Friday, 19 September 2014

If achievement is genetic, how is Education different from Apartheid?

In the nature v nurture debate of child development, academic research is squeezing the role of nurture like never before. A recent study in Nature shows 2 year-olds inherit language learning ability; a study of over 5,000 twin pairs from 2013 estimated 58% of performance in English, Maths and Science GCSEs is accounted for by genes. Some have estimated (£) the genetic contribution to literacy levels as high as 77%. So far all the responses to this nature consensus I've seen, from David Didau and Andrew Sabisky (courtesy of Nobody's Blog) , is to say we can't change the fact some children outperform others and must instead focus on raising every child's attainment an equal amount. This is sometimes referred to as 'moving the curve' and is visualised here, where the y axis is number of pupils, the x axis academic attainment.

The problem with this rallying cry for the profession is that while superficially appearing to benefit everyone equally it ignores the fact that position relative to others is the most significant educational outcome. The opportunities that children at the back end of the curve are afforded by their education are the ones rejected by those with higher attainment. Moving the curve doesn't change this.

If a person's position on that curve is determined by their genes and therefore largely beyond their control then this is far from being an academic, in the sense of unimportant, debate. For if we assign people status in society based on their genetic make up how is our system morally distinguishable from Apartheid?

Under Apartheid you acquire benefits/deficits automatically; at school you have do something to get those benefits, so does that provide an adequate distinction? If we decided who was going to be rich and poor with running races would you justify the system by saying that nature's sprinters still have to turn up and run? No, the system is still clearly biased in their favour. In fact this difference makes our school system more unjust than racial division, because those at the top of the curve are rewarded for hard work by being told they're better than others, while those at the bottom are punished for it by being told they're worse. Natural differences in cognitive ability are magnified and metastasised by encouraging effort from the 'bright' and discouraging it from the 'dim'.

Is it crudely reductionist to say that Education's main purpose is to determine position in society? That's a charge that comes my way by those who seem to imply my attention to this aspect of our school system is my own peculiar fetish. A typical example came this week when Harry Webb tweeted 'Cadwalladered. So you're all about emphasising who comes top and bottom in things?' The only people who ridicule the idea of education as a marker of social status are those who were successful at it, just like aristocrats who dismiss the poor as vulgar for their preoccupation with money. If you come bottom in terms of academic attainment then the key takeaway from your education is that your work is worth less than that of others and that, by extension, you are an inferior human being. That's a horrible experience for a person to suffer.

So the situation is unfair; isn't it God's unfairness not man's? No, for there is nothing fixed or inviolate about a curriculum that only includes academic subjects and so ranks children by academic skill. Even where we stray from this narrow path into areas like Art and Drama we insist they involve significant amounts of writing about Art and Drama. Of the four aims of the new Art KS3 curriculum twoare related to producing art, the other two to analysing work and itshistory. Our insistence that all success is mediated through one type of cognitive ability we know to be largely inherited is perverse. School is where children learn, above all, their place in society. By choosing a solely academic curriculum we choose to make that society an unequal one. We could just as easily adopt a curriculum for equality.

Could schools really make such a powerful switch? Aren't they responding to the job market as an impartial adjudicator of what different types of work are worth? Markets are not impartial, they are an aggregation of people's beliefs and choices. The worth the 'market' assigns to a particular type of work depends on what people in that market have been taught to value. The same applies to the way people appraise their own worth to society; people accept low pay because they've been taught they are less valuable than everyone else. Educators can't wash their hands and say this a problem with society that is not their responsibility to fix. The belief that a person's value depends on their cognitive ability is a direct result of what we teach children at school.

Apartheid was a man made system designed to give people with a certain genetic pedigree privilege over others. Education distributes privilege along a gradient rather than dividing society into black and white but where you fall on that gradient primarily depends on your genes, not your efforts. Someone born with low cognitive ability cannot succeed in our schools any more than a black person can make themselves white. The victims of this injustice are those least able to analyse its causes and express their discontent, so the well educated can bandy terms like 'meritocracy' around without being challenged to say why being born clever makes you anymore deserving than being born with pale skin. If we care about social justice we have to change school; punishing people for who they are is wrong.


  1. My only response is this; good luck with a "curriculum for equality" that produces people in any way productive in a global economy. It's fashionable to sneer at such things, but while producing wealth is certainly not the primary goal of education, it matters. The state's investment requires a return, otherwise the size of that investment will inevitably shrink and disappear (because the state doesn't have any money to spend due to an unproductive citizenry).

    Another problem is the all-encompassing nature of the g factor that underpins IQ. All competence in all domains of life is in some way dependent upon g. Life is an intelligence test. Abilities that one would not necessarily think relate to IQ in fact do so; pitch discrimination, for instance, a key component of musical ability, correlates with IQ at .4. A recent special issue of the journal Intelligence, focusing on talent and expertise, showed that musical prodigies invariably have high IQs. I have never seen any studies on the issue, but I would be exceptionally unsurprised if the same was true for acting talent. It is very difficult to find a domain in which the brighter people, the genetically gifted, do not do better than the less bright (excluding, perhaps, sport).

