Here you seem to argue that Black African immigrant selection looks limited, albeit still significant (and greater than BC selection)
http://occidentalascent.wordpress.com/2012/02/06/is-global-race-realism-still-tenable/

I was being purposefully contrarian in hopes of getting statistically informed replies. I discussed the issue in more detail here:
http://occidentalascent.wordpress.com/2012/09/02/narrowing-the-genetic-environmental-uncertainty-if-you-wish/
For example: http://occidentalascent.wordpress.com/2012/09/02/narrowing-the-genetic-environmental-uncertainty-if-you-wish/#comment-2494
There are too many statistical uncertainties to estimate IQ/educational selection at this point. To be clear, BA are very educationally selected. Since we don't have a good estimate of the genetic educational-IQ correlation in Africa, that doesn't tell us much regarding IQ.

Why don't you dig up statistics for us? I will point you to the sources if needed.

Well those are still Sub Saharan Africans who are now able to match Northern Europeans in Europe without reverting to the mean and after catching up from a lower base in both IQ tests and educational attainment. This hasn't happened before in the debate as far as I can remember...Selection might be part of it yes, but they do start from a lower base and I think are pretty poor financially on average, even less than BC. However even if you can argue selection, the mere fact they can do this after +50000 years is ominous.

I'v discussed all of this literally years ago and I offered some models. As for Black African selection, you can look at the parents of the most recent generation. Here. Or take a typical discussion on the matter:

"The UK data also confirm the skill bias in African migration (Table 6 shows it only for black African migrants, so the picture is not altered by white or Asian migrants). More than twice as many black African migrants have college education or above as the native-born UK population, and half as many are unskilled, roughly the same for both males and females. [Pic Attached]"

BA tend to come to the UK with higher absolute rates of higher qualifications i.e., more college degrees than possessed by the White natives. Now consider:

(a) If Black Africa (the region) had the same average skill level as the UK, one might expect BA kids to do better than native Whites, given that the parents have higher qualifications.
(b) If Black Africa (the region) had much lower skill levels, one might expect BA kids to do even better than in the case of (a), because the parents would be all the more elite relative to the BA African norm.

You basically have a situation in which the children of relatively elite Africans perform on academic measures like average native UK Whites. If these children are performing according to their potential and if etc., then this is consistent with a partial genetic hypothesis. So, again, we have a major confound.

As for "regression to the mean", I've largely changed my mind on that. Regression to the mean is the result of parents not transmitting the portion of their trait values due to non-shared environmental factors and statistical error. For IQ at least, it turns out that this portion of variance has little to no predictive value. When it comes to immigrant selection, this portion would effectively be screened out, since immigrants are not being selected on the basis of IQ scores but on the basis of some predictors. In short, I wouldn't expect much regression in this case.

At what point do you think it will be safe to call the debate and run away? When BC erase the gap with white brits? They are within a trivial hairs length gap from other whites in GCSEs now. Possibly erased it last year. I mean the gap to white brits itself isn't big either. It takes a lot of evidence to build up scientific racialism but it doesn't take a lot to demolish it.

It's rather difficult to demolish "racialism", broadly constructed, because "racialism" need not be "racial hereditarianism". See here and here for conceptual discussion.

As for racial hereditarianism, the ease of demolishing this depends on the proposed magnitude of allelic risk, on the heritability, with respect to the specific trait in question, of the indexes, and on the potency of environmental factors.

The latter consideration is important. Imagine that racial hereditarianism with respect to African Americans in the US is 100% false. If so, environmental factors can be so potent as to cause a 1 SD difference in IQ despite the substantial heritability of measures. If they can be this potent to cause such a difference, then surely they can be potent enough to eliminate one, say in the UK.

So any environmental argument from the absence of differences in such and such regions -- but not in the majority of regions -- runs into a catch-22 of sorts. The argument assumes that differences in genetic potential must closely match with differences in outcomes. Thus, an absence of differences in outcomes is said to evidence an absence of differences in potential. Yet, this premise would be invalidated by the presence of large racial differences in some countries but not others. Thus, if it were shown that Blacks in the UK performed no less than White who had similar backgrounds, then, given the presence of differences elsewhere, it would be shown that a central interpretation of hereditarianism was incorrect. It wouldn't be shown that there were no race differences in genetic potential because the assumption that allowed for this inference would have just been falsified. In this case, we could not rely on uniformity of differences as evidence of genetic ones, so this would weaken an hereditarian argument, no doubt. But it would not "demolish" it or, for that matter, strengthen an environmental argument.

