Could It Be... Genes?

social science's bizarre resistance to considering genetic influence

Hey guys, tomorrow the first post exclusively for subscribers runs tomorrow. Every week there will be one substantive post behind the paywall, and in most weeks there will be an additional weekend post that’s more conversational/freeform/relaxed. It’s $5/month or $50/year. I hope you’ll sign up so I can keep doing this.

A big paper by James Heckman and Rasmus Landersø just came out from the NBER. It uses data from Denmark to look at social spending, socioeconomic equality, and educational mobility. (That is, how well children perform academically compared to their parents, operationalized primarily by comparing parent years of education1 to child grades2.) If you’re unaware Heckman is a rather evangelical researcher (not a bad thing!) who does studies for essentially one purpose: to demonstrate that early childhood interventions of various kinds can make major improvements in a variety of social, educational, and economic metrics3.

Are the results of this paper conducive to Heckman’s project? I can’t say what Heckman thinks, though I’d guess he thinks the result reinforce his perspective. I do not. What Heckman and Landersø find is that Denmark has significantly higher socioeconomic equality than the United States (no surprise) but no higher educational mobility than the United States (no surprise to me, likely a big surprise to some). In other words, all of Denmark’s muscular social programs do not appear to make it more likely that students will outperform their parents in educational metrics. Why? The authors point repeatedly to family influence and its effect on educational mobility. A lot of this is packaged with a methodological debate about how family factors are assessed in the literature4. In the reading of the data that’s conducive to a relentless focus on early childhood intervention, Denmark’s failure to promote intergenerational educational mobility is a black mark on their policy of universal programs. The authors write,

More advantaged families are better able to access, utilize, and influence universally available programs. Universal provision of public services does not necessarily mitigate advantages, and indeed may exacerbate inequality. Targeted strategies are generally more effective, although they are often rejected as politically unpalatable5

What inequality is being exacerbated here? The footnote refers to a paper discussing the tendency of improving economic fortunes for the poor to be outstripped by improving economic fortunes for the rich, creating an increase in inequality even as living conditions improve for those on the bottom. Presumably the universality is bad because some of the benefits go to the rich, and if they did not the advantage of the rich would not grow to the same degree. Still, the authors don’t deny that Denmark is a very equal country relative to peers; it has the lowest Gini coefficient in the OECD, after all. If their universal social programs exacerbate inequality as traditionally understood, it must not be by much. They have remarkably low poverty as well. But Heckman and Landersø are interested in a broader sense of inequality and mobility. For their specific investigation they have (the say) superior ways of measuring different types of social inequality, especially “in human capital formation and education.” Heckman and Landersø think the educational etc mobility they have measured is bad and a reason to avoid “uncritical adoption of Danish policy initiatives.” Denmark social democracy doesn’t produce children doing better in school compared to their parents, so we shouldn’t go chasing their policies.

To me, and to other defenders of Nordic social states, the data can be read the opposite way: yes, educational mobility is low, and yet the country is still remarkably equal, thanks to precisely those universal redistributive programs. Their social programs are shrinking the gap between rich and poor (and raising the floor for the poor) without the need for high mobility. And this just seems far more attractive to me than the obverse; educational mobility theoretically allows people to eventually transcend the economic station of their parents, while universal social programs definitively reduce inequality, reduce poverty, and raise quality of life. This is one of the cornerstones of my book: direct redistribution of cash is far more reliable and far less subject to elite capture than trying to perform economic miracles through education. Another point I have made in my book and elsewhere is that mobility (whether educational or economic) is profoundly overrated by liberals. Why is movement within the system more important than a) the distance between the top and the bottom of the system and b) the overall quality of life within the system? Economic mobility is zero sum. For every family that moves into the top 20%, another falls out.

Anyway, family influence. As you can guess from the title, I think this paper is indicative of a bizarre refusal to acknowledge genetics exist within social scientific research and policy documents. Here’s their discussion of how families might influence the educational outcomes of their children:

Families operate through multiple channels. (i) Through direct parental interactions with children in stimulating child learning, personality, and behaviors. This comes from direct engagement and by setting examples for children to emulate, including supporting, supplementing, and advising schooling and other activities in which children engage. (ii) Through choice of neighborhoods and localities which influence the quality of schooling and the quality of peers. (iii) Through guidance on important lifetime decisions

I can think of another way that children are influenced by family6: through genetic information that is passed on through the reproductive process. Genetic influence would be perfectly consistent with the finding that the United States and Denmark have similar amounts of educational mobility despite significantly different policy and socioeconomic conditions. Educational mobility is similarly low between systems because genetic similarity between parent and child has a large impact on educational outcomes and is unaffected by social policy.

This is an article about how family influences children without a consideration of the most direct and powerful way. The words “gene” and “genes” and “genetic” do not appear in this paper. Neither do “heritable” or “heredity” or “hereditary.” The concept of the transfer of genetic information from parent to offspring simply does not exist in this mental space… in a paper about how families influence the characteristics of their children. I would call this odd, but it’s par for the course in social science research. And I just don’t really get it.

The field of behavioral (population? I think they’re maybe undergoing a rebranding but idk) genetics has been producing research for decades that demonstrates that there’s a strong connection between who your biological parents are and various quantitative behavioral traits, including traits of academic aptitude. Defining this influence verbally is notoriously knotty, particularly given that there are always people eager to police the discussion. But one way or another, a massive amount of evidence shows that there is some meaningful relationship between your genotype and your performance on academic tasks - and this relationship is usually far stronger than the kinds of effect sizes produced in most educational research.

