by Sam Sommers
Just ask Larry Summers.
MATH IS HARD
It’s inescapable: women are vastly underrepresented in the fields of science, engineering, and mathematics. In January 2005 the National Bureau of Economic Research held a conference in Cambridge, Massachusetts, devoted to exploring this disparity. One of the headlining speakers was Larry Summers, then president of Harvard University.
In a lunchtime talk, Summers focused on the gender gap in science and engineering positions at elite universities, evaluating three possible explanations.4 The first, as he referred to it, was the “high-powered job” hypothesis, the idea that women are less likely to consent to the schedule and family sacrifices necessary to attain such a position. Second, he discussed the possibility of gender differences in innate math and science ability. Third, he addressed societal considerations, such as socialization pressures that steer boys and girls toward different disciplines, and the potential for discrimination in hiring and promotion decisions.
Within two months, prompted in large part by this very talk, the Harvard faculty passed a motion of “no confidence” in the leadership of their president. By the following winter, Summers had resigned.
What was so incendiary about his remarks? After all, Summers just articulated, in his own words, “three broad hypotheses about the sources of the very substantial disparities that this conference’s papers document.” And he was anything but dismissive of the problem—to the contrary, he ended his talk by stating that “I think we all need to be thinking very hard about how to do better on these issues.” So why the controversy?
Summers ran into trouble because he did more than present three hypotheses worthy of exploration. He also ranked these explanations in terms of how he saw their relative importance. There would have been no controversy had he stopped after the suggestion that the relative dearth of female scientists and mathematicians could be attributable to 1) family-related pressures or 2) inborn differences in aptitude or 3) societal expectations. But he continued as follows: “In my own view, their importance probably ranks in exactly the order that I just described.”
It’s easy to see why so many at Harvard were distressed by their president’s belief that innate ability plays a greater role in the underrepresentation of females than societal or institutional factors. After all, he’s the guy who signed off on faculty performance reviews and pay raises, and here he was endorsing a view of the gender gap as largely biological and inevitable. Even before he gave this talk, many faculty members already had questions about Summers’s commitment to gender equity: during his administration, tenured job offers to women at Harvard had dropped dramatically, to the point where only four out of thirty-two new hires the previous year were female.5 So while some outsiders and media pundits decried Summers’s ouster and celebrated him as a victim of the overzealous speech police, it’s easy to appreciate the concerns held by those who were working under his immediate supervision.
But even from across town and outside his jurisdiction, I found Summers’s comments disquieting. Living in Boston in the aftermath of the controversy, I felt the need to bestow upon him a parenthetical middle name anytime casual conversation veered in his direction—as in, Larry (No Relation) Summers. This desire to distance myself from his comments reflected not political considerations but rather more scientific concerns. You see, beyond being controversial, Summers’s conclusions were also flat-out wrong. And twice over, at that.
Take a closer look at one brief passage from Summers’s remarks: In arguing his position, he claimed that “the human mind has a tendency to grab [on] to the socialization hypothesis when you can see it. And it often turns out not to be true.”
Really?
Let’s take these assertions one at a time. First, the argument that the human mind tends to grab on to socialization hypotheses. This was an off-the-cuff remark for which Summers offered no real supporting evidence. And it flies in the face of everything you’ve now learned regarding the WYSIWYG mentality.
Just think about it: When the mom at the playground tries to excuse her son’s rambunctiousness, does she latch on to the socialization hypothesis, suggesting that he’s only acting this way because this is what society expects of boys? Of course not—she shrugs her shoulders and says something along the lines of “Well, you know how it is with boys.”
In analyzing the actions of the cheating husband—be he politician, athlete, or next-door neighbor—do we veer toward the socialization account, arguing that society is simply more tolerant of such bad behavior from men versus women, thereby reinforcing the male tendency for infidelity? No, our first move is in the opposite direction, toward debate about whether monogamy runs contrary to the way that men are naturally wired.
And when the best-selling author pitches his manifesto on the psychology of gender differences, does he title it “Men Are Taught to Act Like Martians, Women Learn to be Venusian”? Clearly, no. He opts for the quintessential WYSIWYG thesis, that men and women might as well be beings from different planets.
Despite Summers’s claim, we don’t gravitate toward the socialization hypothesis—quite the contrary. Our knee-jerk reaction to gender disparity is to offer internal, innate, and immutable explanations. Often, it’s only the concern of appearing sexist—our deference to political correctness—that leads us to claim otherwise in public. Larry Summers had it backward.
Admittedly, though, it’s the second half of Summers’s quoted statement that’s the more important part. So what to make of this claim, the idea that the socialization hypothesis “often turns out not to be true”? If Summers was right, it would lend empirical heft to the argument that he was unfairly pilloried for his remarks. To the extent that gender differences in domains such as science and math are consistent across situations—that is, resistant to variations in context and expectation—his conference comments, while still controversial, would at least carry the stamp of research support.
