Human Diversity

by Charles Murray

  But simple quickly becomes complicated. Is the difference between the time men and women spend tending to young children artificially created by culture or driven by inborn male-female differences? How about the attraction of girl toddlers to dolls and boy toddlers to trucks? Male-female differences in college majors? Male-female differences in attraction to casual sex? Are they sex differences or gender differences?

  The sensible answer would seem to be “probably some of both,” with arguments about how much of which. At one level, that’s actually how the academic debate is conducted. The following chapters have hundreds of references to highly technical articles, adhering to normal standards of scientific rigor, published in refereed journals, arguing questions of nature and nurture, with male and female scholars making contributions on all sides on all topics. The tone is usually civil, and the conclusions are usually nuanced and caveated.

  But the women and men who are engaged in this endeavor are a rarefied group of neuroscientists and quantitative social scientists. Few of them seek publicity (many do their work as unobtrusively as possible), and they do not set the mood on college campuses. Since American second-wave feminism took off in the 1960s, the most visible feminist academics have rejected the possibility that there are any significant sex differences from the neck up. In my terminology, they have denied that men and women have any inborn differences in cognitive repertoires. A person’s gender “is an arbitrary, ever-changing socially constructed set of attributes that are culture-specific and culturally generated, beginning with the appearance of the external genitals at birth,” in the words of one of the most widely read feminist scientists in women’s studies courses, Ruth Bleier.15 It’s not a position with a lot of nuance. Gender is a social construct. End of story.

  The most famous illustration of what happens to those who question the orthodoxy is what befell economist Larry Summers. On January 14, 2005, Summers, then president of Harvard University, spoke to a conference on diversifying the science and engineering workforce.16 In his informal remarks, responding to the sponsors’ encouragement to speculate, he offered reasons for thinking that innate differences in men and women might account for some of the underrepresentation of women in science and engineering. He spoke undogmatically and collegially, talking about possibilities, phrasing his speculations moderately. And all hell broke loose.

  An MIT biologist, Nancy Hopkins, told reporters that she “felt I was going to be sick,” that “my heart was pounding and my breath was shallow,” and that she had to leave the room because otherwise “I would’ve either blacked out or thrown up.”17 Within a few days, Summers had been excoriated by the chairperson of Harvard’s sociology department, Mary C. Waters, and received a harshly critical letter from Harvard’s committee on faculty recruiting. One hundred and twenty Harvard professors endorsed the letter. Some alumnae announced that they would suspend donations.18 Summers retracted his remarks, with, in journalist Stuart Taylor Jr.’s words, “groveling, Soviet-show-trial-style apologies.”19 As if to validate that image, Lizabeth Cohen, a Harvard history professor, told reporters after attending the Summers self-criticism session that “[h]e regrets what he said, and I hope that he will prove that by taking constructive steps. We’re going to be in intense discussions with him over the next week.”20

  Since 2005, expanding knowledge about male-female differences has substantiated Summers’s speculations. The next five chapters review that evidence. The basics have been available to interested lay readers for years.[21] And yet elite gender studies departments still refuse to acknowledge the biological side of gender differences.[22] The degree to which the standard social science disciplines have also ignored this literature is an intellectual scandal. Evolutionary biologist Robert Trivers, whom you met in the introduction, has not held back:

  Once you remove biology from human social life, what do you have? Words. Not even language, which of course is deeply biological, but words alone that then wield magical powers, capable of biasing your every thought, science itself reduced to one of many arbitrary systems of thought.

  And what has been the upshot of this? Thirty-five wasted years and counting. Years wasted in not synthesizing social and physical anthropology. Strong people welcome new ideas and make them their own. Weak people run from new ideas, or so it seems, and then are driven into bizarre mind states, such as believing that words have the power to dominate reality, that social constructs such as gender are much stronger than the 300 million years of genetic evolution that went into producing the two sexes—whose facts in any case they remain resolutely ignorant of.23

  Despite the orthodoxy’s devotion to “words that have the power to dominate reality,” the state of knowledge about the observable differences in men and women has advanced enormously in the last 20 years. During those same years, the state of knowledge about sex differences in the brain has been transformed. The next five chapters give you an overview of the most important developments.

  1

  A Framework for Thinking About Sex Differences

  A few decades from now, I expect we will have a widely accepted comprehensive theory of sex differences that is grounded in neuroscience, genetics, and evolutionary psychology. Progress has already been made in that regard, but it is still at the frontiers of scholarship and bears no resemblance to low-hanging fruit. In any case, my purposes don’t require that level of sophistication. A simple framework for thinking about phenotypic sex differences is supported by a growing number of scholars. This framework also links up with recent findings about sex differences in the brain.

  A WORD ABOUT USAGE

  From now on I will usually refer to “sex differences” instead of “gender differences.” “Gender” was popularized in the 1960s to designate socially constructed differences.[1] But it turns out that there is no clear division between biological and socially constructed differences and no point in trying to pretend otherwise—which is what the widespread use of “gender” amounts to. In the technical literature, many scholars who write on these topics have resumed the use of “sex” to apply to all kinds of differences between males and females. So do I.