Of course many people immediately insist that any discussion of genes and behavioral traits, particularly academic traits, must be pseudoscientific racism. But this just isn’t true. It’s perfectly easy to believe that the observed academic differences between groups (like between racial groups or between the sexes) are environmental while the perceived differences between individuals are shaped by genes. In fact this is what the vast majority of people in behavioral genetics believe, at least according to my read of the literature. And while the field is controversial it still seems to operate more or less openly on the campuses of many, many colleges, almost all of them very liberal institutions. (The median American university is a very liberal institution.) That’s because most people in the field are perfectly capable of avoiding pseudoscientific racism while still acknowledging that of course our genes influence our brains and our brains influence our behavior.

This is a point I’ve had to make over and over so you’ll forgive me if I’m a bit curt about it: there is no contradiction between saying that the differences between groups are environmental while saying that the difference within groups is partially genetic. Look: if you timed how long it took people to run 50 yards, but half of them had to do so at the bottom of a pool, it would not surprise you at all to find that the people running on dry land would be faster on average. But would that mean that there is no genetic influence on how fast you can run 50 yards? Of course not. There would be variation among runners within each group and that variation would in part likely be influenced by their genes. Would saying that the variation between runners was partially genetic be an argument that the difference between groups is the result of genetics? Again, of course not; it’s the environment they’re competing in. This is not a remotely complicated way of thinking and I don’t understand why so many bright people profess not to understand it.

Part of what disappoints me about the (probably inevitable) misrepresentation of my book as a race science book, beyond the fact that it just isn’t, is that I am very interested in the possibility of a broad conversation about within group differences in education. The race science accusation forces the between-group conversation to the top instead. The research and policy worlds are fixated on between group differences to the point of obsession - the black/white achievement gap, performance by girls in science and math, the widening graduation gap between improving girls and struggling boys…. Of course those are worthwhile topics. But there’s already so much research being done about them. There’s already so much grant funding to investigate them. There’s already so many policy papers out there about them. In contrast I find it bizarre how little the ed research and policy world seems interested in asking, “why do two students who seem to be the same in so many demographic and environmental and familial ways perform so differently in the classroom?” Isn’t that a question of great inherent interest?

But I suspect that the question is avoided because people are afraid of where the answers my lead.

I understand why discussion of genes and education is fraught. I understand that that’s true even if you remove the racial component as I am here and as I did in the book. But fraught issues do not go away because you don’t confront them. I did a lot of podcasts for The Cult of Smart, and as I talked with various hosts with various degrees of skepticism towards the subject, I was again struck by the fact that the left has done essentially none of the basic work of figuring out how to think and talk and make decisions based on genetics in a way that’s humane and aligned with our values. I know there’s some individuals who have tried to do this work and I’m sure there are some good books out there but as a community there simply is no consensus leftist/socialist framework for the ethics of genetics, unless you count “I don’t want to talk about that” as a framework.

But genetics is not going to go away because we ignore it, and as genetic research develops it will only become more important. It’s conceivable that within 50 years affluent parents are going to be able to pay to conceive a child that has been engineered with a genome that is beneficial to intelligence and more. How can we sort the ethics of such a situation out if we’re stuck on “genes don’t have anything to do with intelligence, that’s racist”? Not doing this work is an abdication of responsibility.

Are the family influences that Heckman and Landersø found actually partially genetic in origin? I have no idea. And I wouldn’t expect them to figure that out themselves; it’s not their area of interest or expertise. But what’s striking is that they seem completely uninterested in the question. So do the people talking online about the paper. Wouldn’t it be very natural, based on these findings, to suggest that exploring genetic factors would be a productive next step for qualified researchers? It’s such a bizarre situation we’ve found ourselves in where neither the writers of a major research article nor a big group of people talking about it even consider the possibility that family effects might be attributable to genes - especially given that genetic research has been some of the most immensely consequential of the past 200 years. Humans are animals. Animals are influenced by their genetic code, including in behavior. Maybe it would be beneficial if the social sciences decided to take that seriously.


Years of education is a variable to measure educational ability when you don’t have more direct means of assessing it, used because it correlates decently with a large number of other metrics.


I have argued many times that movement within the performance distribution over a child’s lifetime is far lower than people want to admit, which is distinct from but related to educational mobility in this sense. People don’t want to admit it, generally, because it offends their sense of fairness.


This can be something of a hard pull because, despite what many people will tell you, a most of the literature about early childhood/Pre-K programs - the most commonly endorsed early intervention - shows truly discouraging results, particularly the larger and higher-quality studies. (I will try and do a lit review sometime in the future.) But Heckman is very committed.


The specific dispute is about the measurements of neighborhood effects, most prominently used by Raj Chetty. I would not claim to really understand that debate at present.


I won’t attempt to wade into the immense targeted vs. universal programs debate here, other than to note that “politically unpalatable” is putting it quite mildly. Targeted programs like food stamps get cut relentlessly in the United States because it’s easy to suggest that the group being targeted is undeserving. Social Security, which is available to anyone who works, is politically impregnable because everyone in the country stands to benefit from it.


I will say that I am glad that they are addressing family/parental effects head on here. One things that happens when you read a lot of educational research is that you often scratch your head and say “what about the parents?” The behavioral geneticists say family effects (the shared environment) have little influence on behavioral outcomes, but you’d still want people to ask the question, yeah? There’s plenty of studies that try to account for parenting effects but there are many many more for which parenting is a complete lacuna. Why? It’s a classic hammer-and-nails situation: we have very few policy interventions for bad parenting, so parenting is avoided as a topic of interest. We’ll come and take kids away if there’s real abuse or neglect but other than that we’re not intervening. (And this is a very good thing.) What this means is that there’s usually just nothing for research or policy types to recommend, and because academic research and (especially) think tanks are founded on the idea that all human problems are solvable if we just care enough, that’s unacceptable. This is why pressure falls on teachers in educational discussions time and time again: they are the element most easily manipulated by policy. Unlike parenting or, say… genes.