throwing out the individual differences baby with the group differences bathwater

Group-level differences in cognitive ability are in the news again, thanks to the quickly-notorious Google memo on the company’s diversity efforts. You can find a lot of stuff written about these differences from qualified people already and I’m not gonna add to the pile, other than to say that from where I’m sitting, if there are any sex differences in intelligence or math ability, they seem not to exist in early childhood and there are plausible cultural and social reasons that they would appear by high school. The science on personality trait differences seems less clear to me but then those constructs are also less concrete. In all of this I’m pretty much in keeping with the liberal mainstream.

But I do want to voice a caution, here, because there’s a natural but unfortunate tendency to make an unjustifiable corollary to arguments of these kind. Regular readers will know that I reject the idea of biological or genetic explanations for academic differences between races. Instead I follow most progressive people in thinking that the differences are socioeconomic and environmental in origin. There, too, I’ve often seen people make the same bad leaps: they tend to reject the idea of innate or genetic differences in individual academic ability or intelligence too. It’s not hard to understand why; talking about genetic differences in intelligence at all may seem like fruit from a poisoned tree, and why not just reject the whole idea altogether? But understanding the difference in group-level claims and individual-level claims is hugely important, both analytically and morally. It’s the difference between contributing to stereotypes that have contributed to marginalization and injustice of vulnerable groups, and accepting the reality that not all individual people are equally gifted in all areas.

And the data here is really, really clear: there are profound differences in individuals in academic or intellectual ability; these differences are generally quite durable over the course of one’s life, although of course there is some variability, as there is in any measurable psychological trait; and there is very strong evidence that a major portion of this difference comes from genetics. I don’t think that boys are smarter than girls or that black kids are less intelligent than white. I do think, and think both the empirical record and common sense shows, that not all people are equally talented in different intellectual domains, and that if you believe that the brain is the product of evolution, we should expect a significant amount of that difference to be genetic in origin, which is in fact what twin studies, adoption studies, and GWAS data show. I’ve written about all this in this space many times before.

You can think about this clearly if you just eliminate the comparison between groups that are supposedly different and look only at within-group distribution. So, for the purposes of this debate, look at women and their various metrics for intelligence and academic success, whether generally or in math/STEM/computer science. Forget about comparisons to men for a moment: within that group, on any properly validated intelligence metric, we find a normal distribution of ability. That is, there’s a mean, and there’s a distribution of about two thirds of the data points within a standard deviation from that mean, and about 95% of the data points within two standard deviations, and the distribution is just about symmetrical. Some women are better than other women on the SATs, IQ tests, quantitative reasoning tests, etc., and in predictable ways. The same exact condition applies when looking at distributions of black students, Asian students, students from Turkey, students who attend public schools, students who are left-handed, students who play Little League, etc. – real, persistent, and predictable differences of ability between individuals.

Now these individual differences don’t have much to tell us about diversity efforts like those at Google, which for the record I support, other than to say that Google is probably looking for those in the very top reaches of these distributions no matter what. But they say a hell of a lot about how we should approach education from a policy level. Policy has to reflect our empirical understandings of reality, and right now, ours doesn’t, as it is based on the false notion that all students can be brought to meet arbitrary performance standards, that there are no intrinsic limits to how well any individual student can perform, and that the purpose of schooling should train every student to be a Stanford-education Silicon Valley superstar. That’s the kind of cheery, optimistic, utterly-unachievable policy goal that comes from thinking that, because there aren’t genetic differences in intelligence between men and women or between races, there are no such differences between individual people either. That’s wrong and destructive and we can’t allow our necessary efforts to oppose bigotry to lead us in that direction.

For a lot of great thoughts on how to ethically consider genetic influences on individual intelligence, I recommend the work of the brilliant Paige Harden.


If you found value in this blog post, please consider supporting this project financially by clicking on the Patreon or Paypal link in the sidebar to the right.