  The People-Things Dimension

  More than a century ago, Edward Thorndike, one of the founders of educational psychology, asserted that the greatest cognitive difference between men and women is “in the relative strength of the interest in things and their mechanisms (stronger in men) and the interest in persons and their feelings (stronger in women).”2 In 1944, Hans Asperger, for whom Asperger’s syndrome is named, hypothesized that the autistic cognitive profile is an extreme variant of male intelligence, which is another way of saying that normal males are more interested in things than people.3 On the female side, the quantifiable existence of a female advantage in “sociability,” as it had come to be called, was developed over the last half of the twentieth century among experts in personality.

  Putting these advances together with some discoveries in biology that I will discuss in chapter 5, Simon Baron-Cohen, director of Cambridge University’s Autism Research Centre, developed a theory of male-female differences that he described for a general audience in The Essential Difference: Male and Female Brains and the Truth About Autism, published in 2003. He coined the words systemizer and empathizer. In Baron-Cohen’s formulation, men are driven to understand and build systems. The defining features of a system are that it has rules and that it does something. It has inputs at one end and outputs at the other. In between are specific operations that translate the inputs into the outputs. “This definition,” Baron-Cohen wrote, “takes in systems beyond machines such as math, physics, chemistry, astronomy, logic, music, military strategy, the climate, sailing, horticulture, and computer programming. It also includes systems like libraries, economics, companies, taxonomies, board games, or sports.”4 Whatever the system may be, men are attracted to understanding what makes it tick.

  Understanding what makes human beings tick? Not so much. “The baby is crying because
it’s hungry” is something men can recognize as well as women (ordinarily, anyway). But entering into and responding to the state of someone else’s mind is a different matter. Empathy is required for that. Most men can do it, but on average, women are attracted to it more and do it better. It’s not just because women devote more attention to it. Entering into someone else’s mind calls on a different set of mental capabilities than the ones required for understanding a system.

  Empathizer as Baron-Cohen uses the word is not confined to understanding what’s going on inside the other person’s head. It also involves “the observer’s emotional response to another person’s emotional state.”5 Sympathy might be one part of the emotional response, but it can also be anger or concern. These responses may be used for altruistic or self-interested purposes. Good empathizers can make effective ministers to the grieving and effective therapists for the psychologically troubled—but, using the same neurocognitive tools, they can also make effective arbitrators of disputes, interrogators of criminal suspects, managers of people, or election-winning politicians.

  Other scholars of sex differences have been finding differences in academic interests, careers, and life choices that break along the lines of systemizing and empathizing but that also lend themselves to the broader and simpler difference that Thorndike identified—in choice after choice, men are attracted to options that have more to do with things while women are attracted to options that have more to do with people. That’s the simple theory of the case I bring to the chapters on sex differences: Women and men divide along the People-Things dimension.

  Lest there be any misunderstanding: I am talking about statistical tendencies, not binary divisions. Many men and women possess trait profiles more typical of the other sex.[6] But these tendencies are strong enough to create distinctively different distributions on important traits of personality, abilities, and social behavior.

  First Interlude: Interpreting How Big a Sex Difference Is

  I warned you in “A Note on Presentation” that I would occasionally be interrupting my narrative to explain technical terms. This is the first such interlude. Some of you are already familiar with the term I will be explaining, effect size, but I urge you to continue reading nonetheless. The interpretation of effect sizes plays a significant role in how one interprets the evidence.

  In the following chapters, I compare men and women on dozens of traits. They are based on many kinds of measures—answers to questionnaire items, scores on tests, and ratings of observed behavior, to name just a few. Researchers need a common metric for expressing the differences that these comparisons reveal.

  To see what this metric must do, think in terms of a simple measure like height. In one sense, an inch gives a common metric for measuring height. You can express the height of anything with it. In another sense, it doesn’t tell us much. For example, how big is a difference of six inches in height? In absolute terms, it’s always the same. But how big is a six-inch difference if we are talking about the height of elephants? The height of cats? The answer depends on the average height of the things you are measuring and how much height varies among the things you are measuring. You need a way to express height in a way that means the same thing for elephants relative to other elephants and cats relative to other cats.

  We need the same kind of metric to talk about sex differences across cognitive repertoires. That metric is based on a statistic called the standard deviation, described in detail in Appendix 1. In many cases, including the ones we will be dealing with, the standard deviation applies to a normal distribution, also known as a bell curve. To get from bell curves to effect sizes, let’s stick with the example of height.

  The contemporary mean height of American women ages 20 or older is 63.6 inches. The comparable mean for men is 69.0 inches. Most people are clustered within a few inches of those means, but successively smaller numbers of people are three, four, five, and six inches from the mean. A tiny proportion of people are a foot or more from the mean. The nationally representative database of people that produced those numbers had these distributions:

  Source: Fryer, Gu, Ogden et al. (2016).

