Lewontin's fallacy: more variation within each race than between

[ApostateAbe]

This post is all about Lewontin's fallacy. I judge it to be the second-most popular style of argument used to or either deny or minimize the importance of biological human races. The #1 most popular style of argument seems to be the continuum fallacy; the continuum fallacy has existed ever since the denial of human races began, but Lewontin's fallacy started in 1972, when the Harvard geneticist Richard Lewontin presented the argument in his paper titled, "The Apportionment of Human Diversity." The argument is summarized as follows:

85% of the total genetic diversity of the human species is contained WITHIN populations, leaving only 15% of the genetic diversity BETWEEN populations. Out of that 15%, only 6% can be accounted for by racial classifications. Therefore, racial classifications are of no taxonomic significance.

This argument is not a formal syllogism, and, as the 6% is well above 0%, the argument seemingly requires the implicit natural inference that the 6% would somehow WASH OUT from being overwhelmed by the 85%, or that the 6% would be overshadowed by the 85%, or something like that. Lewontin himself was never clear on his specific reasoning. Nevertheless, the ambiguous argument won wide appeal, especially among academics in every sort of popular and scholarly publication, and the argument remains popular. The ambiguity may have led to more specific variations of the argument that are more directly false, such as the claim that "two people of different 'races' can share more genetic similarity than two individuals of the same race" (Bamshad and Olson, 2004, "Does Race Exist?"), though Lewontin himself did not express this.

In 2003, the general argument found its most notable critic: the Cambridge geneticist and evolutionary biologist AWF Edwards, who wrote a paper titled, "Human genetic diversity: Lewontin's fallacy," a subtitle that coined the phrase. The main counterpoint of Edwards is as follows:

"These conclusions are based on the old statistical fallacy of analysing data on the assumption that it contains no information beyond that revealed on a locus-by-locus analysis, and then drawing conclusions solely on the results of such an analysis. The ‘taxonomic significance’ of genetic data in fact often arises from correlations amongst the different loci, for it is these that may contain the information which enables a stable classification to be uncovered."

It is not easy to grasp the reasoning as Edwards expressed it, but it is possible after some clarification. The way I make sense of it is that there are THREE dimensions of genetic variation at play with respect to human races, as follows:

1. The genetic variation WITHIN each race.
2. The genetic variation BETWEEN races.
3. The genetic variation WITHIN EACH GENOME.

Lewontin's argument incorporates only dimensions #1 and #2, but racial classifications require incorporating all three dimensions of variation. As an analogy, suppose someone were to claim, "The Washington Monument is not so large, because it is only 3000 square feet, not much bigger than an average house." The argument would depend on only two out of three relevant dimensions.

The reason the third dimensions has relevance with respect to the significance of racial classification is not just because I say so or Edwards says so. The significance becomes more obvious after reflecting on the following three points:

