Book Review: The Genetic Lottery
The Genetic Lottery is by Kathryn Paige Harden. The subtitle is Why DNA Matters for Social Equality. I'm trying to get better at writing quick books reviews for books I want to remember but don’t feel like I have an excessive amount I want to say about them — this is entry #2.
The main theme of the book: DNA matters for educational outcomes. Over the last number of years, the "liberal" political position has generally been to deny that there are differences in intelligence between people that are caused by our DNA. This position came from how discussions of genetic intelligence have traditionally been closely linked with eugenics and racism. However, Harden says that if you care about social justice, if you care about privilege, you need to care about DNA differences in intelligence. Winning the "genetic lottery" is one of the bigger forms of privilege out there. (Almost everywhere in the book, she focuses on educational outcomes, and not "intelligence", per se. The two are related but not exactly the same. But it's fair to say "DNA matters for intelligence" is true as well) In her words:
Considered together, these points are the ingredients of a new synthesis: Genetics is a matter of luck in people’s lives. Appreciating the role of genetic luck in people’s educational and financial success undercuts the blame that is heaped on people for not “achieving” enough and might, in fact, bolster the case for redistributing resources to achieve greater equality.
Given that many people will instinctively associate her position with racism, she's very careful to develop her arguments. Namely, that:
Genetic differences in intelligence are real
But at the same time, research in genetics does NOT prove that the racial IQ gap is due to genes.
How GWAS works
GWAS is "a genome-wide association study (pronounced “JEE-Wos”)...a GWAS correlates individual elements of the genome with some measurable characteristic of people." Let me get right to it — why is #2 above true? To explain this, she gets into the details of how this genetics research works. I'm going to blunder through it:
Our "genes" consist of however-many billions of those G, C, A, and T molecules. A group of those makes up a gene. Naturally, some genes will be physically located close to each other on the double-helix that makes up our DNA. Others will be far apart. When genetic research is done, the whole genome isn't read for each individual; instead, we use statistics and sampling to make an educated guess about which genes are near which others on that double-helix strand. We find one gene in a sample of DNA, and assume that certain others must be nearby as well:
The probability of inheriting a certain version of gene A is independent from the probability of inheriting a certain version of gene B.
Except, that is, when gene A and gene B are very close to each other, physically, on the genome. Recombination shuffles the metaphorical “deck” of paternal and maternal chromosomes, but it does so badly, leaving chunks of cards stuck together. When genes are physically closer, the chances that recombination will occur somewhere in the space between them are smaller. Very physically close genes are likely to be inherited together rather than separated by the recombination shuffle. Genes that are likely to be inherited together, by virtue of their physical proximity to one another, are said to be in linkage.
OK, so that's part one. Then combine that idea with this one: people of different genetic ancestry are different enough that genes which are "physically close" in most people of one ancestral group aren't necessarily close on people of another ancestral group. She says:
About three-quarters of genetic variants are found only in a single continental group, or even in a single sub-continental group, with African ancestry populations showing the greatest genetic diversity. As a consequence, the genetic variants that are most important for a phenotype in one population are not necessarily the most important in another population: a particular mutation in the CFTR gene, for instance, is responsible for over 70 percent of cystic fibrosis cases in European ancestry populations but less than 30 percent of cases in African ancestry populations... [emphasis mine]
The bottom line, then, is that we cannot and should not expect GWAS results to be “portable” across genetic ancestries or socially defined races... Looking across a diverse set of phenotypes ranging from HDL cholesterol to schizophrenia, polygenic indices based on analyses of European ancestry populations are less strongly related to phenotypes measured in other populations, particularly African ancestry groups. When researchers have used an educational attainment GWAS to construct a polygenic index in White-identified, European-ancestry samples from the UK or Wisconsin or New Zealand, then the score “worked”—it captured more than 10 percent of the variance in educational attainment in those samples. But when researchers tested the polygenic index in a sample of African Americans, who are all expected to have at least some African ancestry, it was much less strongly associated with educational attainment.
In short, modern genetic analysis works by comparing a person to something like a "baseline" or "average person" of similar genetic ancestry — this type of analysis can't say anything about genetic differences between different groups. People of different ancestry groups may have totally different genes that lead to higher educational achievement, or they may have the same genes but located at a slightly different place physically on the genome that leads to these types of studies not working. And, of course, people of white European ancestry are vastly over-represented in genetic research.
