by Kenneth W. Krause.
Kenneth W. Krause is a contributing editor and “Science Watch” columnist for the Skeptical Inquirer. Formerly a contributing editor and books columnist for the Humanist, Kenneth contributes regularly to Skeptic as well. He may be contacted at email@example.com.
In a letter to King Frederick II of Prussia, Voltaire wrote of his lover and French noblewoman, Émilie du Chatelet, that he considered her “a great man whose only fault was being a woman.” Privately trained and, among eighteenth-century females, exceptionally well-versed in math and physics, du Chatelet’s French translation of Newton’s Principia Mathematica remains the definitive text.
Nonetheless, as Voltaire’s ironic letter suggests, du Chatelet was less than delighted with the plight of intelligent, science-minded women. Writing in about 1735, she confessed to “feel the full weight of prejudice which so universally excludes us from the sciences.” Plainly, du Chatelet believed it was France’s culture and education system and not the female brain that was responsible for the inequity.
In 2005, the blazing-hot topic of female underrepresentation in high-end science was addressed by American economist and then Harvard University president, Lawrence Summers. “In the special case of science and engineering,” he famously suggested, “there are issues of intrinsic aptitude, and … those considerations are reinforced by what are in fact lesser factors involving socialization and continuing discrimination.”
A notoriously impassioned debate ensued. Within a few months, Harvard’s arts and sciences faculty passed a motion demonstrating a “lack of confidence” in their president’s leadership. Conversely, when asked whether Summers’s comments were intellectually appropriate, prominent cognitive scientist, Steven Pinker, responded as follows:
Good grief, shouldn’t everything be within pale of legitimate academic discourse, as long as it is presented with some degree of rigor? That’s the difference between a university and a madrassa. There is certainly enough evidence for the hypothesis to be taken seriously.
Summers resigned the following year, but his provocative challenge—“I would like nothing better than to be proved wrong”—was not forgotten.
No one seriously disputes the statistical facet of female underrepresentation among the higher echelons of the science, technology, engineering, and math (STEM) fields. Recent data from the U.S—collected by Cornell University researchers, Stephen Ceci and Wendy Williams—leave little room for disagreement. In 2005, PhDs were awarded to women as follows: 30% in math, 21% in computer science, 14.3% in physics, and 8.4% in mechanical engineering. Females were hired to tenure-track university positions as such: 26.8% in math, 20% in computer science, 16.8% in physics, and 18% in mechanical engineering. Finally, full professorships were awarded to women as follows: 7.1% in math, 10.3% in computer science, 6.1% in physics, and 4.4% in mechanical engineering (Ceci and Williams, 2011).
Rather, the real quarrel centers on the disparity’s likely causes. Was du Chatelet correct to attribute the gender gap to persisting discrimination? Was Summers justified in suggesting the possibility of disparate abilities? Or perhaps the issue is substantially more complicated. If so, do the statistics represent a serious social problem demanding intervention, or just a natural and acceptable dissimilarity between human genders?
In the opening pages of a new book on the subject, Ceci and Williams recognize that, although differences in aptitude occasionally register during early childhood, “the size of the male advantage accelerates” later, beginning in puberty (Ceci and Williams 2010, ix-x). By the end of high school, the authors continue, boys are much more likely to be seated at the “right tail of the distribution”—in the top 10%, 1%, or 0.1%.
Ceci and Williams offer three examples to illuminate the phenomenon. First, Honors Math 55 (Advanced Calculus and Linear Algebra) at Harvard University. Reportedly the most intimidating math class in the country, each year the majority of enrolling students drop out within a few weeks. The distribution by the bitter end of 2006 was, for lack of a better word, dumbfounding: according to The Crimson newspaper, “45 percent Jewish, 18 percent Asian, 100 percent male.”
Second, the Scholastic Assessment Test—Mathematics (SAT-M). Twice as many boys as girls achieve a score of 650 (19% versus 10%) or 700 (10% versus 5%). According to Ceci and Williams, “The farther out on the right tail one goes (toward the top 0.01%, or 1 in 10,000), the fewer females there are.” Males, in fact, “are sometimes overrepresented by a factor of 7 or more to 1.
Finally, the Putnam Mathematical Competition—a 6-hour intercollegiate test for U.S. and Canadian students administered every December by the Mathematics Association of America. Putnam winners have gone on to lead illustrious careers in math, and several have become Nobelists and Fields medalists. Predictably, by this point, females are rare among the top five scorers, who are dubbed Putnam Fellows. Since 2000, in fact, only three of 51 Fellows were women.
Clearly, we have cause to be concerned. But researchers aren’t convinced that girls lack the necessary skills. In 2008, for example, a team led my University of Wisconsin psychologist Janet Hyde concluded of American students that, at least “for grades 2 to 11, the general population no longer shows a gender difference in math skills, consistent with the gender similarities hypothesis,” which proposes that males and females are similar on most, but not all, psychological variables. (Hyde et al., 2008).
After analyzing the math scores of more than seven million students from across the U.S., Hyde found “trivial differences” on average, coupled with some “unexplained” evidence of slightly greater variability among males. Unfortunately, the state administered tests were incapable of assessing the students’ relative abilities to solve more complex problems—in other words, to test for skills most crucial for advanced work in STEM careers.