1. You can know with near 100% certainty each percentage of the biological races represented in your genome with only a genetic test, and this is done by examining HUNDREDS of your genetic variants in combination (third dimension). It is now established science used by forensic geneticists, by 23andMe and by Ancestry-dot-com (though they may use the phrase "geographic ancestry" in place of "race"). The primary race matches the race you thought you had (your social race) 99.8% of the time, meaning that your ancestry test results will almost certainly be of no big surprise, in spite of the TV advertisements (see Tang et al's 2004 "Genetic Structure, Self-Identified Race Ethnicity, and Confounding in Case-Control Association Studies"). The people who accepted Lewontin's fallacy did not expect this science, as they didn't realize that the third dimension undercuts Lewontin's argument. For example, based in part on Lewontin's argument, the 2005 article by the "Race, Ethnicity, and Genetics Working Group" pooh-poohed the belief that "group differences arise directly through differing allele frequencies."
2. Dog breeds likewise have more genetic variation within each breed than between breeds. The ratio of those two values is Fst (Wright's fixation index), and dog breeds have an Fst of 0.33 (see Parker et al, 2004, "Genetic Structure of the Purebred Domestic Dog"), meaning there is about 3 times as much variation within than between breeds on average, and yet the many recognized breed differences do not apparently wash out. Like human races, the breed of a dog genome can be correctly genetically identified only through examining many alleles in combination (see Koskinen, 2003, "Individual assignment using microsatellite DNA reveals unambiguous breed identification in the domestic dog"). It doesn't even take greater variation between groups than within groups to make two different SPECIES. Chimpanzees and bonobos are two different species, yet they have an Fst of only 0.5 (see Fischer et al, 2006, "Demographic History and Genetic Differentiation in Apes"), meaning twice as much variation within each species on average than between the two. And yet the many genotypic differences between those two species have not washed out (see for example Rilling et al, 2003, "Differences between chimpanzees and bonobos in neural systems supporting social cognition").
3. For that matter, neither do the most obvious differences between human races wash out from the low Fst. If human races can have significant average genotypic differences on such established traits as skin color, height, immune system functions, lung size, skull geometry, nose width, lactose tolerance, etc., then human races can likewise have significant genotypic differences concerning scientifically-racist traits, in which Lewontin's argument was apparently designed to deny (Lewontin wrote in the conclusion of his paper that "Human racial classifcation [sic] is of no social value and is positively destructive of social and human relations"). But none of the known racial differences have washed out due to the low Fst, as the 6% of the between-race variation is all that is required to account for them.

Maybe the racist claim of racial psychological differences is wrong, but the wrongness can not be rightly inferred from Lewontin's argument. Also keep in mind that the failure of Lewontin's argument does not imply that the scientifically racist inferences are correct. They are two independent sets of issues. Most of the researchers I have cited would almost certainly disagree with the scientifically racist claims that Lewontin's argument targets, but that does not make Lewontin's argument a good one. Other arguments must be used to fight scientific racism, not Lewontin's fallacy.

 

[Bomb#20]

For those interested, Edwards' whole article is available here.

Incidentally, other people pointed out the fallacy before Edwards; he's just the guy who finally got people to pay attention. Edwards says Lewontin's argument came pre-refuted -- he says R.A. Fisher (1890-1962, pretty much the founder of modern mathematical population genetics) had already explained the principle involved.

 

[Juma]

I find this text copied from wikipedia, very illuminating on this discussion on race:

, biological anthropologist Jonathan Marks agrees with Edwards that correlations between geographical areas and genetics obviously exist in human populations, but goes on to note that "What is unclear is what this has to do with 'race' as that term has been used through much in the twentieth century—the mere fact that we can find groups to be different and can reliably allot people to them is trivial. Again, the point of the theory of race was to discover large clusters of people that are principally homogeneous within and heterogeneous between, contrasting groups. Lewontin's analysis shows that such groups do not exist in the human species, and Edwards' critique does not contradict that interpretation.

My bold.

 

[ApostateAbe]

I find this text copied from wikipedia, very illuminating on this discussion on race:

, biological anthropologist Jonathan Marks agrees with Edwards that correlations between geographical areas and genetics obviously exist in human populations, but goes on to note that "What is unclear is what this has to do with 'race' as that term has been used through much in the twentieth century—the mere fact that we can find groups to be different and can reliably allot people to them is trivial. Again, the point of the theory of race was to discover large clusters of people that are principally homogeneous within and heterogeneous between, contrasting groups. Lewontin's analysis shows that such groups do not exist in the human species, and Edwards' critique does not contradict that interpretation.

My bold.

To my knowledge, races have never meant within-race homogeneity, so I would love it if Jonathan Marks provided a citation, but he doesn't.

 

[untermensche]

You can call the fact that humans show diversity and show it based on nearness to relations "race", but it is a meaningless division of diversity and a meaningless concept.

But any meaningless concept, like a god, can take form with human definition and specification.

It doesn't make the term less meaningless.