Intelligence variability is not gender-dependent

Why the Greater Male Variability hypothesis is not an established fact

It is certainly understandable that many men feel that they must go on the defensive in the ostensible intellectual battle of the sexes. After all, we have been hearing for years about women surpassing men in college graduation rates, mean GPA, and income. Wait, one of those doesn’t belong, does it?

Indeed, when I was a boy of around 12 or 13, I was filled with righteous indignation that there was no such thing as a “masculist.” Most of the highest GPAs in my middle school belonged to girls. At various times I was told point-blank that girls were smarter than boys. “Who will defend us poor guys against the aggressive expansion of female achievement?” I despaired.

Fortunately, I have since grown up. I cannot say the same for someone who declares the very real phenomenon of cultural misogyny to be an “invisible bigotry.” I will now assume the role of the “angry mob,” presumably the unthinking reactionaries out to get the daring, truth-seeking proponents of a theory that society is too politically correct to even give a chance.

To start off, it is absurd to suggest that if for every woman complaining of unequal treatment there is a man saying “no, everything’s just fine the way it is,” then the jury must be out on institutionalized sexism. Following this logic, since for every poor single mother who needs help to survive and raise her children there is a well-off person saying they’ll be fine if they just pull themselves up by their bootstraps, we can’t justify giving any governmental assistance. And because every black person who says they experience racism is complemented by a white person declaring reverse racism, everything must even out so there’s no racism on the net; this logic works genuine miracles!

But now allow me to examine this Greater Male Variability (GMV) hypothesis in more detail. The hypothesis states that male intelligence is more variable than that of females. As far as I can tell, it was first suggested in 1894 by Havelock Ellis and was most recently brought to public attention after it was endorsed by economist and former Harvard President Larry H. Summers ’75. Undoubtedly, the idea that male intelligence has a flatter distribution than female intelligence has a great deal of support. The trouble with the hypothesis is that the evidence is not nearly as consistent as Yost would have you believe. For while it is true that many studies in various countries show greater variability in intelligence for boys, this is by no means universally true.

First, this effect is not consistent across race: A 2008 study using Minnesota state math assessments showed that at the 99th percentile, the male-to-female ratio was 2.06 for Whites, but 0.91 for Asian-Americans. There were more math-proficient Asian girls than boys.

Second, it is not consistent across countries: In a 2003 Trends in International Mathematics and Science study, one-third of the 50 participating countries showed either no significant disparity among variances between girls and boys or a disparity showing greater variability among girls. For example, while the variance ratio — a measure that is exactly what it sounds like — for boys versus girls in the U.S. was 1.19, in the Netherlands and Denmark the ratios were 1.00 and 0.99 respectively. If the males really do have greater variability in intelligence (generally and specifically in respect to mathematical ability), and this is in our genes as Yost postulates, shouldn’t the phenomenon be observable everywhere?

Among studies specifically geared toward the mathematically gifted — perhaps most apropos to our venerable Institvte’s policy — we may look to the Study of Mathematically Precocious Youth (SMPY). The SMPY researchers identify children 13 years of age and younger who are mathematically advanced and administer the SAT to them. Near the program’s inception in the early 1980s, the ratio of boys to girls who scored above a 700 on the math section was 13-to-1; in 2005 it was 2.8-to-1.

If you scoff at a 700, even for a 13-year-old (as I suspect you might), we can look at the truly exceptional level. As demonstrated by Hyde and Mertz (2009), nations’ performances on the 2007 Gender Gap Index correlates (r=0.44, p<0.05) with the proportion of girls on their International Mathematical Olympiad teams for the past two decades. They also offer explanations for the SMPY results: 1) increased immigration of Asians and Eastern Europeans have brought with them their cultural norms, 2) second-wave feminism and Title IX have opened math and science opportunities to girls in all levels of education, and 3) girls are taking more mathematics and science courses in high school due to changing graduation and college-entrance requirements.

Perhaps studies of children are not sufficiently satisfying, so let’s look to the awarding of PhDs in physical sciences and mathematics. In 1970, women accounted for 8 percent of such mathematics degrees and 5.5 percent of those in the physical sciences. In 2006 those numbers increased to 32 percent and 30 percent respectively; currently, women account for 48 percent of all mathematics degrees in general. Interestingly, physics shows far more abysmal numbers, a trend that is not well understood; according to the American Physical Society, women still only earn 12 percent of PhDs, compared with 3 percent 30 years ago.

On the whole, it would seem that the slow progress of Western society has had some effect on the number of women in the highest echelons of scientific and mathematical ability and achievement. Given that we are by no means at the level of perfect equality, we have reason to suspect that the admittedly inconsistent gender disparities in variability may one day vanish altogether.

But maybe all this progress is just political correctness and affirmative action. Still, one of the major reasons it is impossible to make any meaningful conclusions is that almost all studies of sex differences in intelligence ­— with all their inconsistencies ­— have been conducted in the West. To illustrate why broader testing might complicate things, consider the results of a recent test of non-verbal ability given to 2,700 sixth graders in the Philippines (Vista and Care, 2011). They found no significant differences in mean intelligence (a typical result) but found that males had greater variability in the upper half of the distribution, while females had greater variability in the lower half of the distribution.

Further, the GMV hypothesis may not even be necessary to account for the greater number of far-below-average males. There is still a genetic explanation, but more mechanical than due to specific encoding on the Y-chromosome and/or its consequent effects on the expression of other genes. For example, Fragile X syndrome is a major cause of mental disability and is a result of a mutation of the FMR1 gene on the X-chromosome. Its incidence is lower in females despite the irrelevance of the Y-chromosome because females have a “spare” X-chromosome that often has the functional FMR1 allele.

I will not deny that there is a great deal of support for the GMV hypothesis — and we certainly shouldn’t shy away from an uncomfortable conclusion simply because it doesn’t jibe with our preconceived notions — but because the studies that support it, especially the famous study of Scottish children, come from societies steeped in gender norms that favor the results the studies produce, it would be disingenuous and premature to consider the case closed in the face of a stubborn body of evidence that supports a totally different conclusion.