A couple of years ago, Daniel J. Benjamin, a behavioral economist and associate professor at Cornell University, noticed a disturbing trend in genoeconomics, the nascent discipline that seeks to tie human genetics to traits relevant to the social sciences, like risk aversion, happiness, or even self-employment.
Most of the work was statistically weak, he found, conducted on small samples of a few hundred people. Benjamin calculated that scientists could legitimately conclude almost nothing from those studies. It was a black mark on a charged discipline, one that invariably brings up the hoary nature-nurture debate and past associations with eugenics.
Benjamin wanted to do more, however, than just criticize the field—though he did that well, especially in a joint review paper published last year. That’s why he and several other researchers began, two years ago, the Social Science Genetic Association Consortium, a group dedicated to pooling data in search of genoeconomic insight. To find any legitimate links, they decided, they had to increase their study size to at least 100,000 people.
Building such a sample is difficult, at least for now; given the amount of genetic sequencing under way, that could change in a couple of years. Across 42 data sets, however, the group did find one variable that was consistently collected in genetic-sampling studies around the world. It was a variable that had existing methods for translating national differences. And it was one likely to have at least some small link to your genes.
It was how far you had made it in school.
The consortium—also led by David Cesarini, an assistant professor at New York University’s Center for Experimental Social Science, and Philipp Koellinger, an associate professor at Erasmus University Rotterdam—has now released the first major paper from its work, published on Thursday by Science. The researchers screened 126,559 people for the study, finding three robust regions in the human genome that connected, in a microscopic way, to educational attainment.
How small? For perspective, the largest link they found could account for only 0.022 percent of the variation in the subjects’ educational advancement, drastically smaller than the largest genetic influences found for, say, height, where one gene might explain up to 0.4 percent of the variation. Over all, the genome variants the researchers surveyed could explain, combined, 2 percent of their subjects’ educational success.
The group’s work is both a negative and a positive finding, in a way. It supports the researchers’ assertion that work finding large genetic ties to human behavior is likely to be flawed, while still compiling the hard statistics needed to justify their field’s existence. In effect, it argues that genoeconomists are ready to play at a higher level, and everyone should start listening.
There’s much to take from the group’s work, and I encourage you to read the study and, more important, a series of questions and answers written by Benjamin and his colleagues. But perhaps the most important lesson for the public is this: There will never be a “gene for educational success” or a “gene for entrepreneurship,” just as there will never be a “gene for intelligence” or a “gene for personality.”
“You just shouldn’t believe anything that says it’s the ‘gene for education,’” Benjamin says. That’s true for pretty much any human trait, down to height and weight, but it applies doubly so for socioeconomic outcomes. “The effect for any gene is going to be vanishingly small.”
This is old news for genetics researchers, who have been wrestling with that reality for the better part of a decade. Genes are great at predicting what proteins the body will make, but they are far, far removed from human behavior, and every step along the way, their influence is weakened by a host of environmental factors. There is no linear chain that runs from, say, genes to personality to educational success; when such connections exist, they form, at best, a dense web.
Even the parts of educational success that can be tied to genetics—given a larger sample of one million people, Benjamin expects the number could rise from 2 percent to 12 percent or so—can still be deeply influenced by the environment. Take the example of phenylketonuria, a metabolic disorder, caused by a single mutation, that makes it impossible for the body to use an amino acid, leading to mental retardation. It’s a disease mediated by the environment; early detection and strict dietary guidelines ultimately allow its victims to lead healthy, normal lives.
What would be a somewhat analogous scenario for educational success? This is purely hypothetical, Benjamin stresses, but say there’s a gene variant that increases the likelihood to read books, and it is the reading, in turn, that helps determine scholastic futures. (I said it was hypothetical.) We could still encourage kids who don’t have the variant to read, raising their chances of educational success. Nothing would be predetermined.
Benjamin hopes the new study will also help with a sort of reverse engineering, pointing back to areas in the genome connected with traits closer to biology than whether you graduated from high school. Indeed, the researchers found the same genetic regions, in a pool of Swedish soldiers, explained 3 percent of the variation in their skill on intelligence tests—more than it did to explain their educational path. That finding seems to support the notion that cognitive abilities are one strand in the long web between genes and graduation. That distance could also help explain why the individual effect sizes for educational success are so tiny; each step away from genes reduces their influence. But that’s far from proved.
So what use is the work? It would be incredibly helpful in designing studies of social interventions, like providing universal preschool or offering prenatal care to poor women. Those social experiments are so expensive that studies of them are often small in number, but if you could remove the genetic effects, that could provide a huge boost to their statistical power. That’s what gets Benjamin truly excited about the work.
Beyond that, members of the group are quite explicit about what lessons public-policy experts should take from their work: none at all. Given the limited genetic influence, even if Benjamin and company do tick it up to 12 percent, it’s unlikely you’ll ever be able to predict people’s educational future with their genome sequence. Don’t expect to see such a genetic test appearing on college-admission exams. (Sorry, Gattaca fans.)
There’s another hypothetical Benjamin uses as our phone call wraps up, about how the environment can influence genetic connections. Imagine if you ran a genetics study of college-completion rates 100 years ago.
“You’d find a very strong effect,” he says, “of the number of X chromosomes you have on whether you’ve completed college.”
