astralcodexten.com
I'm generally a fan of "blurry" definitions where something can qualify as X if it fulfills a few of many criteria. I think trying to create hard rules around blurry areas like race and culture is fool's errand, and Scott does a great job laying out how overly strict definitions can go wrong.
Jump in the discussion.
No email address required.
Notes -
It's an old feel-good rhetorical device to cope with and minimize the importance of differences across races, a device possibly older than most of us participating on this site.
Obviously, when it comes to genetic differences, this can't be true from a smell test alone, as about 15 years ago it was already possible to assign Europeans to their country of origin based on solely their genetic information, and it's much easier when it comes to races. Any of us can download publicly available genetic data, run a PCA on Europeans, East Asians, and Sub-Saharan Africans to get a clean triangle with each of the three populations on a node.
When it comes to phenotypes, this would clearly be trait dependent and false in many cases. The variation in height between Western-born whites and East Asians is almost certainly less than the variation within them, but clearly the variation in lactose tolerance would likely be greater between than within them (it's like 90% one way and 10% the other). By cursory inspection, the variation in skin color is greater between than within European whites and Sub-Saharan blacks. The variation in facial and skeletal morphology is greater between than within whites, East Asians, and blacks, as can be found in forensic anthropology and x-rays.
Of course, just because the variation within groups is larger than the variation between them doesn't mean the differences between them aren't meaningful or important. Men are around two standard deviations taller than women and their within-group variances are similar; it's a coin-toss as to whether if even the variation in height between men and women is greater than the variation within them. For randomly drawn pairs, if the variances are equal at a 2SD average difference and normally distributed, the man will be taller than the woman over 92% of the time.
This is only exacerbated by tail effects, as discussed many times here. If whites have an average IQ a standard deviation higher than blacks and their within-group variances are equal at 15, e.g., 100 vs. 85, then when comparing equal sized populations of whites and blacks, the white population would have over 16 times more individuals with IQs over 130. Blacks would have 7 times more those with IQs under 70 than whites. A randomly drawn white would be smarter than a randomly drawn black about 76% of the time.
Species are also a social construct, as are genus, family, order, etc. and everything in between. What's considered the same or different species (or any group) can be reclassified over time as new information is obtained or as fads change. Yet, hardly would anyone deny the biological underpinning behind classifying organisms. Conservationists generally lean left, but are heavily invested in Organism BioDiversity. Suggest that we should just let gorillas, cheetahs, the red wolf, or the loggerhead sea turtle get replaced and go extinct (what's the big deal, we already have hominids, felids, canids, and turtles? They're all in this together) and see a conversationist react as an ardent purist that would make the most extreme Stormfront member blush.
If you get bitten by a King Cobra, you want King Cobra antivenom ASAP—at least mamba or synthetic Elapidae antivenom as a temporary stopgap—not some antivenom from a random Naja or much less other snake species. While paralysis and tissue necrosis ensues, you definitely don't want a lecture on how species and genera are a social construct and how the King Cobra is not considered a True CobraTM but more closely related to mambas nowadays, although it was considered a cobra at one point. Thus, any random snake antivenom should do. Many different so-called snake "species" can interbreed and can have tremendous overlap in morphology and geography. Are you even sure if it was a King Cobra, or if "King Cobra" is a valid group in the first place? Maybe you're just indulging in and promulgating negative stereotypes involving "King Cobras." After all, you're not some sort of weird bigoted extremist when it comes to snakes, are you?
It's more that there are many ways you could interpret it as an incredibly broad claim, and that some of them are technically true but misleading. I think it usually refers to Lewontin's claim in this article.
Lewontin's Fallacy also works here and usually I mention it in such discussions, but I was thinking about "differences" in a more colloquial sense from the perspective of how a layperson might think of it, before switching to variation when it comes to phenotypes (most people have an easier time imagining phenotypical than genetic variation, especially at a locus-level). Hence the smell test using something more readily visualized, especially since most people are already at least cursorily familiar with commercial ancestry kits like 23andMe.
At a given locus there might be more variation within groups than across groups (not true in many cases, for example, the loci associated with lactose tolerance and skin color would likely have more between-group than within-group variation when it comes whites vs. East Asians, and whites vs. blacks, respectively), but once you consider more and more loci you start to get clean separations. Hence the aforementioned exercise Europeans and PCAs when it comes to Europeans, East Asians, and Western Africans. A randomly selected young athletic man might on average be only about 10% taller, 10% faster, 50% stronger, and have a slightly faster reaction time than a randomly selected young athletic women, with both groups having substantial intragroup variation for each of those traits. However, in terms of overall athleticism, almost always the man will be more athletic than the woman.