    Now, you could of course change this. Early identification of the genetically gifted via either behavioural tests or genotyping could allow for a sustained program of punishment designed to ensure cognitive equality. We could make sure that the gifted were deprived of books, were regularly beaten, and were largely kept in a perpetual state of malnourishment. They would be given inadequate, inferior educations that perhaps terminated at a young age, and would be cut off from the internet and public libraries. I have no doubt that by such a program genuine equality could be achieved and the injustices you lament remedied. It really would work, unlike compensatory education for the less gifted. Inequality of outcome based on inequality of genotypes would be a thing of the past. I do not think such a program desirable, but you may of course differ. That is your prerogative.

    1. If a curriculum that developed skills beyond the academic would harm our economy, why has the CBI called for 'a definition of achievement [that] goes far beyond exam scores'? The educational diet of low achieving students is simplified academic tasks combined with lots of praise for the slightest effort. That's a terrible preparation for the world of work and a missed opportunity when they could be developing useful, practical skills.

      Regarding your second, nightmareish, point you misunderstand what I mean by equality, confusing equally valuable with equally capable. Take any two teachers, one will be better at teaching than the other, but that doesn't mean he or she should be paid more. In the context of school you're right that whatever the curriculum entails, some children will be better at it than others. Just because Andy is better at academic subjects than Paul, it doesn't mean that Paul is better at practical subjects than Andy. That's true, but unimportant. What's important is that we don't take those differences in ability and mark them with signs of official favour. When we train children at sports we encourage them to be the best they can, while remembering the success of their team and the way the game is played are more important than their individual performance. Training them as citizens should be no different.

      I understand that it's difficult to separate cognitive ability from social status. All of us are raised in a system where to have one is to have the other. But that status is conferred by the actions of those in authority not the innate qualities of citizens. The advantage of conferring signs of status on children by grading their work is that the most able are encouraged, but that's clearly outweighed by the disadvantages which are that everyone else is discouraged and divisions are sown between the top and the bottom.

  2. Interesting take, Andrew. One thing I'd note is that many allege that the achievement gap isn't based on true meritocracy based on allegations that some people on the lower end not being able to maximise their genetic potential rather than not having much of a genetic potential to begin with. Some even claim things like racial discrimination to be at fault. They also allege that heritability of IQ & achievement is moderated by SES. I disagree with all the above arguments, but don't have much in the way of counterevidence. How would you reply to suck points, Andrew?

    1. Also, an argument someone made against ability meritocracy when I presented them this source arguing for it:

      I quote the counterargument below. I don't know how to counter it myself, how would you do it, Andrew?

      "Anyway this looks like same old to me since GWAS doesnty show causation onlñy correlation between genes and traits it is extremely vulnerable to environmental feedback effects. Also his conclusion that equality of opportunity will result in higher heritability has a giant logical hole in it because he doesnt consider that fors of discriminbation that also follow family lines (such as ethnic or racial discrimination) will have the same result of making heritability appear higher, so that heritability can just as well be considered an index of e.g. phenotypically motivated discrimination. It also ignores the often observed fact that heritability varies between high income and low income families - with high heritability in the former and low in the latter. This is essentially due to the same effect that Plomin considers, namely that increase equality of options lead to higher heritability. But it also shows that the Bell curve argument that SES differences is a result of IQ meritocracy is incorrect, because LOW SES families have less opportunities to develop their potential. Plomin does not even mention this aspect of the argument which is frequently mentioned in the literature, by e.g. Flynn, Turkheimer, and Mackintosh."

    2. this is what I call the Turkheimer effect (I think others call it something else). The idea that IQ heritability is moderated by SES has replicated rather better in the US, but has not faired nearly so well in Europe. Plomin's group, with their vast sample of 5k twin pairs, found no moderation of heritability by SES (see Hanscombe et al. 2012

      By the way, the original Turkheimer study showing this effect was pretty low quality. The data were old (from the 1950s as I recall), so there's no reason to think the results still hold today given the large social changes in the US since then. More importantly, it was underpowered for the effect it was trying to detect. It takes a lot more statistical power to detect GxE interactions than to detect G+E main effects, and consequently if an underpowered study does find such effects it's quite likely that it's just a false positive. Really, I wish people would just stop quoting this particular study just because it shows what they want it to show, given it's dubious quality. I am told there is a meta-analysis on this coming out shortly, so we'll see what the aggregated results are.

      (even weirder, in the Turkheimer sample there's a SES-IQ heritability interaction, but not a race-heritability IQ interaction. Go figure. In general heritabilities for blacks and whites are fairly similar;

      NB: I do think heritability of IQ moderation by SES probably still does exist in the UK today, but only in the tiny percentage of the very worst households (from an environmental standpoint). Obviously families in the bottom 1% of the SES distribution are unlikely to be included in twin studies, so no conclusions can be drawn, but I'd be surprised if being in this bottom percentile did not lead to lower heritability of IQ.

      This is not a full response - more later.

    3. Great post, Andrew, but I have few queries. Would you have on hand any published mainstream stuff vis-à-vis your criticisms of Turkheimer, something to cite? Also, do you have any stuff from a mainstream journal that replicates John Fuerst's findings in his paper? Other than this, good post. I await your next telegram.