  The dotted vertical lines show the means for women and men. The gray horizontal bar shows the difference between the two, which I call the “raw effect size.” Dividing it by the pooled standard deviations of the two groups gives us a way to express magnitude that can be compared across different traits.

  An effect size is denoted as d. To calculate d for height, I subtracted the male mean from the female mean, producing a difference of –5.4 inches. The pooled standard deviation is 2.9 inches, so d equals –5.4 ÷ 2.9, which works out to an effect size of –1.86. This is an extremely large effect size. Most sex differences are much smaller and the distributions have much more overlap.

  Note that the sign of d (negative or positive) is arbitrary. If I had subtracted the female mean from the male mean, the effect size wouldn’t have changed, but the sign would have been positive. Just so you know, in this book my default will be to subtract the male mean from the female mean in calculating sex differences. Therefore negative d values will always indicate that males are higher than females on the trait in question, whether “higher” means something good, bad, or neutral.

  Two questions are crucial to assessing the importance of sex differences: When is an effect size big enough to be interesting? Should individual effect sizes be treated individually or aggregated?

  When Is an Effect Size Big Enough to Be Interesting?

  Jacob Cohen, who originated Cohen’s d, inadvertently set the standard for interpreting effect sizes (he had a different purpose in mind). His list was subsequently expanded by Shlomo Sawilowsky. Under these guidelines, a d value of 0.01 = very small, 0.20 = small, 0.50 = medium, 0.80 = large, 1.20 = very large, and 2.00 = huge.7

  The guidelines were well-intended but have often proved to be pernicious in practice. As Cohen himself took pains to point out, the importance of a given value of Cohen’s d depends on the specific topic you are examining.[8] In 2019, psychologists David Funder and Daniel Ozer took on what they called the “nonsensical” standard set by Cohen, arguing that the interpretation of effect sizes should be guided by their consequences. In the case of a drug for curing a deadly disease that has a relatively small success rate, the effect of a success is a saved life—a consequence that can be important even if the effect size is small. In the case of a small effect size that has many repetitions, it’s the cumulative effect that’s important. For example, a study that tracked two million financial transactions found that the correlation between a person’s score on a measure of extraversion and the amount spent on holiday shopping is just +.09. “Multiply the effect identified with this correlation by the number of people in a department store the week before Christmas,” the authors wrote, “and it becomes obvious why merchandisers should care deeply about the personalities of their customers.”9 They offered a new set of guidelines based on the correlation coefficient (r). In the summary that follows, I have replaced the value of r with the equivalent value of Cohen’s d.

  The authors argued that an effect size of .10 “is ‘very small’ for the explanations of single events but potentially consequential in the not-very long run,” while an effect size of .20 “is still ‘small’ at the level of single events but potentially more ultimately consequential.”10 Other scholars have advocated similar guidelines for interpreting small values of d.11 But their treatment of “small” collides with the position taken by the most influential work arguing for small sex differences in cognitive repertoires—the “gender similarities hypothesis” originated by psychologist Janet Shibley Hyde in the September 1985 issue of American Psychologist, the flagship journal of the American Psychological Association. Here is her statement of the hypothesis:

  The gender similarities hypothesis holds that males and females are similar on most, but not all, psychological variables. That is, men and women, as well as boys and girls, are more alike tha
n they are different. In terms of effect sizes, the gender similarities hypothesis states that most psychological sex differences are in the close-to-zero (d ≤ 0.10) or small (0.11 < d < 0.35) range, a few are in the moderate range (0.36 < d < 0.65), and very few are large (d = 0.66–1.00) or very large (d > 1.00).12

  The inclusive definition of “small” to include everything up to a d of .35 dictates her interpretation of the literature. Hyde reviewed 46 meta-analyses of psychological sex differences and concluded that of 124 classifiable effect sizes, 78 percent were small or close to zero by her definition.13

  For Hyde, Cohen’s guidelines “provide a reasonable standard for the interpretation of sex differences effect sizes.”14 She acknowledged that in some cases—cure rates for disease, for example—a small effect size can have important effects. But, she argued, “[I]n terms of costs of errors in scientific decision making, psychological sex differences are quite a different matter from curing cancer. So, interpretation of the magnitude of effects must be heavily conditioned by the costs of making Type I and Type II errors for the particular question under consideration.”15

  Type I error refers to a false positive finding—in this case, wrongly concluding that a sex difference has been found. Type II error refers to a false negative finding—mistakenly concluding that no difference exists. Hyde was worried about the consequences of making a Type I error. She went on to give examples of the ways that inflating sex differences have real-world costs. For example, the idea that women are more nurturing than men backfires when it comes to the workplace: “Women who violate the stereotype of being nurturant and nice can be penalized in hiring and evaluations,” Hyde wrote, citing evidence to that effect.16