The issue then is more complex then you are making it out to be.

@Chuck.

"(a) If Black Africa (the region) had the same average skill level as the UK, one might expect BA kids to do better than native Whites, given that the parents have higher qualifications."

That looks highly likely the way its trending(technically they already do slightly better in GCSE A*-C). The other important part is that the environment was suppressing their kids even though the parents did well and it can be clearly seen in the way IQs start off low but increase and close the gap as they age and generationally. How much of that sort of suppression and lack of intervention is happening in Africa to both the lower and higher intellects? I mean the place is till a breeding ground for pathogens.

There are also biological explanations for why gaps could stagnate like in the US IE: epigenetics. It has been shown in tests on mammals that environmentally induced effects can last more than 4 generations even with the environment to reverse it. HBD people like to take epigenetics as a joke, but its serious. It actually adds chemicals to the genetic sequence, doesn't change the sequence of the sequence but it changes the chemical makeup of it and its hereditary. This sort of explanation was not there earlier on in the debate. It can explain the Minnesota adoption studies results, though that study was technically pathetic. It didn't even account for the parents IQ IIRC.

Still I will agree its not over yet and I am over simplifying, but to me it looks ominous even without using the black Caribs.

Correct me if I am wrong but I see higher and higher estimates for Africa since the debate began. Started off with average IQ in the low70s-80s to now mid 80s-90s(eg:Piffer).

If a small difference in average IQ can make a big difference then closing a large part of it will make an even bigger difference.

That looks highly likely the way its trending(technically they already do slightly better in GCSE A*-C). The other important part is that the environment was suppressing their kids even though the parents did well and it can be clearly seen in the way IQs start off low but increase and close the gap as they age and generationally.

Could you make a graph of Black African British GCSE scores for the last decade; that would give us a reference to interpret and predict score changes. As for depressing factors, I would imagine that language would be one; in this case, though, psychometric bias would initially be artificially depressing scores, which is consistent with a selection hypothesis. Imagine that Black African immigrant parents are relatively super selected. And that kids initially perform below average due to linguistic bias. But with time and generation that bias diminishes rapidly. And yet, in the end, BA kids only perform equal to or slightly better than White British kids. This would be consistent with a hereditarian hypothesis.

There are also biological explanations for why gaps could stagnate like in the US IE: epigenetics. HBD people like to take epigenetics as a joke, but its serious. This sort of explanation was not there earlier on in the debate. It can explain the Minnesota adoption studies results, though that study was technically pathetic. It didn't even account for the parents IQ IIRC.

Epigenetics is a silly explanation for IQ differences. I explained why here. You can directly measure the variance due to additive genetics (GCTA), shared environment (unrelated sibs reared together), nonshared environment (MZ twin discordance), and error (test-retest) and then add up the variance components and see how much is left over for other exotic explanations. Not much. Also, early racial theories (e.g., Buffon's) were often epigenetic, so this isn't in fact a new proposal; it proceeds DNA based ones.

As for the MTRAS, this was actually a technically sophisticated study. The authors even estimated the heritability of IQ for the various subsamples. As for biological parents, they used parental education as an index; IQs were known for adoptive parents. The problem, in my opinion, was mainly sample size. Were the results to be replicated several more time using larger samples I would be convinced that a shared (inter-generational) environmental explanation was untenable.

Correct me if I am wrong but I see higher and higher estimates for Africa since the debate began. Started off with average IQ in the low70s-80s to now mid 80s-90s(eg:Piffer).

Galton estimated an innate difference (in genius) of about 22 IQ equivalent points. Lynn has typically estimated that half of his said 30 phenotypic IQ point difference between Black Africans and White Europeans was owing to genes; at other times, though, he argued that there was a 20 point genetic gap. Jensen (and Rushton) estimated the 50 to 80% of the 15 point Black (80% Black African)-- White (98% White European) American difference was due to nature. Piffer recently estimated that there was a genetic difference equal to 10 points between White Europeans and Black Africans. So...

Galton (1800s): 22 IQ points
Lynn (1970s to present): 20 to 15 IQ points
Jensen (1970s to 2012): 10 to 15 points (corrected for % African Ancestry)
Piffer (present): 10 IQ points

So, yes, estimated differences appear to be decreasing.

If a small difference in average IQ can make a big difference then closing a large part of it will make an even bigger difference.

That assumes that genetic and environmental components of IQ have equal predictive validities, no?

