Posted: Wed 02 Jul 2008 21:40 Post subject: 1/3rd of White Americans have recent African DNA claim
Frank, I was hoping you could expound on this claim because I have been quoting it in many conversations, but Mark Shriver has made me have doubts as to the strength of the claim. Not that it is accurate, but that we have enough info to make the claim yet in the first place.
I was hoping you could clarify because I see your source was an interview of Mark by Steve Sailer in 2002. So I wanted to check if there was another source apart from that one.
My correspondence with Mark instead of trying to paraphrase and maybe misrepresent him:
When I asked him about the proportions based on his most recent data he responded:
Quote:
There really is not any good new data on this, but I think the old data likely will be close to what is found in larger samples. Of course there is most likely lots of regional variation as one moves across the US, but the average is something like 1% of the European American gene pool is from West Africa. This would equal about 10% of European Americans having one West African great grandparent.
{I then quoted you and Steve Sailer's articles}
Maybe this is the paper he got those number from. Look at the figure 2b. {Shriver et al, Skin pigmentation, biogeographical ancestry and admixture mapping, 2003}. It's actually quite a difficult issue to communicate. Blacks having white ancestry and whites having black ancestry is simply not the way I would say it. We need to have some words to distinguish the parental gene pools from the contemporary populations. Also, an important point is what level of ancestry is biologically/historically significant? If you think even the great-great grandparent contribution (6.25%) is significant you have to be very careful in interpretations as many estimates of ancestry at this low level will be imprecise. For this reason I like to go from the gene pool estimates. Don't forget too that we are all African if you go back in time far enough. An we are all modified African apes too! We can talk about this if you like.
{To which I mentioned recent African ancestry only, like in the US slavery to recent times era.}
One of how many ancestors? This is the question. If the number you care about is small, it can be increasingly difficult to estimate this number with any accuracy.
One has to be careful in this field with any particular numbers. Steve Sailor made a big point of the change in my ancestry estimate and his criticism is really just an indication of how much he thinks about individual numbers. I'm forwarding you an analysis I asked Indrani Halder about this morning based on these conversations.
How much confidence is there in the 1/3rd of Euro-American population claim? Should I still use it or should I be a little more cautious throwing that statistic around? Thoughts?
Joined: 30 Mar 2005 {Posts: 1047 } Location: New Jersey
Posted: Wed 02 Jul 2008 23:34 Post subject:
For what it's worth, my understanding (by corresponding with Frank and also with Lou Charlton, formerly of DNAPrint) of the different figures given for sub-Saharan admixture in White Americans (originally 30%, then 10%, finally 5%) is as follows:
If you are interested in any markers at all that point to sub-Saharan ancestry in European-Americans, then 30% is the figure to use. If you are interested in the percentage of markers that pass the error limits of the test, then 10% is the figure to use. If you are interested in the percentage that can be guaranteed with 95% accuracy, then 5% is the figure to use.
Just because a result is below the 95% confidence interval does not by any stretch of the imagination mean it is useless. If it is reduced to, say, 80%, it still has a pretty good chance of being accurate. An estimate that falls below the 95% confidence interval gains more importance when it is found in enough people of a given population to be what is known as a signature admixture. For example, an individual south European may type with about 5% sub-Saharan markers. The confidence interval may negate this for this individual. But if the same 5% is found in many southern Europeans, it becomes a signature admixture. In other words, it is very likely accurate to say that this population has on average 5% sub-Saharan genes, because statistical noise wouldn't show up so often.
Oh ok. I just don't want to get caught in a claim that later comes back to bite me. What specific study or studies should I site? Frank's article quotes Sailer, but Sailer doesn't give a specific point of reference, just a general interview of Mark.
Oh ok. I just don't want to get caught in a claim that later comes back to bite me. What specific study or studies should I site? Frank's article quotes Sailer, but Sailer doesn't give a specific point of reference, just a general interview of Mark.
I think you are looking for an authority whom you can quote so that you can "win" an argument with someone. I regret that we cannot help you with this. William has explained the situation very well. If you are interested in knowing the facts, you are much better off referring to the raw data rather than seeking a voice of authority. Here is Heather Collins-Schramm's chart. It sampled 147 White Americans, well sufficient to show the pattern.
Lay a ruler horizontally across the chart, covering up the top axis, thus hiding White Americans with no detectable SS admixture. Now count the visible dots of White Americans. There are about 46. This is one third of the total of 147. So two thirds of White Americans (the ones covered up) have no detectable SS admixture and one third (the ones still visible) have detectable SS admixture.
But the tests are not very reliable in the 0%-2% range. For a more conservative measurement, slide the ruler down to cover everyone with 95 percent Euro admixture or more. This is higher confidence interval and leaves only about 14 dots (10 percent) uncovered. So, you can say that ten percent of White Americans have definite (to a high degree of probability) SS admixture.
But most scientific findings are reported at the 95 percent confidence interval. To see this, slide the ruler down to cover all the dots with 90 percent or more Euro admixture. There are only about 7 dots left (5 percent). So, you can say that five percent of White Americans have certain SS admixture (to the 95 percent confidence interval).
Oh ok. I just don't want to get caught in a claim that later comes back to bite me. What specific study or studies should I site? Frank's article quotes Sailer, but Sailer doesn't give a specific point of reference, just a general interview of Mark.
I think you are looking for an authority whom you can quote so that you can "win" an argument with someone. I regret that we cannot help you with this. William has explained the situation very well. If you are interested in knowing the facts, you are much better off referring to the raw data rather than seeking a voice of authority. Here is Heather Collins-Schramm's chart. It sampled 147 White Americans, well sufficient to show the pattern.
Lay a ruler horizontally across the chart, covering up the top axis, thus hiding White Americans with no detectable SS admixture. Now count the visible dots of White Americans. There are about 46. This is one third of the total of 147. So two thirds of White Americans (the ones covered up) have no detectable SS admixture and one third (the ones still visible) have detectable SS admixture.
But the tests are not very reliable in the 0%-2% range. For a more conservative measurement, slide the ruler down to cover everyone with 95 percent Euro admixture or more. This is higher confidence interval and leaves only about 14 dots (10 percent) uncovered. So, you can say that ten percent of White Americans have definite (to a high degree of probability) SS admixture.
But most scientific findings are reported at the 95 percent confidence interval. To see this, slide the ruler down to cover all the dots with 90 percent or more Euro admixture. There are only about 7 dots left (5 percent). So, you can say that five percent of White Americans have certain SS admixture (to the 95 percent confidence interval).
Thanks a lot Frank. By the way, was the sampling regional or national?
Like most such studies, it was her fellow students (University of California at Davis). Shriver's original study gave the same distribution from students at Penn State.
Although a few studies have been published on regional admixture differences within the A-A population (Parra), only a handful have studied Whites (Kaiser).
Nevertheless, there have been some unpublished studies presented at the Melungeon and Redbone conferences that Mary Lee and I attend. None of these have presented scatter diagrams, but they have computed mean SS admixture in White-identified Melugeons and Redbones at around 5.6-6.0 percent. This is nearly ten times higher than the rates on Schramm's (or Shriver's) scattter diagrams for White college students in PA and CA.
As more studies are done, I am confident that regional admixture patterns for Whites will emerge (they are already clear for A-As). I am willing to go on record as predicting that they will show peaks of around 5-6 percent SS admixture in the Appalachians and in Louisiana, with linear gradients outwards and to the north. As a fellow banjo-picker once whispered to me at a Carter family gathering, "You ain't a true southerner unless you got a little nigger in you."
'
FAQ
Search
Register
Log in
