Blacks and crime in America

[bldysabba]

@Blake

There is no divergence.

Can you explain where you think the paper diverges from my summary of it?

Sure. The paper uses this method from Pritchard to separate out groups in the population based on genetic data

We describe a model-based clustering method for using multilocus genotype data to infer population structure and assign individuals to populations. We assume a model in which there are K populations (where K may be unknown), each of which is characterized by a set of allele frequencies at each locus. Individuals in the sample are assigned (probabilistically) to populations, or jointly to two or more populations if their genotypes indicate that they are admixed.

Using this method and genetic data collected from 3600 odd people, Tang et al find that

Genetic cluster analysis of the microsatellite markers produced four major clusters, which showed near-perfect correspondence with the four self-reported race/ethnicity categories. Of 3,636 subjects of varying race/ethnicity, only 5 (0.14%) showed genetic cluster membership different from their self-identified race/ethnicity.

Which tells me that they aren’t constructing genetic clusters ‘after the fact’ as you said they’re doing -

Originally posted by Blake
By relying upon self-reported race, and then constructing genetic clusters after the fact, such a system becomes inherently unscientific because it is, obviously, self referential. For such a system to be scientifically valid it would need to start from assigning genetic clusters and then assigning races on that basis. But we can’t do that.

After re-reading your post and the report, one potential way they could be gaming this is that they’ve picked the 360 genetic markers (upon which they base the genetic distance calculations) in such a way that they’re different between the racial groups that they chose in advance. In which case you’re right. But I see nothing in their report to suggest this is true.

[Blake]

I suspect that you are misunderstanding the procedure used to produce those genetic clusters. The Tang paper uses an “approach …similar to that of… Rosenberg et al except that the FBPP population primarily represents a United States–based sample”.

Now if you read Rosenberg, which outlines the approach in somewhat more detail you will note that “Of 4199 alleles present more than once in the sample, 46.7%
appeared in all major regions represented… Only 7.4% of these 4199 alleles were exclusive to one region; region-specific alleles were usually rare, with a median relative frequency of 1.0% in their region of occurrence.”

Further, in the Tang paper, differences in the frequency of alleles between “races” found that 60% of alleles were not significantly different. IOW most alleles were the same between “races”.

Note also that the software used structure “places individuals into K clusters,
where K is chosen in advance”. IOW the researcher has to decide in advance whether to produce two groups or to produce 4 groups because he has 4 pre-selected “races”. Note also that Tang et al found that “allowing for more than four clusters did not yield stable results: multiple runs of structure produced varying cluster configurations”. IOW the researchers decided upon four groups either because that was the level at which the software failed or because they had pre-assigned 4 races. It wasn’t some sport of objective standard. They could just as easily have stopped at 2 races: East Asian and the rest of the US. Or they could have concluded there were no races because there were no stable results produced.

IOW these aren’t unique DNA sequences we are talking about. They are simply combinations of unbiquitous genetic material selected post facto precisely* because *they correspond to our groups. The study didn’t just look at 4 markers. It had to look at over 300 markers, and construct groupings that had not just (Marker B and Marker C) but (Marker 365 and Marker 120 and marker 1 and Marker 287 and… ) or (Marker 563 and Marker 210 and marker 67 and Marker 233 and… ) for Black. Note that these two “Blacks” share no genetic commonality at all that they don’t also share with Whites.

[bldysabba]

@Blake

I suspect that you are misunderstanding the procedure used to produce those genetic clusters. The Tang paper uses an “approach ..similar to that of… Rosenberg et al except that the FBPP population primarily represents a United States–based sample”.

Now if you read Rosenberg, which outlines the approach in somewhat more detail you will note that “Of 4199 alleles present more than once in the sample, 46.7%
appeared in all major regions represented… Only 7.4% of these 4199 alleles were exclusive to one region; region-specific alleles were usually rare, with a median relative frequency of 1.0% in their region of occurrence.”

Further, in the Tang paper, differences in the frequency of alleles between “races” found that 60% of alleles were not significantly different. IOW most alleles were the same between “races”.