Ok, but instead of merely trading replies, why don't we try to increase our background knowledge. See if you can find period information on Carib and African GCSE scores (relative to native White scores). Transform % differences into standardized ones. Graph it.

Still I will agree its not over yet and I am over simplifying, but to me it looks ominous even without using the black Caribs.

I find "negative HBD" to be just as interesting as the positive form. I don't see any reason for why racial groups which have evolved separately for tens of thousands of years needed necessarily to have developed roughly equal levels of general intelligence, do you? Surely, cultural environments have not been the same. The absence of large differences, then, tells us something important about this trait and the variability in it -- perhaps, for example, there is tightly balanced selection. This is interesting as it might tell us something about future limits of IQ enhancement.

Let me be clear: I appreciate your skepticism. More of this is both needed and deserving. It would be more helpful to me, though, if you could put together some statistical arguments; perhaps try to improve on this one, by addressing some of my criticisms of it - for example, concerning the degree of migrant selection, the utility of GCSE as a measure of g, the degree of regression to the mean, etc. and by updating the score (non) gaps.

Galton (1800s): 22 IQ points
Lynn (1970s to present): 20 to 15 IQ points
Jensen (1970s to 2012): 10 to 15 points (corrected for % African Ancestry)
Piffer (present): 10 IQ points

Jensen in The g Factor said it clearly that Galton's measure of IQ was supposed to be a pure measure of g. It wasn't IQ test, but chronometric test, the kind that Jensen has always recommended, in place of paper-pencil test. Jensen (2003, p164) in "Galton's legacy to research on intelligence" repeats the argument he made in The g Factor. When the IQ test (paper-pencil) has g-loading of 1, the expected B-W IQ gap was close to Galton's expectation : 20 points (and not 22 points). Furthermore, both Galton and Jensen weren't talking about genotypic differences. So, the comparison with Piffer is invalid.

Galton’s notion that the average White–Black difference in general ability, represented as Spearman’s g, is 1·33 SD can now be evaluated with all of the data available in 1998, based on 149 psychometric tests given to samples totalling 43,892 Blacks and 243,009 Whites. The regression of the White–Black difference (in SD units) can be extrapolated on tests’ g loadings to the point at which a hypothetical test would have the maximum possible g loading of 1·00. On such a hypothetical pure measure of g, the White–Black difference would be 1·31 SD, or equivalent to 20 IQ points (Jensen, 1998a, pp. 376–379). This is the best estimate we have of the White–Black difference in psychometric g. One may wonder if the proximity of Galton’s conjecture of 1·33 SD to the present finding of 1·31 SD represents an amazing perspicacity or an extraordinarily lucky surmise.

For Lynn, it was probably 20 points difference, because he expected the genotypic african IQ to be 80, given that nutrition generally rises IQ by 10 points (he said that in the book Race Differences in Intelligence: An Evolutionary Hypothesis, if my memory is correct). However, as Wicherts pointed out in a series of papers in response to Lynn, the african IQ is probably under-estimated because of the sub-optimal condition under which the africans took the IQ tests. Sometimes, the administrators reported that the africans were "bewildered" as they didn't understand what they were supposed to do. Some have problems holding a pen, etc. So, the 70 IQ estimate shouldn't be trusted, neither is the 80 genotypic estimates proposed by Lynn.

With regard to non-additivity, I can suggest this paper here. That may be relevant.

It takes a lot of evidence to build up scientific racialism but it doesn't take a lot to demolish it.

This is the kind of silly arguments made by the egalitarians you are talking about. I read it every time. They look at the BW gap; if it decreases, it must be an environmental cause. If it vanishes, it must be an environmental cause and the hereditarian hypothesis is invalidated. As I said to you before, the magnitude of the gap is irrelevant if you can't explain it with your preferred theory. If, say, during the time that the BW gap has diminished, the racial differences in all indices of environments have stayed the same or have increased, would you see it as a validation of the environmental hypothesis ? You shouldn't.

The correct thinking :

Draw a theory (logic, principle) capable of explaining the data.

The incorrect thinking :

Look the data to infer a theory.

the expected B-W IQ gap was close to Galton's expectation : 20 points (and not 22 points). Furthermore, both Galton and Jensen weren't talking about genotypic differences. So, the comparison with Piffer is invalid.

