Some people will justify everything

A week ago, a transsexual neurobiologist named Ben Barres made the standard case against the biological view of gender differences in mathematical/scientific ability, adding some personal anecdotes about how people started treating him more seriously after his gender reassignment surgery.

Apart from the personal touch, the bits quoted in the Pandagon link above (Barres' article is behind a subscription wall) are nothing I didn't see on Pharyngula during the Larry Summers hoopla. I know the standard reply to them, I know my attack on the standard reply, I know the criticism of my attack, and I know the answer to the criticism.

However, Gene Expressions Classic's take on the situation isn't even up to the standard reply's quality. Agnostic hashes up a large collection of facts pointing out that men and women do differently on various tests, and that IQ is inheritable. Nowhere does he produce evidence that the differences are in fact inheritable - indeed a comprehensive psychological study shows that in children's early years, differences in cognition are minor.

There is, of course, the argument that small differences in mean or in standard deviation lead to large gaps at the higher end. If A and B are populations of equal size whose standard deviations are equal and whose means differ by half a standard deviation (say A has the higher mean), then 4.1 as many members of A as of B will be at least two and a half standard deviations above the mean. The same applies if A and B have equal means but A has a standard deviation 2/9ths higher than B's.

But looking at actual data of women in science disproves either of these propositions. Spatial perception and mathematical aptitute are not equally important to all sciences; the ranking is roughly math > physics >> chemistry > biology. But at Stanford, women's representation goes biology > physics >> chemistry > math; at Columbia, it's biology >> math = physics > chemistry. Furthermore, at both institutions the overall percentage is higher than the national percentage, even though either a greater mean or a greater standard deviation will make the difference the greatest at the highest level.

I don't know if the single indication of gender bias in the academia that GNXP attacks is the only one in Barres' article, but it's certainly not the only one that is known. In 2001 Stanford commissioned a comprehensive study about gender disparity within the university, finding among other things disparity in pay, attrition of young female professors due to a variety of factors, and even discrimination in lab size in engineering (though not in science). There are studies showing that a female name on the byline is enough to cause professors to devalue a scholarly paper.

And finally, comparing different countries' level of female representation in science and engineering reveals little connection to quality. The countries with the best universities are not those that have the least female scientific fields, except the elephant in the room that is the United States. However, there is very strong correlation between discrimination in the public sector and among the middle and upper classes, and female representation.

So in that light, Agnostic's claim that it is biology and not discrimination that causes women to be vastly underrepresented in science seems like a complete non sequitur. Even if that the studies proving large personality differences between men and women are not flawed - and that's a pretty big if - these differences are entirely attributable to social conditioning. There's certainly no evidence that biology is in any way involved, given that cognitive science has produced no evidence of a socially meaningful innate mental difference except women's better language perception.