Note also that the software used structure “places individuals into K clusters,
where K is chosen in advance”. IOW the researcher has to decide in advance whether to produce two groups or to produce 4 groups because he has 4 pre-selected “races”. Note also that Tang et al found that “allowing for more than four clusters did not yield stable results: multiple runs of structure produced varying cluster configurations”. IOW the researchers decided upon four groups either because that was the level at which the software failed or because they had pre-assigned 4 races. It wasn’t some sport of objective standard. They could just as easily have stopped at 2 races: East Asian and the rest of the US. Or they could have concluded there were no races because there were no stable results produced.

IOW these aren’t unique DNA sequences we are talking about. They are simply combinations of unbiquitous genetic material selected post facto precisely* because *they correspond to our groups. The study didn’t just look at 4 markers. It had to look at over 300 markers, and construct groupings that had not just (Marker B and Marker C) but (Marker 365 and Marker 120 and marker 1 and Marker 287 and… ) or (Marker 563 and Marker 210 and marker 67 and Marker 233 and… ) for Black. Note that these two “Blacks” share no genetic commonality at all that they don’t also share with Whites.

I’ll try and figure it all out and get back to you. Don’t hold your breath
Also, thanks for trying to explain. If I had more time right now I would have kept the conversation up. Cheers.

[Blake]

@bldysabba

After re-reading your post and the report, one potential way they could be gaming this is that they’ve picked the 360 genetic markers (upon which they base the genetic distance calculations) in such a way that they’re different between the racial groups that they chose in advance. In which case you’re right. But I see nothing in their report to suggest this is true.

Two points.

Firstly they didn’t need to pick genetic markers in such a way that they’re different between the racial groups that they chose in advance. The nature of *structure *is that it will do that for them to a large extent. The 93% of alelles that don’t reliable cluster won’t be used because … multiple runs will produce varying cluster configurations. The fact that anybody who didn’t identify themselves as one of the four desired racial groups was excluded from analysis pretty much ensures the rest.

Secondly, the clusters produced suggest that East Asians are distinctly different to the rest of the US, who are quite closely related. IOW we should Blacks to be most closely related to whites and Hispanics. Or they could have stopped at White/Hispanic, Black and Asian races. So why did they continue until they had four clusters? Why not stop at 2? Or go to 6 and conclude that there were no clusters?

[Blake]

And just to put the issue to rest, here’s a more recent offering from Tang. Genetic populations, sure. But even with a tiny sample from limited geographic range, nothing approaching a genetic race.

“in the Middle East and South/Central Asia, there are multiple sources of ancestry. For example, Palestinians, Druze, and Bedouins have contributions from the Middle East, Europe, and South/Central Asia. Burusho, Pathan, and Sindhi have an East Asian contribution. Hazara and Uygur share a similar profile of combined South/Central Asian, East Asian, and European ancestry. In East Asia, only the Yakuts share ancestry with both Europe and America, although these contributions are small. Although much of sub-Saharan Africa, Europe, and EastAsia appears to be homogeneous in Fig. 1A, finer substructures can be detected when individual
regions are analyzed separately.”

Middle Eastern populations are an unsurprising mixture of South Asia African, European genetics in addition to their own characteristic genetic signature. Europeans blend seamlessly into South Asian and ME populations. South Asia into East Asia, East Asia into Oceania.

And definitely no genetic races: "withinpopulation variation accounts for most of the genetic diversity in humans…differences between populations accumulate over a large number of loci.

[Chief_Pedant]

As I’ve mentioned before, while it’s a popular notion that genetically defining “race” is an important preliminary to figuring out if outcome differences are genetic, it’s actually irrelevant. This argument is advanced in error.

What you need to be able to show is that two groups you are comparing differ in prevalences for gene sets. That’s it. You don’t need to show the two groups are internally related, or have specific markers, or anything else.

It’s not clear to me why folks are so persistently confused about this.

Suppose, for instance, I make this statement: Cystic fibrosis is more common in whites.

I get challenged: “Define whites.”
I reply, "The Self-Identified Race/Ethnicity, ‘White’ "

I make a similar statement for blacks: Black males have higher levels of testosterone than whites.

