Sunday, October 27, 2013

Why Michael Gove's department is confused about genes and education; and why you probably are too.



In the Guardian on Saturday October 12 it was reported that Michael Gove's special adviser Dominic Cummings had "provoked outrage" by claiming that "up to seventy percent of a child's performance is related to his or her genes".

Now it seems that Mr Cummings has been mis-quoted here. I happen know this because he told me himself in a tweet. His twitter name is @odysseanproject – which I suppose will amuse fans of Diary of a Nobody. Mr Pooter’s favourite joke about his friends Gowing and Cummings was that Gowing was always coming and Cummings was always going.

But I digress.

What the Guardian ought to have said and (it seems) Dominic Cummings did say is that "up to seventy percent of the variation in children's performances is related to their genes".

So why is that different from saying "up to seventy percent of a child's performance is related to his or her genes"? To see why, we need only consider the following simple thought experiment:

Imagine you adopted two randomly chosen children born one the same day (Mary and Jane perhaps) and gave them exactly the same upbringing, environment, life experiences, and education. (Of course that would be impossible in practice, but this is only a thought experiment.) Now imagine that we tested them both (several times perhaps to make sure one of them was not having an off day) at eighteen years old and Mary got straight Bs and Jane got straight Cs.

The variation in the results of the two individuals must, I hope you see, be entirely due to their respective genetic makeups.

Now let us repeat the thought experiment but provide much better education. This time (we could imagine) Mary gets straight As and Jane gets straight Bs. The variation in the results of the two individuals must still be one hundred percent due to their respective genetic makeups. The improvement in results is, however, entirely due to the change in environment - specifically the improvement in education.

This observation illustrates why Dominic Cummings's statement (as mis-reported) is drivel. The seventy percent figure relates to the explanation for the variation in a population not to the performance of an individual.

Asking about the relative contributions of genetics and environment to a particular child’s performance is a bit like asking what whether the height or the length of a rectangle contributes most to its area. Such a question makes no sense.

Asking about the relative contributions of genetics and environment to variation, on the other hand, makes perfect sense.

Some things in a human population may be vary a lot – like personal income. Other things in a human population may vary much less – like height – you do not find people who are two million meters tall for example.

The degree of variation in a population can actually be quantified. (This is quite complicated, and there are different ways of doing it, but let us just stick with the basic idea.) Once we have quantified the amount of variation, we can talk about what factors contribute most to that variation.

If, in the case of school children and academic performance, we took away one of the contributions to variation (which we could do in theory) by breeding a cohort of school kids who were all genetic clones (which would take away the variation due to genetics) or by giving a cohort of school kids exactly the same education (which would take away the variation due to quality of education); in either case, the amount of variation in the population would be reduced. It would obviously be reduced more if you took away whatever was making the biggest contribution.

If we pretend for the moment, and for the sake of simplicity, that education and genes are the only factors (of course there are many others such as social class, but let us keep things simple) what may seem slightly paradoxical is that if we gave all children exactly the same education, though this would reduce the variation in the population, it would increase the relative contribution of genetic factors - it must do so because all variation in a population receiving exactly the same education must be down to the genes.

Of course, as I expect almost everyone agrees (regardless of their politics) the variation in academic achievement (and many other attributes) of the population depends on a complex mixture of factors. Teasing out the relative contributions of the various factors is far more tricky than you might think. Even if we take something like height - which is far easier to measure objectively than academic ability and is undisputedly highly heritable (tall parents tend to have tall kids and vice versa) - it is still far from clear to what extent the variation in human height around the world is down to genes or environment.

I have no idea what the correct figure is for the genetic contribution to the variation in academic achievement in the population at large, but (though I am very much on the political left) it wouldn't surprise me at all to learn that the true figure is even higher than seventy percent.

But, given the fact that nobody knows the facts for sure, people at either end of the political spectrum are wont to provide ideologically-driven rather than data-driven answers to the empirical question: How much is nature and how much is nurture? Hence the irate tone of much of the discussion on this topic in the media this week.

The left's commitment to egalitarian principles lead them to conclude that it must be mostly due to nurture. Only if we believe that, they suppose, can we imagine a future where social inequities are put right through progressive social intervention.

The right's commitment to in-egalitarian principles lead them to conclude that it must be mostly due to nature. Only if we believe that, they suppose, can we justify the claim that doing anything to improve the lot of the hoi polloi is a waste of time.

So why do I claim that both sides get the whole thing rather back to front?

Let us conduct another couple of thought experiments:

First let us first suppose that we have the most extreme case possible of the frequently encountered left-wing belief about the way the world is. Everyone in our imaginary society is a genetic clone with an exactly equal genetic endowment of academic potential and any differences in ultimate achievement will be entirely due to how we nurture the individuals concerned. How would we then structure our education system? We should have to choose individuals completely arbitrarily from the pool and train some of them up to be clever enough to be surgeons or rocket scientists or whatever; and - at the other end - some of them to be just clever enough to tie their own shoe-laces so that they could perform jobs requiring very little intelligence - like the job of Education Secretary I suppose.

But isn't this more or less what right-wing education policy has always been (and what the likes of Michael Gove and Dominic Cummings seem to want to go fully back to): a system where people are picked arbitrarily from the pool on the basis of social class (rather than innate ability) and given the training they require to fulfil their allotted stations in life?

Now, instead of a society of genetic clones, let us imagine a society where everyone is born with different potentials. No matter how well I had been nurtured, I could never have become a Premiere League football player; and the likes of Michael Gove could, no matter how well he had been nurtured, never have understood averages or become a professor of thermodynamics.

...a bit like the world Dominic Cummings and other right-wingers (probably largely correctly) believe we do inhabit.

In this world, it no longer makes sense to choose people arbitrarily from the pool and nurture (only) them. The only policy that makes sense is to nurture everybody so that each person achieves the best he or she is capable of and those who come out on top represent those who started out with the best genes rather than those who were fortunate enough to be given an education.

...rather like the sort of education system left-wingers tend to argue for in fact.

Okay, I've over-simplified here and rather caricatured the various political positions, but I hope I have also successfully made a serious point: the thinking about nature and nurture, on both left and right, is often terribly confused.





A version of this post was included in the @pod_delusion podcast of 2013-10-17.