Unfortunately for Summers, the data aren’t kind to him on this count, either. But don’t feel too bad for Not-My-Uncle Larry—I’m sure he’ll land on his feet somewhere. You know, like director of economic policy at the Obama White House, for starters.
ALMOST A DECADE BEFORE Summers’s talk, three researchers at the University of Michigan—Steve Spencer, Claude Steele, and Diane Quinn—set out to test just how entrenched the gender difference in math is.6 Their first study was straightforward: they recruited twenty-eight male and twenty-eight female college students to take a difficult standardized test. These students were all high achievers—to be eligible for the study, they had to have scored in the eighty-fifth percentile or better on their math SAT.
The study’s results were also straightforward: the men did much better than the women. In fact, the average score for male test takers was more than twice as high as the average for females, a gender difference practically begging for a simple Mars/Venus/Larry Summers explanation.
But the researchers refrained from jumping to conclusions about sex-based differences in math aptitude. Instead, they kept digging. They wondered why such a gender gap would emerge when all these men and women had enjoyed previous success in math. The women in the study spent just as long working on each problem as did the men—why were their scores so much lower? To answer these lingering questions the researchers ran another study that included more than a simple gender comparison. This time, they also varied the context in which the test was taken.
Specifically, half of this new group of men and women again took a math test under normal circumstances. The other half received a different set of instructions that changed the entire context of their experience. Before the test, this second group was told that while some previous research had found evidence of gender differences in math ability, other studies hadn’t. These students heard that the test they were about to take fell into the latter category—it had been found to avoid any type of gender disparity.
This little change in procedure made a big, big difference
.
Students who took the test without additional instructions once again exhibited a gender gap: men’s scores were almost three times higher than women’s. Students who took the test under the impression that it was gender-balanced? No gender difference at all. The average scores for these men and women were nearly identical.
It’s a pretty amazing finding. All the students took exactly the same math test. Under normal circumstances, the average man outperformed the average woman, a disparity consistent with the idea of entrenched, even inborn differences in math ability. But tell students that the test was engineered to be “gender-neutral”—you know, no word problems involving football blitzes, testicle care, and the like? Then the sex difference vanished entirely. Just like that.
How a test is described is far from the only situational factor that can eliminate (or exacerbate) the gender gap in math. Asked to solve math problems in mixed company, women don’t perform as well as men, but this underperformance goes away when the test is administered in single-sex groups.7 Shown a series of ads depicting girls as fixated on boys and shopping, women do poorly on a subsequent math test, yet there’s no gender difference after they watch commercials about intelligent and articulate women.8
And in a study that sounds eerily like a recurring nightmare I had in junior high, researchers even examined the effects of taking a math test while wearing a bathing suit.9 How, exactly, did they pull this off? Participants were told that the study was about consumer preferences. Led to a private dressing room with a full-length mirror, they were presented with a rack of swimsuits—trunks for men and one-pieces for women—and asked to try on the one closest to their size. Then, much as those who climb Everest must stop every few thousand feet to reacclimate, students were told that they’d have fifteen minutes to get used to the unfamiliar clothes they were now wearing. To pass the time, they could help some researchers in the neighboring department of education by completing an ostensibly unrelated test of mathematical aptitude.
Clad in water-repellent Lycra, men outperformed women on the math test. For the fortunate participants asked to try on and evaluate a sweater instead, the gender difference was far smaller.
What do we learn from these findings? Namely, that the gender disparity in math isn’t entrenched and unavoidable. It depends on context, perspective, and expectation. It’s actually surprisingly fragile. You can’t put much stock in the notion of inborn, immutable differences in aptitude when minor tweaks to a test’s instructions—or, for that matter, a nice cardigan—reduces or even wipes out the gender gap.
These very different studies converge upon a situational conclusion: remind women of the low expectations society holds for them in math and they will, indeed, underperform. Whether in the form of purportedly scientific conclusions regarding the genetic superiority of the male brain10 or a bathing suit that conjures up thoughts of female objectification, simple reminders of gender-based stereotypes are threatening enough to undermine actual math performance.
Think of how little it takes in this research to lead women to worry that they might confirm the expectation that they aren’t cut out for math. Merely having men in the room is enough. In fact, just thinking about math does the trick: even though women do fine when they’re told a math test is gender-neutral, their default tendency is to assume otherwise. And this message that girls can’t do math is a self-fulfilling one. Recent education research demonstrates that it’s reinforced in the classroom itself, as female elementary school teachers’ own anxieties about math predict increased anxiety and decreased performance among their female students.11
The ease with which women can be prompted to think about low math expectations reflects a reality in which the onslaught of gendered messages begins early in life and never really lets up. And I’m not just talking about my daughters’ quilts. Remember the notorious talking Barbie doll from the early 1990s that cheerfully reminded girls that “Math is hard”? OK, so the exact quote was actually “Math class is tough,” but same idea. Thanks to movies, TV shows, toys, and blankets, young girls don’t even have to leave the comfort of home to learn what’s expected of them when it comes to math. The answer is not very much.
