Popular Posts

Caveat Emptor

The opinions expressed on this page are mine alone. Any similarities to the views of my employer are completely coincidental.

Thursday, 12 December 2013

Plomin on the radio

I caught part of Bob Plomin's interview on the Today programme this morning. I guess he is flavour of the month because of the Dominic Cummings kerfuffle. What I heard  - he was talking about genetics and variation in exam performance - didn't sound particularly controversial - though I should emphasize I didn't hear the end of  the interview, so who knows where it went.

Of course simply putting the words genetics and education in the same sentence is likely to lead to a knee jerk explosion of indignation that precludes almost all rational discussion. Though I'm against anything that simply brackets off a subject area from rational investigation I do in fact have some sympathy for the view that the faux naive: "I'm just a scientist telling it like it is, it's for the politicians to decide what to do with it" can be a thin veil for something more sinister. Jensen, Herrnstein, Murray & co were, and in the latter case, are,  not just scientists. They also had and have policy preferences they want to push.

As far as I can tell this is not true of Plomin and it seems to me that he is a much more serious scholar. I first came across his work by chance, just before he moved to the UK.  An American colleague,  a man of impeccable liberal credentials, knew  and spoke highly of him. On the debit  side you could put down the fact that Plomin was a cosignatory of the infamous Mainstream Science of Intelligence editorial in the Wall Street  Journal. Among the list of 52 signatories are a few crazies that I'd be wary of having my name linked with but guilt by association is not guilt.

I read some of Plomin's early papers but I have to confess that I'm not tooled up to follow the details of his most recent work, even those pieces aimed at a more psychological audience. To understand these papers you have to have a greater understanding of genetics than I do as well as an easy familiarity with the latest techniques for drawing statistical inferences from gene sequence data. I strongly suspect that one has to read these very closely to fully appreciate the message and it would be foolish to jump to conclusions without understanding that the devils, if there are any, will undoubtedly be in the details.

So I'm forced back on my meagre stock of knowledge, mostly gleaned through reading the polemics generated by past rounds of the genetics/IQ controversy. Still, I think you can get some of the way  towards making intelligent judgements with a little learning. So here are a few banalities to keep in mind:

1) Heritability is a population level concept ie it is about variability (differences around the average) in the population. 

2) The more you make environments similar, the greater the proportion of  trait variability that will be accounted for by genotypic variability. If, hypothetically, the environment is a constant, then the only possible sources of variation are chance (random "errors") and genes. This might be seen as a paradox for the left.

3) But, we might not give a damn about heritability. Height is a highly heritable trait, but in some populations average height has increased to an astonishing degree in the post-war era. Why? Probably better nutrition more evenly distributed throughout the population ie environmental improvement (and equalization).

4) Additive decompositions of phenotypic variance are too simplistic. It would be truly amazing if some geneotypes always produced better performing (OK I know there is a lot of baggage in that expression) phenotypes in all environments. In fact we know they don't, the literature on agricultural experiments is littered with examples of this. So when somebody tells you that they have decomposed population trait variance into genetic and environmental components remember to ask them about the interaction term (and whether it has just been assumed away?).

5) Genetic/environmental interactions are important because human individuals, unlike plants, choose, to some extent, the environments they live in and again, to some extent, make those choices on the basis of expected outcomes. People that hate reading (or never learned to love reading) are more likely to choose to live in houses without books, so whatever genetic basis reading ability has is not going to reveal itself in that sort of environment.

NB I'm not saying that Plomin ignores or doesn't appreciate any or all of these points. I am saying that other social scientists should appreciate them before they unleash their knee-jerk reactions.

No comments: