Angela Saini is a British science writer who belongs to what I call the Cordelia Fine School of Science Journalism (CFSSJ): a school whose members have an explicit ideological bias that colors all of their popular writing. In the case of Fine, her ideology is that there is essentially no evolutionary/genetic difference between the brains and neurology of men and women, and so any behavioral differences we see are of purely social origin. Further, hormones play little or no role in behavioral differences between the sexes. Fine’s motive is good—to reduce sexism and bias—but her modus operandus is not, for it involves misrepresenting science.
To buttress Fine’s ideology, her books aim at debunking every study that contradicts her preconceived thesis, even though there is now convincing indication of not only genetically based behavioral and morphological differences based on hormones, but also of differences in the brain. In contrast, Fine goes easy on studies that support her thesis.
In other words, the CFSSJ is characterized by tendentious science writing and confirmation bias, with the bias occurring in how studies are discussed. In the case of Angela Saini, her ideological bias is that all races are equal in any important aspect of biology, and that investigation of differences between races (or, as I call them, “populations” or “ethnic groups”) is liable to play into what she calls “scientific racism”. Ignore the fact that scientists have been trying for decades to debunk the misuse of science in buttressing racism, and most journalists, particularly those with little knowledge of genetics, haven’t been particularly helpful. Some, like Carl Zimmer, know their onions, while others, like Saini, apparently can’t grasp the fundamentals.
I’ve been struck, especially in the CBC interview with Saini shown below, by her willingness to make insupportable statements about differences between groups. I haven’t yet read her book on the topic, Superior: The Return of Race Science, as our library is closed and I don’t want to pay to support ideologically-based biology. But I’ve read other writings of hers, reviews of her books, both pro and con, as well as listened to YouTube videos and lectures. None of this has disabused me of the notion that she’s a member in good standing of the CFSSJ. Further, as we’ve discussed here, she’s misrepresented the situation at University College London by asserting that the scientists there have papered over the college’s history of eugenics and racism.
I’ll give you just one example of how Saini misrepresents the truth in the service of ideology. I was especially concerned about this one because the misrepresentation crops up frequently in discussion of genetics and “race”, and it’s time that people get the issue clear.
Here’s just one Q&A from that show, but it’s an important one:
Let’s move into the modern era then. Biologists have come up with a really strong scientific critique of the idea of race. Can you take me through that?
Well, for 70 years since this consensus after the Second World War, all that biology has done is reinforce the fact that we are so similar. We imagine the genetic differences between racial groups.
For example, I am of Indian origin. My parents [were] born in India. But if I were to randomly pick a South Asian person on the street and randomly pick a white, Canadian person on the street and test their genomes, it’s perfectly statistically possible for my genome to have more in common with a white person than with the Indian person. That’s how almost complete that overlap is. So we are incredibly similar as a species, and the vast majority of difference that we see is accounted for by individual difference.
Now I’ve tried to parse her statement in a way that it would be correct, but I can’t. In fact, the only way you can say that there’s any validity to her claim of no difference between the South Asians/white Canadians and South Asians/South Asians comparison is to construe “perfectly statistically possible” to mean that you might be able to find one or a couple of South Asians who, throughout their genomes, were more similar to some Canadians than they were to other South Asians. But you will almost never find that. We know this from the genetic data that already exist. You could equally well assert that it’s “perfectly statistically possible” for all the oxygen molecules in your room to move to the other side of the room at once, suffocating you. The error is taking what is possible and making people think that this is what’s common or probable.
In fact, all the genetic data we have shows that Saini’s implications about genetic similarity are wrong. If you want to validate her claim, you would have to look at gazillions of nucleotide bases in the DNA sequences of white Canadians and South Asians (I presume Saini means Indians, as she’s of Indian descent), and show that, on average, the proportion of DNA sites that had identical bases in Canadians vs. South Asians was about the same as the proportion of DNA sites that had identical bases in two randomly selected South Asians.
And that, according to the data we have, is not the case. Because of genetic similarities between populations that are spatially (and historically) contiguous, if you find identical bases at one DNA site, it becomes more likely that you’ll have identical bases at other sites. This is easily shown by combining DNA data from different regions of the genome (different “genes” or “SNPs”) to conduct a cluster analysis of overall similarity. And when you do that, you find that populations cluster based on history and geography. Assuming that, say, you’re not sampling a recent Indian immigrant to Canada as a “white Canadian”, or a Candian who lives in Mumbai as a South Asian, you can pretty well diagnose someone’s geographic ancestry—their “population”—from their genes. Here’s an example from 2015 on a very small scale, showing clustering within the British Isles (click on screenshot to access the paper):
Heres a diagram of the clustering, showing how easily someone’s population can be diagnosed from a large sample of DNA bases. Look, for example, at the demarcation between Devon and Cornwall—populations separated only by a river!
What this shows is that if you use information from the whole genome, people’s origins can be largely diagnosed, even on this small scale that used half a million DNA sites—a small fraction of all the DNA sites, which number 3 billion in humans). If you looked at South Asia versus white Canadians, you’d get even more differentiation. Saini’s claim that it’s likely or probable that you could find more similarity between a Canadian and South Asian than between two South Asians is palpably false.
I think where Saini went wrong is that she committed what’s known as “Lewontin’s fallacy,” named after my Ph.D. advisor and discussed in a paper by the geneticist A. W. F. Edwards. What Lewontin originally asserted, correctly, was that if you take all the genetic variation present in the human species, and apportioned it among individuals, among populations within a so-called “race”, and then among “races” (defined as the classical “races”), you find that of all the variation, 85% can be found among individuals within a population, 8% among populations within a “race”, and only about 6% between “races”. In other words, individuals within a population contain nearly all of the existing genetic variation of our species, and when you add different populations or different races, you don’t beef up the variation much more.
Lewontin took this to mean that there are no such thing as genetically differentiated races (as I said, I prefer, because of the historical freighting of “race”, to use “ethnic groups” or “geographically differentiated populations”). And that’s where he made his error. Lewontin is right if you look at each gene separately and then average the apportionment of variation among different genes. But genes among populations are not independent. As I said, if you’re different at one gene among ethnic groups, you’re more likely to be different at other genes as well. In other words, the structure of genetic variation, because of history and evolution, is correlated. I quote Wikipedia on Edwards’s refutation of Lewontin’s conclusion:
Edwards argued that while Lewontin’s statements on variability are correct when examining the frequency of different alleles (variants of a particular gene) at an individual locus (the location of a particular gene) between individuals, it is nonetheless possible to classify individuals into different racial groups with an accuracy that approaches 100 percent when one takes into account the frequency of the alleles at several loci at the same time. This happens because differences in the frequency of alleles at different loci are correlated across populations—the alleles that are more frequent in a population at two or more loci are correlated when we consider the two populations simultaneously. Or in other words, the frequency of the alleles tends to cluster differently for different populations
In Edwards’ words, “most of the information that distinguishes populations is hidden in the correlation structure of the data.” These relationships can be extracted using commonly used ordination and cluster analysis techniques. Edwards argued that, even if the probability of misclassifying an individual based on the frequency of alleles at a single locus is as high as 30 percent (as Lewontin reported in 1972), the misclassification probability becomes close to zero if enough loci are studied.
The cluster analysis mentioned by Edwards was used in the analysis of the British populations given above.
And if the misclassification probability of an individual becomes close to zero when you add more bits of DNA, as it does, then Saini is simply wrong. Yes, races are not nearly as genetically differentiated as early biologists thought they were, and yes, most of the variation in genes can be found in single populations and not among populations or “races”. But you can still genetically diagnose people as to ethnicity by looking at a big chunk of their DNA. South Asians will be more similar to other South Asians than to a white Canadian.
Most of the bits of genome used in these analyses don’t really do much, or have no functional significance in geographic differences in behavior, morphology, or physiology. But populations also differ in meaningful “adaptive” ways because of natural selection. Lactose tolerance and oxygen-carrying ability of the blood are two famous traits, and this paper by Sarah Tishkoff gives many more. Here’s a figure from that paper. It clearly shows that different populations differ in adaptive traits:
Now just because you can diagnose someone’s ethnicity or geographic origin from their genes does not in any sense buttress racism. All it shows is that our genomes reflect our historical and evolutionary ancestry. Saini, in her desire to show that there are no differences, doesn’t seem to realize that the genetic differences used to diagnose people do not place any races above others—there’s no support for any inherent superiority or inferiority of groups. But rather than admit the truth about genetic difference and then say it doesn’t matter morally or politically, Saini would rather throw out the inconvenient data. This is the hallmark of the CFSSJ: if the data go against your ideology, either ignore them or deny them. Or misrepresent them.
This has already gone on too long, but I’ll support my thesis about Saini’s ideologically based science by directing you to another review of Superior: The Return of Race Science, as well as to an Amazon review which is remarkably thorough. Both reviews discuss Saini’s insupportable and misleading claims about genetics. Looking over her claims (another is that there is no genetic variation within populations affecting cognition; see quote from Amazon review below), I can only conclude that in many places crucial to her thesis, she doesn’t know what she’s talking about.
The first requirement for writing sgood cience journalism or popular science books should be this: be sure you understand the science. Or, as Davy Crockett said, “Be sure you’re right first, and then go ahead.”
From the Amazon review of Superior:
On page 221, Saini says, “The question of whether cognition, like skin colour or height, has a genetic basis is one of the most controversial in human biology.” To be clear, this sentence is referring to the causes of individual variation in cognition, not the causes of differences between group averages. The question of whether or not group differences have a genetic basis is indeed controversial, but in 2019, making such a statement about the heritability of individual variation is equivalent to saying that it’s controversial whether or not global warming exists. Ideas such as the existence of global warming or the heritability of cognitive ability are controversial among some political activists, but among professionals in the relevant fields, these questions have been regarded as settled for more than twenty years.
If she really says that on page 221, it’s a howler. The heritability of IQ, for instance, is around 50%, which means that of the variation of IQ scores within a population, half of that variation is due to variation in genes.