More options
Context Copy link
More options
Context Copy link
I have not performed this exercise, but my understanding is that the first two PCs you'd use as the axes of this plot will usually explain a negligible percentage of the total variance in a genomic dataset. PCA will always rank axes by variance, but that doesn't mean the top few PCs are any good, in an absolute sense, at reducing the data dimension while preserving structure. Even if the PCs that let us reliably identify these clusters happen to be the relatively 'best' PCs we could use to collapse the multidimensional data onto a graph, they could still suck, in an absolute sense.
Depending on the publication, they can get up to 30% from the first two PCs, although it can be low like high single digits.
Regardless of the number, they're not sucky nor negligible when you can do a PCA using 5,000 random SNPs from 300 random individuals (100 individuals each of the three populations of Europeans, East Asians, and West Africans) and get a clean triangle. Pick another 5,000 random SNPs from another 300 random individuals and you'll get another clean triangle. Rinse and repeat. Quite robust.
If anything, it's a testament to how different genetically Europeans, East Asians, and West Africans are, that one can get consistent, clean separation in such low dimensionality (e.g., 2), robust to using different individual and different loci. One can always add more PCs to get more total variation captured, but it's just unnecessary for separating those three populations.
More options
Context Copy link
that is, assuming 1 gene corresponds to 1 trait and all traits are equally important. But suppose some trait (say, amount of melanin in skin) is determined by 10 genes and one population has 35% frequency of dark allele and other population has 65% frequency of dark allele. Most genetic variation is within populations but looking at their resulting color, they are visibly distinct. Height or intelligence are much more polygenic.
There are lots of genes related to immune system in which variants are no worse or no better, they just make it harder for germs to jump from one individual to another, and selection will favor diversity in this gene. Other genes would be selected in one direction and can reach fixation. Pick traits you are interested about, and then estimate how much PCs explain it.
More options
Context Copy link
The original argument was about - is there a genetics-related meaning of "race", or is it just a social construct?
I'd argue that race is, genetically, as real as the color of a fruit or vegetable. If you had huge matrix with tens of thousands of low-level properties of various fruits, the 'color' variable would explain very little of the total variance, but color is still a property worth discussing, a common kind of variation. The question of how much of traits we care about race 'explains' is a different one with a higher bar for evidence, but the PC does show that "race" does have some meaning, even if a weak one. It rebuts "race is a social construct" when itself used as a rebuttal to more substantial arguments about race.
Undoubtedly there is a genetics-related meaning of race, in the sense that there are identifiable genetic markers that discriminate (heh) between people of different racial categories.
I should have made clear in my reply above that I was specifically questioning the implication, in the post I was replying to, that a PCA plot showing distinct racial clusters can rebut 'the old feel-good rhetorical device" that there's more genetic variation within than between. It does not necessarily do this, in the situation where the PCs showing those distinct clusters themselves explain a negligible fraction of overall variation.
But that's fine! Rebutting "more variation within than between" does not seem not necessary for race to have a genetic basis.
More options
Context Copy link
genetic differences and variations exist, but race as a category is socially defined. Socially defined does not negate usefulness (as with many systems in our society). Money is a social construct, but we don't consider it useless. The genetic variation and differences are real, but you also need to keep in mind most of these broad claims about IQ or even physical characteristics fail to take it a step further and discuss the numerous factors that influence such differences (socioeconomic, environmental, cultural and geographical).
This is technically true, but I think what you mean to do is imply that the social construct "race" doesn't map perfectly onto "ancestry / genetics" in a way that affects our interpretation of things like "race and IQ are related". I think "race" and "ancestry" are close enough that it's not important to clarify you mean "ancestry's effect on genetics" when you say "race".
The sophisticated ones don't! It is true that BasedHitler1488 on twitter has a very inaccurate view of the literature on the heritability of IQ or the association of that with race, but that's pure a weakman.
The science on the topic of the heritability of IQ and physical characteristics among individuals is extremely clear and mainstream, and directly deals with factors "socioeconomic, environmental, cultural, and geographical!"
The race-causes-iq-differences arguments are not scientific consensus. Imo this is mostly because saying 'black people are dumber because genes' is something that most existing Americans, including most smart ones, (for various reasons) have extremely strong negative reactions to. But they do directly deal with "socioeconomic, environmental, cultural, and geographical" too.
For some intro reading, check out JayMan: https://jaymans.wordpress.com/jaymans-race-inheritance-and-iq-f-a-q-f-r-b/ https://jaymans.wordpress.com/hbd-fundamentals/
Note that JayMan is black!
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link