It’s tempting to see genes everywhere, lurking in every shadow. For geneticists trying to understand a disease, or Bret Stephens writing an unprompted article about "Jewish genius," genes seem powerful and mysterious, as if they could potentially contain the answer to any of life's questions.
Genetics’ allure can draw people away from more obvious explanations for problems. Here’s a hypothetical. Imagine a poor neighborhood on the side of a highway. If you notice that people living in poor neighborhoods next to highways get asthma more often than rich people across town, you could study their genomes and find some genes common in poor asthmatics. Some of those might even be for genes expressed in the throat and lungs, and then suddenly it seems like poor people are genetically predisposed to having asthma, all while ignoring the much simpler explanation that poor people are breathing in car exhaust while rich people aren’t.
Viewing genes as a determining factor while ignoring larger systemic and societal issues is misleading. For example, scientists recently went looking for a link between individual variations in people’s genomes and their income. Their results, which found that they could predict an individual’s income, in very small part, by looking at individual variations in their genomes, were published in Nature Communications in December.
Using data from about 300,000 white British people, W. David Hill at the University of Edinburgh and colleagues compared the genomes of people grouped in five income brackets. The lowest bracket was made up of those making less than 18,000 pounds per year, the next those making between 18,000 and 30,000, up to the fifth bracket, which contained those making more than 100,000 pounds per year. The data comes from the UK Biobank, a repository of anonymous personal data from about half a million British people for use by health researchers to study essentially whatever they want. The Biobank collects data on a variety of personal characteristics, from things like height and weight to herpes status and sodium in urine. Individual’s genomes are also collected.
Hill’s group highlighted specific locations in the genome that were more or less common within each income bracket—a technique called a genome-wide association study (GWAS). Many of the locations they found were in genes expressed in the brain and were correlated with performing well on intelligence tests.
From this analysis the researchers generated a “polygenic risk score” (PGS), a number that essentially adds up all the genetic variants associated with a trait, which in this case was income. The higher the PGS, the idea goes, the stronger the predictive power. In a second data set of about 30 thousand people, they looked for the same genes that were associated with income in the first set. Though they noted up front that differences in income have environmental and cultural explanations, with this analysis, they were able to account for about 2 percent of the difference in people’s income just using their genes.
What does that mean? Here’s an example: if I make $50,000/year and my friend Mark makes $100,000/year, $1,000 of the difference in our incomes could, in this scenario, be chalked up to the different genes we have. Maybe Mark has genes which give him faster-firing neurons, which make him smarter, which increases his income. The other $49,000 of the difference would come from other things like the environment, or class, or luck, according to the scientists.
The stated goal of the paper was to understand the links between socioeconomic status and health. If genetics play a role in income, the idea is that that can go some way towards lowering health disparities. Without much further explanation on how it will help, the authors say, “An understanding of the causes underlying the association between socioeconomic position (SEP) and health is likely to be helpful to minimize social disparities in health and well-being.”
The paper’s authors did not respond to inquiries, though they included an FAQ section in the paper explaining that "our results do not imply that an individual is in some way predestined to end up earning a certain amount of money.” A Nature Communications spokesperson directed me to the FAQ and declined to comment further.
If your gut reaction to this research is that there’s no plausible way for genetics to meaningfully impact income, or for those meaningless differences to impact health disparities, you would be right. Research like this offers a genetic map to nowhere. It argues for nothing and proves nothing, since it offers no real-world explanation for the problems it’s supposed to study.
People will see what they want to see in such a study. Charles Murray is the author of The Bell Curve, a book stating, among other things, that intelligence largely determines socioeconomic status and that differences in this area explain different statuses between black and white people. Here’s him tweeting about it. Here’s Claire Lehmann, the editor of the right wing apologia magazine Quillette, tweeting some eyes emojis at it. These people use these kinds of studies to reinforce the idea that class differences are at least partly innate, and so the rich and powerful inherently deserve to be rich and powerful because of something within them.
The methods used in the paper are standard-operating-procedure. “It’s a boilerplate [study]. Methodologically, it’s all standard stuff,” said David Baranger, a neurogeneticist at the University of Pittsburgh.
But, there’s a catch. “Almost anything that you can measure these days with a really huge sample, which is what we're dealing with here, there's always going to be an effect,” Baranger said. “It’s always up to the field to decide, ‘is this effect large enough to be something that we care about?’”
With a sample size as big as the UK Biobank you could pick any characteristic you wanted, like, say, enjoying an evening cup of tea. Then, comparing thousands of tea-drinkers to non-tea-drinkers, you might find that there are genetic variations more common in tea-drinkers. But that’s as far as the genetic analysis could take you. Whether tea-drinking is actually driven by genes, and whether that makes any sense, would simply have to be decided by humans.
It’s worth noting that the main statistical crux used here, the PGS, was shown in another work to overblow effects like these by a factor of 10, in some cases. That’s to say that the difference between people when measuring a trait like income can be easily overstated when relying on a PGS.
One of the authors of that study, Arbel Harpak, a geneticist at Columbia University, said via email: “We…devised a test for the presence and substantial effect of [confounding effects] on polygenic scores, and household income lit up as…clearly affected by culturally/environmentally-mediated factors.”
Harpak and colleagues’ study compared two different types of GWASs: one made up of random people, and another that’s made up of family members. If differences in traits, like income, come from genetic sources, the tests will have the same outcome. If a trait is confounded by environmental and cultural factors, the two tests will have different results, which is exactly what Harpak saw. Things like years of smoking, years of schooling, and, yes, household income were heavily affected by the environment. "We really have very little idea on how reflective [UK Biobank] associations are of direct genetic effects,” Harpak said.
The UK Biobank also isn’t a cross-section of humanity since it’s disproportionately made up of white people (the UK itself is overwhelmingly white). People in the Biobank also tend to be wealthier and healthier than non-participants (a common bias in epidemiological data called the “volunteer effect”). That homogeneity can stifle its predictive power. For instance, a genetics study derived from Biobank data to predict risk of schizophrenia was more accurate when used for white Europeans than with anyone else.
The study is emblematic of genetics as a field not seriously reckoning with its potential for misuse by 21st century pseudoscience. Plunking down a paper on how genetics affects income without making any effort to reflect on how exactly this kind of work has been used to justify barbaric practices (like the 20th century eugenicist drive to sterilize anyone deemed an “imbecile”) is irresponsible at best. At worst it indicates a belief in genetic essentialism, the idea that characteristics like intelligence and ability are defined by exclusively by genes.
Substantial differences in health deriving from socioeconomic status are a structural problem and can only be meaningfully addressed through structural changes. There is no huge societal problem that can be solved with a genetics study like this, and any attempts to do will only be fuel for those who think that social inequalities are natural and unchangeable.
Computational genetics is so new and so powerful that it’s easy to look at everything through its lens. Here’s a GWAS for being hot. Here’s a GWAS for being able to dance. Here’s a bot that pumps out GWASs for anything you can think of, like “poultry intake,” or whether you felt loved as a child. Here's a GWAS for same-sex sexual behavior.
If you want to address poor people getting sick and dying while the wealthy aim towards living forever, increase access to health care, and improve working and living conditions. Ignore the 2 percent effect and focus on taxing the top 1 percent of people who own more wealth than the bottom 90 percent combined. That will actually help.