Are Men Smarter than Women?

Psychometricians have been arguing about this for a century. Now, there's a new meta-analysis.

Until the invention of IQ testing in the 20th Century, few male intellectuals assumed that women were as smart as men on average. Schopenhauer’s 1851 essay “On Women” was particularly extreme, but probably not unrepresentative:

Women are directly adapted to act as the nurses and educators of our early childhood, for the simple reason that they themselves are childish, foolish, and short-sighted — in a word, are big children all their lives, something intermediate between the child and the man, who is a man in the strict sense of the word. …

The nobler and more perfect a thing is, the later and slower is it in reaching maturity. Man reaches the maturity of his reasoning and mental faculties scarcely before he is eight-and-twenty; woman when she is eighteen; but hers is reason of very narrow limitations. This is why women remain children all their lives, for they always see only what is near at hand, cling to the present, take the appearance of a thing for reality, and prefer trifling matters to the most important. It is by virtue of man’s reasoning powers that he does not live in the present only, like the brute, but observes and ponders over the past and future; and from this spring discretion, care, and that anxiety which we so frequently notice in people. The advantages, as well as the disadvantages, that this entails, make woman, in consequence of her weaker reasoning powers, less of a partaker in them. Moreover, she is intellectually short-sighted, for although her intuitive understanding quickly perceives what is near to her, on the other hand her circle of vision is limited and does not embrace anything that is remote; hence everything that is absent or past, or in the future, affects women in a less degree than men. This is why they have greater inclination for extravagance, which sometimes borders on madness. Women in their hearts think that men are intended to earn money so that they may spend it, if possible during their husband’s lifetime, but at any rate after his death.

And so forth and so on.

But then in the 20th Century, scientists started measuring intelligence rather than just opining about it, and whaddaya know? Men and women weren’t all that different.

But Charles Spearman, who came up with the concept of a general factor of intelligence g, pointed out in 1904 that males and females scored roughly the same across the various subtests of intelligence he had tried out.

In 1916, Lewis Terman, in making up the first American IQ test, the Stanford-Binet, decided that so far as he could tell, males and females were about equal in intelligence, so he was going to favor including questions on which the sexes score about the same over questions on which they don’t. If he asked, say, “Who has a higher lifetime batting average, Ty Cobb or Tris Speaker?” and 80% of the boys and 55% of the girls got it right, he’d figure that was a bad question and junk it. (Not that he ever considered questions as bad as my example.)

This custom, of course, tended to lock in his finding of equality of the sexes.

This puzzles critics of the race gap found on IQ tests. If test-makers can censor questions so that boys and girls score about the same, why can’t they select questions on which blacks and whites score the same?

The answer is that the sex gap, whatever it may be, is a lot smaller than the race gap. So you can have a rule of preferring questions without a sex gap without sacrificing much, if any, predictive validity. But nobody has ever figured out a way of getting blacks and whites to score the same without making the test fairly useless for predicting real world performance.

This is tied into a more general observation about intelligence testing: it’s easy to come up with a not bad intelligence test (all sorts of different approaches will more or less work) but it’s extremely hard to come up with a really good cognitive test.

For example, consider a simple question like who averages higher, black or whites? Just about any test will come up with the finding that whites average higher than blacks.

But once you get into trying to answer harder questions, like: is there a sex gap? Are there different sized sex gaps between blacks and whites? Is the size of the difference in the size of the sex gap between blacks and whites changing over time?

You quickly run into all sorts of complex technical issues.

To return to the sex gap question:

When careers opened up more to women in the early 1970s, it very quickly turned out that women were decent performers in all sorts of white collar jobs. Maybe not equally likely to be right edge of the bell curve performers at certain high end jobs, but businesses found that women did pretty good at a wide variety of careers.

Now, there’s a new paper based on a meta-analysis of studies covering 16 million people:

Paywall here.