In early 2010, two members of Hyde’s team collaborated with Villanova University psychologist Nicole Else-Quest to probe the issue more broadly and inclusively by evaluating data gathered from two previous studies, the Trends in International Mathematics and Science Study (TIMSS) and the Programme for International Student Assessment (PISA) (Else-Quest et al., 2010). TIMSS tests were generally easier and more sensitive to curricula or institutions; PISA exams were more difficult and emphasized math literacy and its practical application. Together, these data sets represented 493,495 students aged 14-16 from 69 nations.
In terms of math achievement, Else-Quest’s results, like Hyde’s, substantially supported the gender similarities hypothesis because the sizes of all mean effects were “very small, at best.” The largest effect was in the space/shape domain of the PISA, consonant with historical evidence of male superiority in mental rotation skills. The team was quick to point out, however, that gender disparities in this area can be mediated through appropriate education, as other studies have shown.
Even so, Else-Quest offered two possible explanations for the PISA gender gap. The first is rooted in the greater male variability hypothesis, which predicts no discrepancies on average, but more top performers among males. Boys, however, didn’t outperform girls on the most challenging TIMSS problems demanding creative or strategic reasoning. “Thus,” the team cautioned, “comparisons between TIMSS and PISA regarding test difficulty should not be overplayed as support for the greater male variability hypothesis.”
The second explanation spotlights society-based gender inequities. More pertinent here, Else-Quest suggests, is the gender stratification hypothesis, which posits performance gaps closely related to cultural variations in opportunities for females. Indeed, effect sizes revealed considerable variability across nations, and, despite similar achievement levels, boys regularly reported more positive attitudes and affects toward math. So, do societal valuations of math proficiency among young females affect achievement? Consistent with stratification, the team judged that “girls will perform at the same level as their male counterparts when they are encouraged to succeed, are given the necessary educational tools, and have visible female role models excelling in mathematics.”
At about the same time, Ceci, Williams, and Cornell colleague Susan Barnett reviewed more than 400 articles and book chapters to reconcile competing claims of biological and sociocultural causation (Ceci et al., 2009). In the end, they pronounced the evidence for each contention to be both contradictory and inconclusive.
First, if underrepresentation were solely the function of ability, women should still occupy at least twice as many high-end science positions as they do. Second, although women still experience unequal childrearing responsibilities in many or all cultures, such inequity should result in women having inadequate time for all professional careers to the same degree, which doesn’t seem to be the case.
Disparate abilities and cultural attitudes might play important roles, the trio agreed, but only a “confluence of factors” can account for all salient data. “Of these factors,” they concluded, “personal lifestyle choices, career preferences, and social pressures probably account for the largest portion of variance.” Math-proficient women tend to prefer non-math careers and are more likely to relinquish them as they advance. They are also more likely than men to possess outstanding verbal competence and, thus, the additional option to flourish in law, the humanities, or medicine.
According to Ceci and Williams, “The tenure structure in academe demands that women who have children make their greatest intellectual achievements contemporaneously with their greatest physical and emotional achievements, a feat fathers are never expected to accomplish” resulting in career choices “men are not required to make.” But in order to counteract the “childbearing penalty,” as he terms it, they suggest that universities consider deferred start-up tenure track positions and part-time work that segues into full-time tenure-track employment as children mature.
Finally, on February 7 of this year, Ceci and Williams published a hard-hitting and no doubt divisive paper addressing persistent and pervasive claims of sex discrimination in interviewing, hiring, and grant and manuscript reviewing. (Ceci and Williams, 2011). After reviewing twenty years of data, Ceci and Williams—married with three daughters of their own—decided that the evidence of discrimination against women in math-intensive fields is “aberrant, of small magnitude” and “superseded by larger, more sophisticated analyses showing no bias, or occasionally, bias in favor of women.”
In agreement with their most recent work, Ceci and Williams surmised instead that the gender gap results primarily from women’s career preferences and fertility and lifestyle choices, “both free and constrained.” Adolescent girls tend to gravitate toward careers focusing on people as opposed to things, the couple found, and female PhDs interested in childrearing are less likely to apply for or maintain tenure track positions. As a secondary explanation, Ceci and Williams again pointed to evidence for upper tail disparities in cognitive ability.
The authors briefly addressed the thorny question of solutions as well, emphasizing the need to move beyond historical causes. But if the existing bases of female underrepresentation are mostly a function of female preferences—for non-math or less math-intensive careers, or for reproduction and childrearing—is it really “underrepresentation” in any meaningful sense of the word? If so, does it represent a problem justifying remedies involving sacrifices from others, average taxpayers in particular? Perhaps some arrangements would benefit many and harm none. But others implicating the commitment or reallocation of valuable resources ought to be vetted thoroughly at all levels of society.
Ceci, S.J., Williams, W.M. 2010. The Mathematics of Sex: How Biology and Society Conspire to Limit Talented Women and Girls. New York: Oxford University Press.
Ceci, S.J., Williams, W.M. 2011. Understanding current causes of women’s underrepresentation in science. Proceedings of the National Academy of Sciences, USA, DOI: 10.1073/pnas.1014871108.
Ceci, S.J., Williams, W.M., Barnett, S.M. 2009. Women’s underrepresentation in science: sociocultural and biological considerations. Psychological Bulletin 135(2): 218-261.
Else-Quest, N.M., Hyde, J.S., Linn, M.C. 2010. Cross-national patterns of gender differences in mathematics: a meta-analysis. Psychological Bulletin 136(1): 103-127.
Hyde, J.S., Lindgerg, S.M., Linn, M.C. 2008. Gender similarities characterize math performance. Science 321: 494-495.