Heritability of academic prowess

Interesting work by Krapohl, et al. in PNAS. I love twin studies. They look comprehensively on the heritability of various factors that contribute to academic achievement, particularly performance on the British* GCSE exams administered during high school.

They report that scores are 62% heritable, and that the different traits are from 35 - 58% heritable, with intelligence being the 58% figure. Their other traits were "personality", self-efficacy, well-being, etc. We'll have to look at the paper itself to see what they mean by personality, and what specific traits do the work. I'm guessing they mean Big Five personality, and that conscientiousness will be one of the major factors. (Openness to Experience often shows up as an important Big Five trait for academic achievement, but it's mislabeled – on its face, it's more about urban sophisitication, intellectualism, or urban liberalism than openness at a pure psychological level of analysis. We'll have data on this soon.)

What does it mean for something to be "X% heritable"? I think this should be spelled out more clearly in news reporting of heritability studies, because I don't think the public will generally know what such statements represent. It is almost certain to mean that X% of the variance in Y (in this case, academic achievement/test scores) is accounted for by genes, by one's genetic inheritance. This is commonly expressed as R-squared in linear regression. 35 - 58% are huge figures for percent of total variance explained. We almost never see such large figures. To get the correlation, r , just take the square root of .35, .58, etc. The square root of .58 is .76. That's a huge correlation.

What does it mean to say that X% of the variance in one variable is explained by variance in the another variable – what does it really mean, down to its bones? That's complicated, and modern language – English or any other – is not well-suited for describing the nature of such statistics in a sentence or two. Something that gets lost how we express such statistics is the reality of exceptions, and the number of such exceptions. Also, people often convert correlations to probabilities, which is completely wrong. That not what correlations are. A .76 correlation between X and Y does not imply – at all – that if a person is high on X, they're likely to be high on Y.

And in this case, the particulars of which genes/alleles do the work, or which sets and combinations, will probably matter a great deal. They'll have different implications for the heritability of these outcomes and traits. We don't know all the details yet. I think the intersection of genetics and environment is extremely interesting, as is the intersection between genetics, environment, and willpower/self-regulation (which is itself partly heritable...)

And the intersection between genetics, environment, willpower, and ideas. What does that mean? Ideas will be folded under "environment" for most purposes by most researchers. People are exposed to radically different ideas in different settings, eras, and cultures. We know almost nothing about the force of ideas. It has not received nearly as much research attention as things like income, race, genetics, etc., probably because those things are easier to measure. What's the impact of repeated exposure to the Reverend Dr. Martin Luther King Jr.'s I Have a Dream speech? Or Obama's speeches? Or Oprah, Feynman, Rand, Cyrano de Bergerac, The Shawshank Redemption,  etc.? How sticky are they? How much are they remembered and used by those exposed to them? I suspect the ideas we're chronically exposed to by our parents and teachers will have more impact – the implicit and sometimes explicit beliefs about the world, the feasibility of success or happiness (which don't have to be the same), etc. But who knows – the ideas we get from books and movies might have a large impact.

We capture some ideas in the form of worldview measures like belief in a just world, external vs internal locus of control, belief in free will vs. determinism (see Vohs & Schooler), dispositional optimism, or just cultural psychology. I think we've just started to explore that whole area, even if some of the measures have been around for a while, and I don't think we've done much from a developmental standpoint to track exposure to and assimilation of certain ideas throughout childhood (I've always envied developmental psychologists, and I think that social psychology could probably make a quantum leap if we all stopped what we were doing and dived into developmental psychology for two years, read everything we could, then reapproached our research questions.)

This has always lurked in the corner as a weakness in how we approach genetics and evironment. We study, necessarily, the status quo, the world as it is, the world as we find it. Then we make inferences about the role of genetics and environment in X, Y, or Z. We declare that academic achievement is "58% heritable" and people will run around saying that as though it's a stable, fixed truth, like the acceleration of gravity on earth (even though the researchers don't necessarily think of it that way.) But what if the environment were different? What sorts of things about the environment could be fundamentally different than the environment(s) that existed for the participants at the time and place the researchers conducted their study? Ideas could definitely be different. And some ideas might interact with the phenotypes at issue. And some important ideas might not even exist yet. I think that will be a very interesting area to explore in the coming decades. I don't mean to say that particular ideas and values will definitely have a huge impact on the outcomes we normally focus on – reality is whatever it is, and I don't know what it is. This is what science is for – to find out. This is potentially very rich – ideas vary in structure, content, source, and many other ways. Exposure to and embrace of ideas can vary in many similar ways.

A slightly related development: For decades, we thought Mischel's marshmallow studies represented a finding where children with more self-regulation, or self-regulatory capacity, fare better as adults – and that we could capture this when they were very young, like 4-5 years old. And the basic finding might still be true. But now we have innovative research that suggests that the credibility of adults might have been a factor. When an adult tells a child that if they refrain from eating a treat, they will get a bigger reward in the future, whether they believe the adult should matter. Some children might be more exposed to less credible adults – they might be lied to more. They might not take for granted that some dude in a lab coat is telling the truth. In that case, it would be a better strategy to just eat the treat in front of them while they can, and not pin their hopes on an adult being trustworthy. If that drove the Mischel results, and if such children were more likely to be low-SES, than it wouldn't necessarily be a story about self-regulatory capacity (though it still could be, in a couple of ways.) It could just be that a low-SES childhood predicts a low-SES adulthood, and that low-SES children were less likely to trust adults' promises, given their experience. (Thankfully, last time I checked, parents' income accounts for only 25% of the variance in childrens' adult income in the US, so the idea that environment is destiny is simply false. It's usually lower for daughters, which could be explained by cultural changes.)

It's a fascinating issue, and it also touches on the need for more diversity in social psychology (and all branches of research psychology.) The idea that the children from low-SES homes might not trust adults as much might not occur to a lot of researchers who never grew up in such settings. We need more Mexicans, more blacks, more white people who grew up poor, definitely more Asians. And like we argue in our BBS paper, more intellectual diversity, more people who depart from the prevailing white liberal narrative. The best argument for biodemographic diversity is that it leads to greater diversity in ideas and assumptions about human behavior and the factors that drive it. I for one don't care about racial diversity for esthetic or reparative justice reasons – I care about it because it gives us a broader set of intuitions and experience, and ultimately better and longer-lasting research.

How is this related to my starting point, the genetics and academic achievement study? Well, it would be a very Joe move for it to not be related at all. I start in one place and end up somewhere else all the time. But it's related here in that what many people took to be a proxy measure for a heritable trait – self-regulatory capacity at age 4 or 5 – might have actually been a measure of an environmental variable. It might have been about what the children had learned about the trustworthiness of adults (and learning is environment for our purposes.) There are lots of ways we can make that type of mistake, even if we conclude that in this case the basic finding still holds, which we might yet conclude. We can mix up conceptual opposites like nature and nurture by making bad assumptions, based on the narrow cultural firmament of contemporary academia – the assumptions of urban white liberals, in essence. There's nothing wrong with being urban, white, or liberal, or all three – my point is simply that this is one culture in a constellation of cultures.

Anyway, I love twin studies like Krapohl's.


* The drama over the exams in Harry Potter is more easily understood if you know about the British exam system, the A-levels, etc. Americans have no equivalent, even in states that mandate exams for high school graduate.