Now these two physiologic differences (and thousands of others) hold true. They hold true because those two SIRE groups have different prevalences for those genes. Period.

You may complain they are stupid groupings (they are). You may complain there are no genetic markers, or that the categories confuse and conflate hundreds of sub-categories (they do) or that there is plenty of genetic admixture blurring the edges (there is) or occasional exceptions (there are).

None of that–none of it–gets around the simple fact that these two SIRE groups–and all the other current standard SIRE groups–have different prevalences for genes. This is because of the general history of how human groups have flowed across the world, and have tended to stay within a particular population.

Whether or not you like the standard SIRE groups ultimately depends on whether you are a splitter or a lumper. But either way, you cannot take any scientific issue with the fact that lumping into SIRE groups creates partitions in which there are different prevalences for genes, and it is those genetic differences which help to drive the phenotypic differences we see.

[Blake]

This just gets funnier and funnier. After literally years of loudly proclaiming that races are genetically distinct and biologically definable, the evidence has become so overwhelming that he has given up. He no longer makes such a claim.

Now he is saying that, you don’t need to be able to show that races are genetically different in order to show that races have genetic differences.

Priceless.

None of that–none of it–gets around the simple fact that these two SIRE groups–and all the other current standard SIRE groups–have different prevalences for genes.

As I already pointed put at length in both this thread an previous threads, any self identified group will have different prevalences for genes (or more accurately allelles). Every single fucking one.

Democrats and Republicans will have different prevalences for genes.
People who watch “NBC News” and people who watch “Fox News” will have different prevalences for genes.
People who wear hats and people who don’t wear hats will have different prevalences for genes.

That can not help but be true because every human being has a unique genetic composition. You *must *always be able to find a unique gene prevalence for any group whatsoever.

The problem is that you then have to make the leap from “Republicans and Democrats have different prevalences for genes” to “The behaviour of Republicans and Democrats is due to different prevalences for genes”

And that is what you comprehensively failed to do and will always fail to do. And you will fail because, as you have finally admitted, race has no biological reality. It’s a social construct with no genetic basis, and as such any racial differences must also be constructed socially and can not have a genetic basis.

This isn’t some obscure debating point. It’s an inevitable conclusion once you accept, as you finally seem to have done, that race is a social construct that has no genetic basis.

[Chief_Pedant]

@Blake

Democrats and Republicans will have different prevalences for genes.
People who watch “NBC News” and people who watch “Fox News” will have different prevalences for genes.
People who wear hats and people who don’t wear hats will have different prevalences for genes.

This a baseless and silly assertion. While any two individuals might have different genes, at a group level I am unaware of any studies confirming this sort of pap.

The science for the gene prevalence differences between the SIRE groups of “black” and “white” is extensive, well documented, and un-challenged.
Typical examples include a markedly different prevalence of diseases such as cystic fibrosis or sickle cell disease, physiologic differences such as muscle enzyme levels, and the sorts of genetic patterns which are used to accurately predict from DNA if a decomposed body would likely have self-identified as “black” or “white.”

Your assertion that any two groups have average genetic differences shows a profound confusion, and you have not a shred of evidence to back up your point.

The reason that SIRE groups differ in gene prevalences has to do with the historic migratory patterns of human groups out of africa and the patterns of their subsequent dispersion, re-mixing and so on (and more probably than not, an intermixing of at least some subgroups with the Neandertal populations).

Perhaps when you have a chance to educate yourself a bit more you can return with less strident and more educated replies, and while you are at it you might stop misrepresenting my position. If you are so confused that you simply cannot make a coherent reply, then might I ask you to simply state your own position and stop trying to summarize mine.

This statement below, for instance, is not only inaccurate, but just plain stupid.

@Blake

"*After literally years of loudly proclaiming that races are genetically distinct and biologically definable, the evidence has become so overwhelming that he has given up. He no longer makes such a claim.

Now he is saying that, you don’t need to be able to show that races are genetically different in order to show that races have genetic differences.*"

I have no way of determining if it is simply a matter that your own capacity to reason is deficient, or if you just want to be inflammatory, but I’d like to think that in GD the minimal obligation is to quote, in context, and not present a wildly inaccurate “summary.”