Tuesday, January 18, 2005
Why There Are So Few Women in the Hard Sciences: Part II
Time to dive in deeper. Lawrence Summmers' little speech suggested three different explanations for the scarcity of women in hard sciences: the eighty-hour weeks one is assumed to work, possible biological differences in talent between the sexes and discrimination against women. I wish to look at each of these, and I'm going to begin with Summers' second hypothesis, the one about biological differences between the sexes. The other two questions will be addressed in my next post if I ever get there.
It is useful to note at the very beginning that all these causes are likely to be extremely intertwined. For example, if people like Dr. Summers believe that women are less able to do hard sciences, and if people like Dr. Summers happen to become the presidents of major Ivy League universities, well, then there is quite likely to be discrimination against women in these places even if there was no innate sex difference in scientific talent. Likewise, if many employers accept the idea that women shy away from an eighty-hour week then they are going to be less likely to look at female applicants for such jobs seriously which can cause discrimination. And so on.
Also, it is interesting to ponder on the reason why hard sciences are called hard. What is it that is hard about them? Most seem to think that this label serves to distinguish them from soft sciences, but then, once again, what is so soft about some sciences? In reality, the hard sciences are in many ways easier than the soft sciences (which tend to deal with human data) because the empirical evidence has so much less static in it. It might be interesting to ask whether this hardness has anything to do with the idea of hardness in penis comparisons. I have no opinion on this!
Ok. Onwards and upwards. Are women less talented in hard sciences for purely biological reasons? The correct answer to this question is that we don't know. I can imagine many readers turning red here and starting to spew liquid on the keyboard, but bear with me for a while.
How would we find about such innate differences? The obvious answer would be by using genetic study for it, but as far as I know genetic study is currently not in a position to answer such complicated questions. If anything, we are learning that the interplay between inherited tendencies and the environment is much more complicated than we previously thought. It is even possible that the environment turns genes on and off, and in that sense there might not even be any such thing as purely genetic influence.
We don't have data on purely genetic influences in this field, or at least we don't have data that could be guaranteed to contain nothing else but the influence of our genes. What we have instead are three types of evidence which have been widely used to argue that differences in the numbers of women and men in hard sciences is biologically based: studies of differences in cognition between groups of girls and boys or women and men, studies which analyze the impact of some known change in the fetal stage or the impact of some known medical condition (such as autism) on cognition, and teleological studies which really use these same data but pedal backwards from them to various interesting stories about the division of labor between prehistoric housewives and brave map-reading warrior-hunters. I'm not going to say anything about the third group because discussing pseudoscience doesn't add anything (except hilarity) to what I'm trying to convey here.
The first type of evidence is the most important one of the three, because it uses large samples of relatively randomly picked individuals from all types of societies. The problem it has is that the measures of cognition elicited by asking children, teenagers and adults are unavoidably not going to be pure measures of genetic differences. The environment and the general culture have had time to work on the study subjects beforehand, and factors such as the quality of schooling the person has had, the family income and the general societal norms all can be shown to influence the findings.
Keeping this in mind, it's possible to note that most cognitive tests show some average gender differences in mathematics. Boys are, on average, better at certain types of mathematics problems than girls, especially in word problems, and this is true from a very early age and across various cultures. As Virginia Valian points out in her book Why So Slow, this might paradoxically be caused by the girls understanding words better in communication: most word problems require the solver to decide which parts of the statement are important for the solution, which parts are not and whether there is enough information for the solution. In the actual understanding of speech all parts have a function, even if some of the functions are half-hidden. Not knowing this may make a solver better at using the words mathematically.
Such gender differences are unlikely to be purely biological. This is because the differences between children in different countries are far greater and because the gender differences within countries have been declining for some time. But some part of these differences could be purely biological. What their significance for the hard sciences participation rates might be is more complicated. Consider the one test which shows the greatest sex difference of all: the mental rotation of three-dimensional figures, in which boys outperform girls pretty much everywhere. Most studies suggest that ability to do well on this test and other mathematical ability are not correlated. Thus, it's not possible to explain women's scarcity in mathematics in general by using this single test as the explanation.
The tests we develop and apply in cognitive studies are not necessarily neutrally selected. Every researcher has a gender identitity, after all. This is important to remember. For example, there are tests in which girls outperform boys by the same large and consistent margins as boys outperform girls in the mental rotation of three-dimensional figures, such as the one on perceptual speed, but these tests are not studied as intensively as the more familiar ones are. Even more generally, the tests we use are spot measures, may not reflect all important skills and tend to be overapplied to the young and then ignored for the rest of the individuals' lives. There is evidence suggesting that women have different lifetime patterns of mathematical abilities from men, and very little research exists on this and its possible meaning. For example, women outperform men in old age.
Many who argue for the biological explanation for the sex-skew in hard science practitioners say that what is really important to analyze is not the average scores on all these tests but the proportion of boys and girls who score exceptionally well. Because these tests have greater variability for boys, there are many more boys than girls with exceptionally high scores, and this fact alone could explain why most mathematical geniuses appear to have been men.
This is an interesting argument. Consider a different test, that of writing skills. Girls outperform boys on this test, and the variability is much greater for girls (see Dianne Hales: Just Like A Woman for references on the evidence). This means that there are many more girls in the high-scoring part of the test and therefore we should expect the geniuses of great literature to be overwhelmingly of the female denomination.
The reality is very different. Either we have been excellent in compensating boys for their "innate" deficiencies in writing talents or something else has affected the outcome, too. Or perhaps we should stop taking all this so terribly seriously.
A different avenue towards trying to find innate sex differences in mathematical ability uses subjects who are known to differ from the average in some pre-birth induced way. One such case is the genetic disorder of congenital adrenal hyperplasia (CAH) which leads to an overproduction of androgens in the fetus' adrenal glands. Both boys and girls with this condition have very high levels of circulating androgens, and girls with CAH show many forms of behavior which are regarded as traditionally male (boys are harder to diagnose so there is less data on them).
Girls in CAH score very high on the mental rotation of three-dimensional figures, as might be surmised, but they don't score any better in general quantitative tests than other girls.
Simon Baron-Cohen has proposed a similar approach to testing gender differences in cognition by using children with autism as the study sample. He believes that autism is a form of an "extreme male brain" and that the greater mathematical abilities of many autistic children are evidence of the biological nature of gender differences in mathematics. The problems with his arguments become evident if you read his book on the topic (The Essential Difference: Men, Women and the Extreme Male Brain), which I have done. He presents no new evidence for his arguments and he even goes on for some pages about the hypothetical "extreme female brain" which doesn't exist, but if it did exist, Baron-Cohen believes that it would be favored over the "extreme male brain". Which sort of shows where he biases lie.
And if you weren't convinced about them yet, you could always take the very objective test in the Appendix of his book which will tell you how biologically male your brain is by answers to questions such as:"Can you fix your own electrical problems?" "When you look at a piece of furniture, do you wonder about how the joins were made?" (What about when you look at a beautiful dress?) "Do you like to chitchat more than you like to collect coins?" Or you could take it here.
Ok. Enough for one post. I'm not an expert in this area, though I'm a sort of a Renaissance goddess, so the usual caveats apply.