    4. Also what I find problematic about race and intelligence topic is this paper:
      It seems to be mainstream in the field and I can't find much criticism of it.

    5. Also to add, the person referred to this book as well:

      So I wonder what you'd have to say for that.

  3. as to your correspondent's other points, I am unsure what to make of them, as they are poorly expressed. It is just a truism that more equality of opportunity leads to higher heritability, since equalized environments mean that a greater percentage of the variance MUST be genetic in origin. It cannot be otherwise. Now, flip it around, and you can respectably argue that high heritability does not imply equality of opportunity, but to do so you have to argue that heritability is being systematically overestimated somehow. People do sometimes to try to so argue, but generally their arguments are less than convincing, and in the modern era genome-wide complex trait analysis (GCTA) is validating many twin study results without any assumptions about the equality of MZ-DZ twin pair environments (a point I should have made at researchED).

    I am not going to argue about race and intelligence in public.

    Re the Turkheimer study, the Hanscombe et al paper linked to above talks about the large statistical power (they do a formal power analysis) needed to detect GxE interactions of this kind. Doug Wahlsten is a (largely misguided IMO) critic of behaviour genetics, but in a 1990 paper he made the good point that you need way more power to detect GxE than to detect main effects (“Insensitivity of the Analysis of Variance to Heredity–
    Environment Interaction,” Behavioral and Brain Sciences 13: 109–120.) A point not made in that paper is that with underpowered samples you get both false negatives and more false positives (see Kate Button's Nature Reviews Neuroscience paper). NB I am not critiquing Turkheimer personally - all data has its limitations, that is not his fault, and he is not responsible for other people citing one finding from an underpowered study using old data as established fact that applies everywhere at all times.

    More later.

    1. Your first point makes sense to me, and I would kindly request if you have any examples of the GCTAs you mention, and perhaps the arguments made by opponents. And I don't quite understand why you wouldn't want to discuss the tightly intertwined topic of race and intelligence seeing as you have already linked a paper from an online journal specializing in the topic which I regularly read myself. Wouldn't racial confound affect the SES-IQ correlations? For third point, do you have any papers from mainstream journals which explicitly make the point of Turkheimer's study being too weak and underpowered to draw any proper conclusions from it? I would like to use it as a citation on Wikipedia's articles, which still use Turkheimer, 2003 since they don't have any papers that criticize it. Anyway, looking forward to your next telegram.

    2. Basically, if your points are corroborated in the mainstream academic discourse, then I'd love to see it and cite it, otherwise others don't have much reason to accept the arguments in relation to amendments of a popular online encyclopedia.

    3. Environment probably does impact IQ to some degree as adoption studies have shown around 12 point IQ bump for children who move from low to high SES homes (Duyme, Dumaret, and Tomkiewicz (1999)).

      It's also unclear in the case of the correlation between high IQ and success whether the cause is the high intelligence or the confidence that comes from being raised to believe being clever makes you socially superior.

      However, solving this mystery is not important. If people with high IQ succeed because of their intelligence, teaching them they are superior is unnecessary. If confidence is the cause then it's unjustified.

    4. I only have so much time and don't really have the inclination to trawl through the literature to find if someone has made the same points re that study that I have. Wikipedia will sort itself out in due course anyway, and there's no need for Turkheimer's study to be explicitly criticized at all anyway; a point that far larger studies at different times and places have not universally replicated the effect (though some have), with citations to the appropriate literature, will do. That will get the key message across that Turkheimer's study is just one data point amongst many, and the story of the aggregated data is complicated at best.

      For GCTA, just google the acronym plus any cognitive trait you're interested in. Plomin cites a bunch of GCTA studies here and Peter Visscher's lab has a list of their papers here;

      The SES and racial differences in intelligence are not quite as difficult to distinguish as you may think, as controlling for SES generally only decreases the B/W IQ difference by about 5 points (and this "control" is obviously pretty meaningless anyway as any discussion of the famous "sociologist's fallacy" will quickly show).

    5. Yes, my point indeed is that race differences in IQ seem to be more innate/genetic than do SES differences. But how would one tackle papers like Nisbett, Flynn, Turkheimer et al.2012 which argue that group differences are environmental in origin?

    6. Look. I am not going to hand-hold you through a literature that goes back to Jensen's 1969 HER article and much further, with its roots in the early decades of the 20th century. Google exists. Use it. Nearly all Arthur Jensen's papers are freely available online, as are those of JP Rushton. Or you could email Heiner Rindermann or Gerhard Meisenberg. This is not difficult. I have better things to do with my life than you talk you through it all when the vast majority of the relevant literature is freely available and that which is not is not difficult to get hold of. Quite frankly, at this point I doubt your good faith - this is all starting to feel like a sting operation to get me to say the wrong thing. If you are genuinely interested in this question, google is your friend.

      Anyway, there is actually a much stronger argument for genetic SES differences (Herrnstein's syllogism) than race differences, because we can see the mechanism for the former, but not the latter.