among people ages 25–54 employed as mechanics or repairers, only 0.6 percent were women in 1971–75. In 2014–18, 2.1 percent were women—a proportional increase of 350 percent. But seen as a percentage of all employed women with no more than a high school degree, mechanics and repairers constituted 0.1 percent of employed women ages 25–54 in 1971–75 and 0.2 percent of them in 2014–18—such a tiny number that the increase in women mechanics had no effect on the trendline. In other blue-collar jobs, there were remarkably small increases in the percentage of jobs held by women. Among people ages 25–54 with no more than a high school degree employed in manufacturing jobs in 1971–75, an average of 21 percent were women. In 2014–18, that figure was 23 percent.
In sum: The effect of the feminist revolution on the vocations of college-educated women was real but quickly reached a new equilibrium. For women with no more than a high school education, it is as if the feminist revolution never happened.
Cross-National Sex Differences in Vocational Directions
As I discussed in chapter 3, the results of the 2015 administration of the PISA tests were familiar from previous ones, showing a small math effect size (–0.05) favoring boys and a moderate reading effect size (+0.32) favoring girls. In 2018, psychologists Gijsbert Stoet and David Geary published an analysis of the 2015 round of PISA tests that cast a new and informative light on what those results imply for the attraction of males to Things occupations and of females to People occupations.
The authors created a measure of the individual 15-year-old’s perso
nal strengths in science, math, and reading scores relative to their overall ability. The details are given in the note,[57] but it comes down to this: Suppose 100 students from a magnet school for the gifted take the PISA tests, and so do 100 students from an ordinary high school. Joan, from the magnet school, has raw science, math, and reading scores of 700, 650, and 600 respectively, for an average of 650 points. Her science score is 50 points above her personal average. Joe, from the ordinary high school, has raw science, math, and reading scores of 400, 350, and 300 respectively, for an average of 350 points. His science score is also 50 points above his personal average. So for both of them, despite the great difference in their raw scores, science is their relative strength. But how can we compare their respective relative strengths? The measure Stoet and Geary devised is a statistically suitable way for doing so—what I will call relative-strength scores (the authors called them intraindividual scores). We can then add up the 100 relative-strength scores for each school and compare them by sex, making statements about sex differences (if any) between the relative strengths of boys and girls in the two schools.
We can do the same thing comparing nations. The following table summarizes the 2015 PISA results for 62 countries showing mean scores by nation, relative-strength scores, and correlations with the Global Gender Gap Index.58
RELATIVE-STRENGTH SCORES FOR PISA-2015 WERE DIFFERENT EVEN WHEN MEAN SCORES WERE ALMOST THE SAME
Subject: Mathematics
Mean scores
Human Diversity Page 11