Contrary to Larry Summers’s suggestion, you can’t explain the gender gap in the sciences and math without considering the major role played by social forces. A predominantly biological account doesn’t square with the data. Not to mention that it never really made sense in the first place why testosterone would draw men to the Pythagorean theorem like some mathematical version of a monster truck rally. How, exactly, is the Y chromosome supposed to help with long division? Sure, over generations and generations, natural selection can lead men and women to evolve in different ways, but why would any of them involve trigonometry?
As one last example, consider that in 1983 boys outnumbered girls thirteen to one in the ranks of students scoring 700 or better on the math SAT. Almost thirty years later, that ratio is less than three to one. Two and a half decades is a long time: long enough in the United States for five different presidents and a doubling of postage rates. But it’s nowhere close to long enough for evolution to have reversed course or for hardwired differences between the male and female brain to have evened out. Those explanations just don’t cut it.
Arguing that the gender gap in math performance results primarily from inborn, inescapable differences in aptitude is more than just politically incorrect. It’s also wrong.
READY TO RUMBLE?
Perhaps you never really bought into the idea of male math superiority. So maybe the fragility of that gender difference doesn’t strike you as particularly surprising or impressive. Well, then, how about the granddaddy of all gender differences? What about the well-documented conclusion that men are more aggressive than women? Could situational forces really have anything to do with that gender gap?
As cited in the opening of this chapter, scores of studies have found males to be more physically aggressive than females, regardless of age or culture. Moreover, much like science and engineering are fields disproportionately dominated by men, so is the act of homicide, both in terms of perpetrators and victims. And there are biological explanations for such gender differences: testosterone has been directly linked to aggression in studies involving people as well as animals.
But the nature of this gender difference depends, first and foremost, on how you define “aggression.” When talking in strictly physical terms, males are more aggressive than females. However, dictionary definitions of aggression aren’t limited to physical acts: rather, they describe a more general category of hostile behavior intended to cause injury. When you cast this wider net, you find that women are actually as aggressive as men—it’s just that their aggression often looks different.
Child development research has found that starting in early elementary school, boys are more likely to engage in direct forms of aggression like physical domination and verbal assault, while girls more often practice indirect efforts to cause harm.12 This Mean Girls route to aggression focuses on the manipulation of social relations. Such as, for example, campaigns to convince the group not to be friends with a particular child. Gossip. Or announcing to Mrs. Robbins’s entire fifth-grade class that Sam’s shirt wasn’t “a real polo shirt” because the guy on the horse was holding a flag instead of a mallet.
Damn you, Knights of the Round Table® and your deceptively haute logo!
In strictly physical terms, aggression is more of a male tendency. But defined more socially or relationally, women assume the lead in this Pyrrhic battle of the sexes. And when you consider “aggression” in more general terms, it becomes difficult to identify a consistent gender difference in either direction.
Even if we stick to the realm of physical aggression, however, the gender gap isn’t as entrenched as we think it is. When behavioral scientists study aggressive behavior in a research laboratory—by, for example, giving adults the opportunity to administer electric shocks or blast another person w
ith loud bursts of white noise through headsets—men regularly emerge as more likely to aggress. This gender difference goes away, though, when participants are first provoked.13 That is, while women are indeed less likely than men to initiate an aggressive interaction, they tend to be just as physically aggressive in response to insult or direct threat.
Moreover, you’re already familiar with another contextual factor that can eliminate gender differences in aggression: direct orders from authority. Milgram’s famous obedience study was also an examination of aggression, as respondents believed that they were giving painful (and even lethal) shocks to a fellow participant. Though most would have predicted otherwise, Milgram found no evidence of a gender gap in aggressive behavior: women in his study performed no differently than men, administering shocks just as great in number and voltage.
So women are just as physically aggressive as men after provocation or in the face of direct orders. And women aren’t lacking in the general drive to aggress—rather, they just tend to channel this impulse in different, less physical ways than men do. These conclusions don’t jive with the idea of a gender difference in aggression based on innate or biological factors. Instead, it seems more like women have many of the same aggressive tendencies as men, but they’re compelled to keep these feelings at bay most of the time. It’s as if something else causes them to hold back, or at least to aggress in ways that are less overt or easily recognized.
It’s those quilts again. Or, more precisely, those ubiquitous societal norms regarding gender. Women aren’t supposed to be aggressive, but men in a scuffle are just boys being boys. Thus, women often refrain from showing an aggressive side unless they have an obvious excuse for it. Or they try to inflict harm, but in subtler ways.
