Greg Cochran contradicted Steve Sailer in a comment at Steve’s blog. I asked him to elaborate, and here is his response:
Total genetic diversity is higher in Africa: most of that is neutral diversity that doesn’t do anything, just as it is in any other group. Is a given Swede closer to another Swede than to any sub-Saharan African, by a measure of genetic distance (one that does not consider the actual effects of those genetic differences)? Yes, always. If this true if we restrict the measure to noncoding parts of the genome? Yes. Is it true if we restrict the measure to coding part of the genome, the fraction that makes proteins? Still true. Is it true if looked at a measure of phenotypic difference, which would correspond to a genetic distance in which genetic differences were weighted by their effects? Even more true.
What’s the right way to define race? Not as a partially inbred extended family. It depends on the level of differences you are considering. At the finest level, any population (a group that mates within itself fairly freely and has for quite a while) that is different enough in some interesting observable quality from some other population could be considered a race. At a coarse level, large subdivisions of human that have been geographically isolated and had low levels of mixing until fairly recently: this corresponds moderately well to traditional groupings such as Caucasoids, Australoids, Amerindians, etc. Things have become a good deal more mixed (in some areas) since the development of decent sailing ships, especially after Columbus.
We now know there are interesting differences in ancestry/origins, and those may have some surprising implications. It looks as if Eurasians all have some Neanderthal ancestry while sub-Saharan Africans do not, or at least have a lot less. ( I’m sure we’ll find some in groups influenced by Arab gene flow, such as the Somalis and some peoples in coastal east Africa). That Neanderthal genetic contribution should have furnished some adaptively important alleles, in my opinion: but we don’t know that for sure yet. We might know fairly soon. There may have been other regional mixtures with other old-fashioned hominids: there are hints of this for Pygmies, Bushmen, and Australoids. To some extent, some of the existing racial groupings may be the shadows of past subspecies – although all populations seem to be mostly descended from a fairly recent African population.
July 26, 2010 at 6:06 am
I once heard Henry Harpending on Electic Politics podcast on race. He also agreed that race was a hard concept to define and acknowledged that race critics and/or deniers had a point. He defined race, I’m slightly paraphrasing from memory here, as a common evolutionary regime – regime being the same environmental and cultural forces interacting with the (human) species.
The ‘partly inbred extended family’ definition seems too static. It doesn’t emphasize enough why and how races are (or become) different from each other. But that’s just my two cents.
July 26, 2010 at 8:28 pm
I’m jealous. I asked him to clarify too but he answered you. Probably because you’re well-known and I’m just some guy.
:(
Good job though; interesting post.
July 26, 2010 at 9:05 pm
Cochran misinterpreted Steve’s comment to mean that a white guy is closer to any given African than to his own brother.
Reread the quote he was responding to. Steve wasn’t making that claim, and didn’t state anything inaccurate.
July 26, 2010 at 9:30 pm
Malcolm Gladwell (unsurprisingly) provides a perfect example of the Common Educated Person Misinterpretation of African genetic diversity Steve was referring to:
“… in almost any single African population-a tribe or however you want to define it-there is more genetic variation than in all the rest of the world put together… So you can expect groups of Africans to be more variable in respect to almost anything that has a genetic component… to the extent that running is influenced by genetic factors you would expect to see more really fast blacks-and more really slow blacks-than whites but far fewer Africans of merely average speed.”
The “Crazier” misinterpretation of African Genetic Diversity Steve was (parenthetically) referring to was this:
Radley Balko: “A black man and a white man from Manhattan, for example, are likely to be more genetically similar than a black man from Manhattan and a black man from Nigeria.”
Cochran misinterpreted Steve’s comment as an assertion that the Balko statement is correct for neutral genes.
July 26, 2010 at 10:03 pm
Here’s a similar misinterpretation from Tim Wise:
“On average, humans will vary by eight to nine times more within a population group than between a population group so that I as a European-descended person will be eight to nine times more different from another European person genetically than I would be from someone… who was Asian, or who was African.”
Actually, as much as I’d like to distance myself from this logic, I make a similar error in the very same post. I do this because a paper by geneticist Michael Bamshad published in Nature purported to demonstrate empirically that people from different racial groups are more similar to each other one-third of the time.
I should have known better. The Bamshad paper was confusing because it explicitly disavowed Lewotin’s Fallacy and then surreptitiously committed it to make a similar point.
The flaws of Bamshad’s claim were quickly debunked in the comments by Cochran, and were later published in Witherspoon 2007:
“Thus the answer to the question “How often is a pair of individuals from one population genetically more dissimilar than two individuals chosen from two different populations?” depends on the number of polymorphisms used to define that dissimilarity and the populations being compared… if genetic similarity is measured over many thousands of loci, the answer becomes “never” when individuals are sampled from geographically separated populations.“
July 26, 2010 at 10:15 pm
B Lode: Perhaps for sufficiently trivial values of “well known”. As long as there are enough “some guys” who occasionally get a response, we can all be informed. Such are the wonders of the internet.
I admit I misread Cochran as saying that Africans are more diverse on non-junk genes. Does anyone know of research that has been done on that?
I actually think Balko’s statement doesn’t seem that crazy. “Blacks” in America can have majority European ancestry, and I think the ones in the north are more likely to (for Manhattan, Harold Ford!). But I don’t think it qualifies as “likely”, particularly when specifying Nigeria (I believe the major source of slaves in America).
Another criticism of Gladwell here. More importantly, he’s wrong about hockey.
July 27, 2010 at 7:11 am
“I admit I misread Cochran as saying that Africans are more diverse on non-junk genes. Does anyone know of research that has been done on that?”
Africans are more genetically diverse on non-neutral genes, but that doesn’t necessarily imply greater trait diversity– a la Gladwell– because of natural selection (e.g. non-African populations are certainly more diverse in pigmentation).
But diversity in physical (and presumably psychological…) traits under more relaxed selection, such as cranial and tooth shape, map on to genetic diversity: Africans are more variable, and variation decreases with distance from Africa.
““Blacks” in America can have majority European ancestry, and I think the ones in the north are more likely to”
The most recent research I am familiar with failed to detect greater white admixture in Northern blacks.
I don’t think Balko was referring to admixture, but rather to the same fallacy in the Time Wise quote, where the diversity within groups somehow completely inverts the relationship between groups.
July 27, 2010 at 10:26 am
>>”At the finest level, any population (a group that mates within itself fairly freely and has for quite a while) that is different enough in some interesting observable quality from some other population could be considered a race.”
That sounds a lot like a “partially inbred extended family”.
July 27, 2010 at 2:18 pm
Steve Sailer’s definition is a sociological one as much as a biological one. In his post where he presents that definition, he says that the conflict between Ulster Catholics and Ulster Protestants is a racial conflict. I’d be surprised if there are any phenotypic differences of any significance between the two populations. However, since Ulster Catholics only marry other Catholics, and Ulster Protestants only marry other Protestants, their sources of information and attitudes towards the outside world are formed differently.
So in biological terms, Sailer’s definition may not be that useful. In sociological terms, it seems to be very much so.
July 27, 2010 at 6:38 pm
Oi vey I’m getting lost.
Let me step up to the plate and we can see if the inherent complexities of population genetics can blow a fastball by me.
Africans are more diverse than
(a) any other race
(b) the rest of the world combined in terms of
(c) neutral genes which
(d) used to be called “junk DNA” and
(e) in terms of non-neutral (coding) DNA, but
(f) they are not likely to be more diverse, least not than the whole rest of the world combined, on some / most / all traits.
Other than A or B, and the “traits” F, I’m pretty sure that is what Jason Malloy has said. Is that right?
Now guys, let me swing at another: is it true that there is more genetic difference between two human races (can’t remember which one) than there is between red wolves and coyotes? Or did I make that up in a feverish dream?
The Parra, Marcini, et al. article Mr. Malloy linked to is pretty interesting. It looks like the most Euro- of the African-American populations are in NYC, Pittsburg … and New Orleans (about a fifth). Charleston and Philly are the most Afro (about one eighth Euro). So there’s not much of an axis visible from those data.
July 30, 2010 at 8:12 pm
Oh well. Having killed this thread, I’ll lay this lovely bouquet on its headstone.
“… there are larger genetic differences between a West African, a Northwest European, and an Northeast Asian, then there is between two separate species, a wolf and a coyote …”
July 30, 2010 at 11:28 pm
Jason Malloy:
Cochran misinterpreted Steve’s comment … Steve wasn’t making that claim, and didn’t state anything inaccurate.
Steve did make an inaccurate statement and Greg corrected it without any misrepresentation.
Sailer: “genetic diversity is highest in African populations … this is the source of … in increasingly crazier variants, that two white Americans might [be] less related to each other than to a black African … This is true for neutral (“junk”) genes that aren’t selected [comma] not for functional genes.”
Of course it is not true for neutral markers. That’s Cochran’s point, obvious and indisputable.
flenser:
That sounds a lot like a “partially inbred extended family”.
Only because “partially inbred extended family” sounds a lot like “population”.
July 31, 2010 at 6:08 am
Nanonymous,
Fair enough; I take “didn’t state anything inaccurate” back. He is wrong that African diversity is restricted to neutral genes.
Rather the primary criticism here is wrong. You are misinterpreting his comment as well. Here is how his comment was intended, without the misleading parenthetical:
“By the way, this is the source of the widely held dogma/ urban legend for the quasi-educated that black Africans are the most genetically diverse people on Earth… This is true for neutral (“junk”) genes that aren’t selected not for functional genes.”
Ok, wrong. But this next portion nested in the center was intended only to note illogical beliefs that have mutated off of the original “error”. It was not intended to mean that Steve believes these ideas are all true for neutral diversity:
“… or, in increasingly crazier variants, that two white Americans might less related to each other than to a black African, or that you and your brother are less similar to each other genetically than you are to an African.”
July 31, 2010 at 12:45 pm
Sailer: “genetic diversity is highest in African populations … This is true for neutral (“junk”) genes that aren’t selected not for functional genes.”
Nanonymous: “Of course it is not true for neutral markers.”
Jason Malloy: “He is wrong that African diversity is restricted to neutral genes.”
Unless someone is stumbling into misleading syntax, I just don’t think there is any room for agreement between these positions. I’m not disputing anything, since I don’t have the numbers (I WOULD REALLY LIKE TO SEE THE NUMBERS), but it seems like it is really really difficult to get this information across using ordinary English. (I read Cochran’s three-word correction of Sailer’s assertion like five times before I gave up.)
July 31, 2010 at 3:04 pm
B lode, I’ve tried to explain the confusion.
Nanonymous is saying that it’s not true that Swedish people are more related to Africans than other Swedish people on neutral markers. He is entirely correct, but that really is not what Sailer was trying to claim.
Nanonymous, Cochran and I are all in agreement that Africans are more diverse on both neutral and non-neutral genes, while Steve incorrectly claimed otherwise.
August 1, 2010 at 4:07 pm
“I admit I misread Cochran as saying that Africans are more diverse on non-junk genes. Does anyone know of research that has been done on that?”
There was a troll on Overcoming Bias yesterday playing the ‘blacks are closer to chimps’ card to imply racial inferiority (But somehow not the superiority of Amerindians. Such people make it harder for those with better information and intentions to jump in; something provocateurs can easily exploit).
It’s easier to understand both population ancestry and differences in genetic diversity if you think of each split since Out of Africa as a smaller population branching off from a larger one. Each time the smaller break-away population takes only a subset of the extant variation of the larger ancestral population.
Europeans are a subset of the African gene pool, Asians are a subset of that smaller European gene pool, and Native Americans are a subset of that even smaller Asian gene pool, etc.
Most of that lost genetic diversity is neutral because most genes are neutral. But the smaller % of genes that aren’t neutral are siphoned through the same bottlenecks. It doesn’t matter what kinds of genetic variation scientists look at:
“Studies of protein polymorphisms, as well as studies ofmtDNA, Y-chromosomal, autoso-
mal, and X-chromosomal DNA variation, indicate that African populations are the most variable and ancestral, as expected under an RAO [Recent African Origin] model. Phylogenetic analyses of mtDNA and Y-chromosomal haplotypes indicate that the most ancestral lineages are African specific and that all non-African lineages can be derived from a single ancestral African haplogroup, consistent with the
RAO model. Africans have the largest number of population-specific autosomal, X-chro-
mosomal, and mtDNA haplotypes, with
non-African populations harbouring only a subset of the genetic diversity present in Africa, as would be expected from a genetic bottleneck during migration out of Africa.” [See link for references]
As I showed above, this applies to phenotypes under more relaxed selection as well (Africans have more HLA, etc, diversity as well, but this is expected from a selection perspective as well).
Non-African populations, on the hand, have undergone considerably more natural selection than Africans, and picked up many new genes from the large mutation-generating population sizes of the Eurasian genetic highway.
August 2, 2010 at 7:20 pm
Jason, the commenter daedalus2u was referencing Peter Schonemann as someone whose results have not been addressed by mainstream intelligence researchers. I’d never heard of him, have you?
Speaking of that branch population issue, I’ve always been suspicious of the claim that we can’t claim to be any more related to common chimpanzees than bonobos (and we can’t say our shared ancestor was more like one than another). There are a lot more chimpanzees than bonobos, who seem to be restricted to a small environment under conditions that would have been unusual. Using the same logic that shows non-Africans to be a smaller branch off the main line, could we conclude similarly for bonobos?
August 3, 2010 at 5:38 pm
“Jason, the commenter daedalus2u was referencing Peter Schonemann as someone whose results have not been addressed by mainstream intelligence researchers. I’d never heard of him, have you?”
Actually there are more technically sophisticated criticisms of Jensen in the literature than Schonemann, and there is certainly no evidence that these researchers were ever suppressed by the race realist cabals behind every major statistics and psychology journal.
Perhaps not coincidentally, Jensen’s worthy challengers are similarly Dutch surnamed psychometricians like Conor Dolan, Jelte Wicherts, and Gitta Lubke.
These scientists argue that Jensen’s statistical methods are flawed (though for different reasons than Schonemann), and instead offer Multigroup Confirmatory Factor Analysis (MGCFA) as an appropriate method for testing test bias, Spearman’s Hypothesis, and even the genetic basis of group differences.
So far they seem to have concluded that MGCFA shows that intelligence tests are not racially biased and that SH is shown still tenable (in some form), though remains not strictly demonstrated.
August 9, 2010 at 9:53 pm
What’s the right way to define race? Not as a partially inbred extended family. It depends on the level of differences you are considering. At the finest level, any population (a group that mates within itself fairly freely and has for quite a while) that is different enough in some interesting observable quality from some other population could be considered a race.
Put in more elementary terms, he is talking about a bunch of people who share common ancestors (family) that other groups do not have and who mostly mate with people who share the same ancestors (partially inbred).
So this really does not contradict Steve’s definition, except to the extent that you want to argue that the observable phenotypic difference is a sine qua non of being a race, in which case Cochran’s definition amounts to ” partially inbred extended family with a characteristic phenotype.”
At a coarse level, large subdivisions of human that have been geographically isolated and had low levels of mixing until fairly recently:
This is still a partially inbred extended family; the only difference is that at the coarser levels breeding is not free within the group, which implies that the coarser-level races can be broken down into smaller races. All of the various Congoid tribes share common ancestry that Caucasoid tribes didn’t; put another way, the split between Congoids and Caucasoids occurred before the spli between Mediterraneans and Germanics, or between the various Germanic tribes, etc.
Cochran’s definition of “population” is simply the lowest level of “race” before it becomes impossible to break smaller races out.
August 10, 2010 at 10:38 pm
The phenotype version of race sounds like the old version in which there was a “Nordic” body type within ethnies. Being an offspring of modernity, I’m more comfortable with the more socially/genealogically-derived model.
August 21, 2010 at 5:18 pm
“There may have been other regional mixtures with other old-fashioned hominids: there are hints of this for Pygmies, Bushmen, and Australoids.”
What does he base this on?
August 23, 2010 at 11:34 pm
I don’t know. I’m guessing that their DNA contains chunks which seem very old due to diversity but are also very distinct from other populations.