Towards the end of the book, she sums it up with this phrase:
genetics are a useful mechanism for understanding why people from relatively similar backgrounds end up different.… But genetics is a poor tool for understanding why people from manifestly different starting points don’t end up the same.
Culturally, this also means that whether your genes make a difference in life outcomes depends on the cultural context around you: "when life opportunities depend less on a family’s level of financial and cultural resources, genes can make more of a difference." If you live in a society where nobody goes to college, then you won't have much educational achievement even if you've got the best genes for it.
As mentioned above, since some of these ideas had traditionally been linked to racism and eugenics, Harden is very careful to develop her arguments, spending full chapters talking about how science works, or what it really means for something to cause something else. For example:
When I teach Introduction to Psychology, I have students practice talking about psychology studies using the following language: “This study was about Construct X, as measured by Y.” For example, this study was about happiness, as measured by people’s ratings on an item that asked how happy they felt today. Or, this study was about social anxiety, as measured by how much cortisol in saliva increased when people were asked to do a short speech in front of unsmiling judges. This language exercise, hopefully, helps them learn to pay attention to how researchers measure abstract concepts like happiness and anxiety, and to be curious about how those measurements might be flawed.
She encourages us to think about a cause as “what is the average difference these things made in the world?”, or in other words, what would the outcome be in a hypothetical world where this particular "cause" didn't exist? The thing is, of course, that those hypothetical worlds...don't exist. We can do studies in the lab, or simulations on the computer, but we can't actually find out what my life would have been like if I had a slightly different genome. She says: "This is what has been called the “fundamental problem of causal inference”: we almost never get to observe what might have been for a single individual."
She does a good job connecting DNA (a microscopic thing!) to the societal level (much larger than microscopic). Sometimes genes act more or less "directly" to cause a certain outcome — single genes that cause a specific disease will cause that disease whether you live in 21st century Finland or 12th century Spain. But other times, genes might "cause" a certain outcome through the way they interact with the environment:
If, for example, a nation refuses to send children with red hair to school, the genes that cause red hair can be said to lower reading scores…
Height could be another example here: genes for height can make us tall, but in a society where nobody is getting proper nutrition, no one is going to end up very tall, and those same genes will be less correlated with height. Another gene/culture interaction is that genetic differences may lead to some feedback effects (which perhaps means we could improve outcomes by avoiding this feedback):
children who had more-advanced cognitive function at age 2 received more cognitive stimulation from their parents at age 4, even controlling for the parent’s previous parenting behavior.
She admits that we don't yet have tons of good examples of equity-based solutions for genetic differences, but there are a few:
eyeglasses are an equity-promoting intervention. People with good eyesight are not being given vision-enhancing surgery to make their vision extra acute.
And while DNA is important for educational achievement, it is only one factor of many:
27 percent of rich children with the lowest polygenic indices graduated from college, compared with 24 percent of poor children with the highest polygenic indices
(A "polygenic index" is, roughly, a number that measures how many genes someone has that are correlated with a certain life outcome—educational achievement in this example.)
Other interesting tidbits
I learned a few things about genetics in this book. Number one is that the sample size in these studies is huge:
An early, high-profile study of 5HTTLPR published in 2003 included 847 people; the definitive rebuttal of that study published in 2019 included 443,264—about 500 times more people.
The second was that, while our genes come from our parents, it's also interesting that our parents have other genes that don't end up getting transmitted to us:
So the chances that DNA from any one specific genealogical ancestor from nine generations ago still lurks in your genome is exceedingly small.
This allows us to do an interesting analysis: we can separate the effects of genes and family environment by looking at whether the parent's genes that don't get passed on to their child are associated with the child's educational achievement:
Are a parent’s untransmitted genes nonetheless related to their child’s life outcomes? If so, then, there is an association between parental genes and child phenotype that cannot be due to genetic inheritance from parent to child; the association must be due to some part of the environment provided by the parent.
[A] study found that for physical traits like BMI or height, your parents’ genes didn’t make you taller or fatter unless you actually inherited them; the untransmitted alleles were uncorrelated with the child phenotype. For education, on the other hand, your parents’ genes are still associated with your own ultimate educational attainment—even if you didn’t inherit those genes. By ruling out biological inheritance as a mechanism for why parental characteristics were correlated with their children’s outcomes, the study showed that it must be the environment provided by the parents that was shaping the children’s educational trajectories. Taking genetics seriously allowed the effects of the environmental privilege to be seen more crisply, providing a direct rebuttal to the eugenic argument that the apparent social determinants of inequality are “really” just unmeasured genetic differences.