It’s Only “Good Science” if the Message is Politically CorrectPublished 16 November, 2010
There’s another article on sex differences that appeared Sunday, published in the Guardian by Madeleine Bunting. The basic idea seems to be that there’s all this Bad Science – her term – that says that men and women are hardwired to be different, but now – yay! – there’s Good Science, which shows that men and women are both from Mars, rather than having separate metaphorical planetary origins.
I thoroughly recommend the article. It has a lot of my favorite mistakes in science reporting, including the following. The reason, she argues, that women might be better with language could either be “that this [difference] is linked to brain structure” or that “it has an evolutionary explanation,” an “or” which obviously confuses levels of explanation, a topic about which I’ve written sufficiently frequently that I won’t bore readers with a rehash.
But there’s something more interesting than that here, where she says that the explanation for this sex difference in language (which I suppose she must therefore acknowledge exists) could be due to “activity type.” She notes that “If that activity type is looking after small children or repairing drains, it will affect how they use language.” Basically this style of argument says that there is a sex difference, but that it has a particular developmental pathway in which language changes depending on what one is up to, which itself (systematically) depends on one’s sex. Maybe this is inconsistent with what some people think, but that sounds at least remotely plausible to me (and not even in principle inconsistent with the two explanations above). I think the real issue here is again “hardwiring.” Bunting thinks that evolutionary psychologists require that sex differences be “hardwired,” developing independent of all aspects of development. If development does matter in contributing to the sex differnece, she reasons, then this runs against the theory. Interesting.
My eye was caught by some quotes from Liz Spelke, who finds that infants don’t categorize by race, but that they are “predisposed” to categorize by gender. I am, of course, very sympathetic to this claim. But, Bunting writes that Spelke “thinks it’s possible that it served some adaptive purpose in our evolution, but that actually gender is a very bad indicator of behaviour because there is so much variability within each sex. For instance, if one man likes bananas, that is no reason to assume another does.” First of all, it seems to me that if one man likes bananas, that is actually a good reason to assume – or, at least, increase your estimate of the chance that – another does. It’s just not a reason to assume that men but not women like bananas. Food choices have all sorts of social stuff going on with them, so who eats what is actually good for inferences. But even so, this seems to be saying that if a dimension can’t capture all behavioral variation, you might as well ignore it. Infants and the old folks like mushed peas; should we ignore people’s age when trying to predict what they’ll get up to? That doesn’t seem right.
Anyway, the last paragraph begins with the remark: “Good science will challenge the tendency to stereotype.” It’s hard to know what, precisely, this is supposed to mean, but the point is that the article closes with some Good Science, the work on stereotype threat. I haven’t followed the literature on this very closely, but my reading of this was that while it was true that stereotype threat increased gaps in performance between groups, removing it did not eliminate it, suggesting that these gaps have other antecedents. (I have in mind Paul Sackett’s comments on this literature, though I want to admit that I have not tracked this closely, and I think those comments focused mostly on race.) Good Science, you see, is all about the political message.
Sackett, P. R., Hardison, C. M., and Cullen, M.J. (2004). On interpreting stereotype threat as accounting for Black-White differences on cognitive tests. American Psychologist, 59, 7-13