In hereditary genius, Galton notes:

"First, the negro race has occasionally, but very rarely, produced such men as Toussaint l'Ouverture, who are of our class F; that is to say, its X, or its total classes above G, appear to correspond with our F, showing a difference of not less than two grades between the black and white races, and it may be more"

Jensen (1973 p. 70) noted that 2 grades would be about 20.85 points or 1.39 SD. I see what I did: I computed the number for 1 grade and then mentally rounded it up to 11 points and then doubled it to get 22. I don't know why Jensen (1998) gives a different number. I would trust the earlier Jensen.

This doesn't tell us what Jensen's estimate of the genetic component of the differences was, though.

So, the 70 IQ estimate shouldn't be trusted, neither is the 80 genotypic estimates proposed by Lynn.

You are correct: "My reply: In my book Race Differences in Intelligence (2006, p.70), I estimated that sub-Saharan Africans have a phenotypic IQ of 67 and a genotypic IQ of 80. Thus, I estimated that the adverse environment reduces their IQ by 13 IQ points and that genetic factors reduce their IQ by 20 IQ points, compared with Europeans. Thus, 43 percent of the low IQ in sub-Saharan Africa is attributable to adverse environments and the remaining 57 percent is attributable to genetic factors (Lynn, 2012)."

With regard to non-additivity, I can suggest this paper here. That may be relevant.

It would depend on which specific trait we are talking about. "IQ", for example, can refer to both "general intelligence" and to "cognitive related ability as measured by IQ tests". Differences due to psychometric bias insofar as this is cognitively related ( i.e., not having been taught how to count) can reasonably be said to be "cognitive related ability" (despite not being general ability). Thus, purely additive g differences could lead to the construction of environments (e.g., schools) which, on the population level, lead to lower levels of "cognitive related ability" in access of g differences.

Regression to the mean involves two parts, 1) statistical regression due to imperfect measurement, 2) regression due to lower than unity heritability. These are conceptually different...For people selected on some g-correlate like educational attainment or income, this would also select them on g as well as other traits important for that trait, as well as environmental factors and luck. Luck is not heritable, so there would be regression in the children both due to the non-perfect heritability of ed. attain. (which is perhaps 30-50%), and the indirectly selected for g. This does not require the U2 part of g to be predictive (or "real").

(a) The shared environmental component is almost by definition intergenerationally transmitted, so why would one regress due to this.

(c) If the non-shared +error component is non-predictive, as I suggest, selecting on the basis of predictors will not select on these components.

Shared environment also has regression towards the mean.

Imagine that trait X is 100% due to shared environment; what degree is gen2 supposed to regress? Why? Shared environment is sometimes defined in terms of the parent-offspring correlation and non-genetic inter-generational transmission. So it isn't obvious to me why this would be considered a "luck factor".

Galton (1800s): 22 IQ points
Lynn (1970s to present): 20 to 15 IQ points
Jensen (1970s to 2012): 10 to 15 points (corrected for % African Ancestry)
Piffer (present): 10 IQ points

So, yes, estimated differences appear to be decreasing.

That was an estimate based on 1000 Genomes, phase 1 data comprising a small sample. The phase 3 has a bigger sample (N=26). Moreover, it was not based on regression analysis but simply using the White-East Asian difference as a reference.
I factor analyzed the 4 IQ increasing alleles using the new 1KG release. Factor scores and regression results are downloadable from here (https://docs.google.com/spreadsheets/d/1XDXK9MmhmQYeI3lr1mVtveGhPcDaTvbT0Ch1jS9EsxI/edit?usp=sharing).
The correlation between the factor scores extracted from the 4 IQ increasing alleles and Lynn and Vanhanen 2012 IQ is 0.9.
I ran two regression. First regression I used all the available IQs. In the second regression I used only IQs from developed countries (to eliminate the confouding effect of socioeconomic/environmental disparities). This left 9 cases, but the correlation between genetic factor scores and measured average IQs was even better (r=0.98), possibly because the average IQ of developed countries more closely mirrors the genotypic potential.
Col. G (yellow) shows the predicted unstandardized scores based on regressing IQ on the genetic factor, using only developed countries, adjusted as Greenwich IQ (UK set to 100).
This gives IQs for Blacks around 81-84 (Afr.Caribbean: 83.3; US Black: 83.7; Esan Nigeria 81.8; Gambian 81.9; Luhya Kenya 81.2; Mende Sierra Leone 83.5; Yoruba 81.7.
Considering only Native blacks (from Africa), the average genotypic IQ is 82, about 17 IQ POINTS LOWER THAN THE NORTH EUROPEAN IQ (99), AS AVERAGE OF BRITISH, UTAH WHITES AND FINNS.
Hispanics genotypic IQ is around 93 (Mexican 93.5, Puerto Rican 93.4) similar to the South Asian IQ (Bengali 91.3, Indian Telegu 93.5, Punjabi Pakistan 94.8).
Odd result is the very high IQ of Vietnam(105.9), which indicates that Vietnam's IQ could be depressed by bad environment.
Chinese and Japanese IQ are a bit lower than Rushton or Lynn's estimates (103-104) but it still has top position and is more similar to the IQ observed in Chinese Americans which is around 103. Probably a case for Chinese culture/writing system improving the IQs of Chinese Natives vs Chinese Americans. Otherwise, given such a small difference it could simply be random noise.

