The genes and geography paper is out

Our paper "Genetic correlates of social stratification in Great Britain" is out in Nature Human Behaviour. I and Abdel Abdellaoui, the lead author, have written a non-technical piece about it in The Conversation, and here's another article from Abdel. The short version: we look at genetic differences across different areas of the UK, in particular rich and poor regions, and find that they are increasing. We use polygenic scores – DNA-based estimates of a person’s characteristics, like their education level, height, personality or proneness to disease. Over the past half-century, people with high polygenic scores for educational attainment have been leaving poor areas and moving to richer ones, making Britain more unequal. Here, I just want to give some more personal reflections on the paper.

First, I’m very happy with this work. Though I only contributed a small part to a much larger body of evidence pulled together by Abdel, I still feel proud to be part of this project. I am mostly a sceptic about “scientism” in the social sciences – the idea that we can discover timeless truths about society, the way the hard sciences do for the natural world. But there is genuine science to be done, and in my view this is an example. Much is still unclear about the import of these results, and there are still many interesting questions to explore, but the paper proves its core claims beyond reasonable doubt.

Second, it’s been an interesting process. The paper’s topic – genetic differences across different regions – is controversial. Early on, I predicted that people would either see our results as obvious, or as unconscionable. The preprint generated a lot of tweets. When I put an early chart from the paper on Facebook, one person accused me of “shitting out” a random result and unfriended me. I was also accused of being part of the Bremainer elite. Heh. Needless to say, this paper makes no claims about the rights and wrongs of Brexit. I expect more of this in future (and, to people outraged by our findings: would you prefer us to change the data?)

Then, I am still very uncertain about the practical implications of the results. Partly this is because some scientific uncertainties still remain. Our Conversation article glosses over the fact that geographic clustering of genetic variation raises awkward technical issues for geneticists. DNA data does not magically rule out confounding variables. If people with high polygenic scores for educational attainment live close to each other, how much of the “effect” of the score is really due to the environment they live in? Within-family studies, which give true causal estimates of these effects, due to the “lottery of meiosis” – the fact that siblings randomly get different amounts of Mummy’s and Daddy’s DNA and hence different polygenic scores – limit this concern. But it still needs to be investigated further.

But even if all the scientific uncertainty were resolved, I’d still be unsure how to respond to our findings. Our Conversation article gets at one point: genetic differences exist between regional populations, and these are likely to explain some of the regions’ economic performance, and maybe also other outcomes such as health outcomes – though this is still unclear.

What should we do about this, though? How serious is it? H. G. Wells’ Time Machine is a nightmare vision of the future, where the human race separates into privileged, childlike Elois and the underground Morlocks, descendants of the working class, who feed on them.

London 2019: Angry Morlocks savage an isolated Eloi. Source

Should we be that worried? Eh, probably not. We still know so little about the pathways from DNA to complex social outcomes like education. But should we be somewhat worried? Maybe.

One point is quite simple. We show differences between rich and poor area averages for many polygenic scores. For a generation or more, geneticists fought to persuade social scientists that individual differences in outcomes may reflect genetics rather than the environment. I’d say that battle is mostly won. With these results, we can start having the same debate about geographic differences. Commentators and policy reports often point to regional inequalities as evidence of social injustice, or of institutional failings in the worse-off places. (Here is a recent report from the Coalfields Regeneration Trust, about the very areas our article focuses on; here's The Mirror blaming Thatcher, though in fact, these areas were declining from the 1920s onwards....) Now, another potential explanation is available.

That does not mean there is nothing to do about these inequalities – as anyone who wears spectacles should know, genetic does not mean unalterable. But it does challenge the presumption that all geographic inequalities must be down to social or institutional differences. It ain’t necessarily so.