If you cite this, please cite Piffer and a link to this forum thread and google docs url.

Piffer also, in the above link estimates UK Telegu IQ at 88 , UK Sri Lankan at 81, Baghdad Bengali at 85, and Texas Gujarati at 93,. These seem questionable considering (significantly higher) South Asian IQ and gsce performance in the UK, and socioeconomic levels and known IQ data(e.g.:Richwine) from the US. I believe immigrant South Asian Pisa scores are also higher, though low in India.

Piffer also, in the above link estimates UK Telegu IQ at 88 , UK Sri Lankan at 81, Baghdad Bengali at 85, and Texas Gujarati at 93,. These seem questionable considering (significantly higher) South Asian IQ and gsce performance in the UK, and socioeconomic levels and known IQ data(e.g.:Richwine) from the US. I believe immigrant South Asian Pisa scores are also higher, though low in India.

Correction; I misread the spreadsheet, which gives some higher S. Asian estimates.
Still, Immigrant S, Asian scores esp. the Texas Gujarati do seem low in light of Indian (1st and second generation) performance in the US esp. given immigrant selection.

Piffer also, in the above link estimates UK Telegu IQ at 88 , UK Sri Lankan at 81, Baghdad Bengali at 85, and Texas Gujarati at 93,. These seem questionable considering (significantly higher) South Asian IQ and gsce performance in the UK, and socioeconomic levels and known IQ data(e.g.:Richwine) from the US. I believe immigrant South Asian Pisa scores are also higher, though low in India.

Correction; I misread the spreadsheet, which gives some higher S. Asian estimates.
Still, Immigrant S, Asian scores esp. the Texas Gujarati do seem low in light of Indian (1st and second generation) performance in the US esp. given immigrant selection.

A genotypic IQ of 97 for Gujarati Texas Indians is low? It's much higher than the average IQ of India estimate by Lynn (around 82).

Piffer also, in the above link estimates UK Telegu IQ at 88 , UK Sri Lankan at 81, Baghdad Bengali at 85, and Texas Gujarati at 93,. These seem questionable considering (significantly higher) South Asian IQ and gsce performance in the UK, and socioeconomic levels and known IQ data(e.g.:Richwine) from the US. I believe immigrant South Asian Pisa scores are also higher, though low in India.

Correction; I misread the spreadsheet, which gives some higher S. Asian estimates.
Still, Immigrant S, Asian scores esp. the Texas Gujarati do seem low in light of Indian (1st and second generation) performance in the US esp. given immigrant selection.

A genotypic IQ of 97 for Gujarati Texas Indians is low? It's much higher than the average IQ of India estimate by Lynn (around 82).

I believe Richwine's second generation estimate was 115, and Indians in the UShave an above average SES and educational attainment (due in part to positive immigrant selection).

Piffer also, in the above link estimates UK Telegu IQ at 88 , UK Sri Lankan at 81, Baghdad Bengali at 85, and Texas Gujarati at 93,. These seem questionable considering (significantly higher) South Asian IQ and gsce performance in the UK, and socioeconomic levels and known IQ data(e.g.:Richwine) from the US. I believe immigrant South Asian Pisa scores are also higher, though low in India.

Correction; I misread the spreadsheet, which gives some higher S. Asian estimates.
Still, Immigrant S, Asian scores esp. the Texas Gujarati do seem low in light of Indian (1st and second generation) performance in the US esp. given immigrant selection.

A genotypic IQ of 97 for Gujarati Texas Indians is low? It's much higher than the average IQ of India estimate by Lynn (around 82).

I believe Richwine's second generation estimate was 115, and Indians in the US have an above average SES and educational attainment.

I doubt it's 115 based on real IQ tests but it's remarkable that the Gujarati get higher genotypic scores than other S.Asian groups.