According to scholar Arthur Jensen, on a scale where white Americans average IQ 100 (SD = 15), African Americans have a mean IQ of 85 (SD = 12).  The question is, what does this actually tell us about the number of extremely high IQ African Americans?  The question was discussed on the West Hunter blog back in June, where the blogger wondered how many African Americans qualify for Nation Merit recognition (somewhat equivalent to an IQ of 139+ as measured by the PSAT, though the exact cut-off fluctuates a bit from state to state):

Consider black Americans. If their mean IQ is one std lower than whites, while the width of their distribution is lower (12 points instead of 15) – then for them the NMSPQRT threshold  is 4.475 standard deviations. Instead of 1 in 200, the fraction of winners is less than 1 in 200,000.

This is too simple: IQ can’t be exactly Gaussian, blacks in the US are not perfectly homogeneous, etc.  But it does show the trend: such high scores are much, much, much rarer in groups with low mean scores. It’s also the case that any black kid with such a high score would get a far better offer from Harvard.

However our very own Lion of the Judah-Sphere wrote in the blog’s comment section:

Hate to brag— okay no, I don’t, I love bragging— I’m black and was a National Merit Finalist. Granted, it was in a southern state with a low cutoff score. Strangely enough, out of about a dozen people in my school who made it to the semifinalist level, three of them were black (me and two black females).

But given the stats cited above and other anecdotes, the blogger found Judah-sphere’s claim very had to believe.

A few people have asked me to write about the IQ standard deviation of different races, and I’m always reluctant because one thing I’ve noticed is that although blacks have a smaller SD than whites, there are a lot more high IQ blacks than their small SD would predict.   For example, in the book The Bell Curve, they attempted to estimate the number of blacks in America with IQs of 125+ and find that the actual frequency of high IQ blacks is much greater than the Gaussian curve would predict.  Note, they were using a scale where an IQ of 100 and an SD of 15 were equivalent to the mean and SD of all Americans, not just white Americans, which is why the reported IQs for blacks are a few points higher than Jensen’s stats:

…As of 1994 there were approximately 32.8 million blacks in America.  If the estimate is computed based on the mean IQ (86.7)  and standard deviation (12.4)  of blacks in the NLSY, a table of the normal distribution indicates that only 0.1 percent, or about 33,000, would have IQs of 125 or higher.  If one applies the observed distribution in the NLSY and asks what proportion of blacks are in the top five percent of the AFQT distribution (roughly corresponding to an IQ of 125), the result, 0.4 percent, implies the answer is about 131,000.  There are reasons to think that both estimates err in different directions.  We compromise with 100,000.

I wish they had elaborated on what reasons there are to think both estimates err in different directions!

On page 454 of The Bell Curve, they note that in 1993, “only 129 blacks, 234 Latinos nationwide had SAT verbal scores in the 700s–and these represented all-time highs–compared to 7,114 whites.”  They are referencing the older, much harder version of the SAT that existed before the re-centering of the scale circa 1995.

Back in 1993, there were probably only 3 million 17 year-olds in America (including perhaps 2.25 million whites and perhaps 360,000 blacks), and virtually 100% of the brilliant ones wrote the SAT.  Thus, if 7,114 out of 2.25 million white 17 year olds scored 700+ on the verbal SAT (one out of 316), then, on a scale where white America averages 100 with an SD of 15, this can be roughly equated to a deviation IQ of 141 (just over the national merit level).

Now, if on the same scale, U.S. blacks supposedly have a mean of 85 with an SD of 12, the Gaussian curve predicts that of the perhaps 360,000 African American 17-year-ods in 1993, only one of them should score 139, and not a single one should score 140+.  Instead 129 scored 141+.  How do we explain this?

Perhaps the old SAT is not normally distributed at the extremes.  I forced it to fit the bell curve among whites which meant 700+ was assigned an IQ of 141 ( based on the fact that only one in 316 whites corresponds to the +2.73 SD level on the normal curve), but in reality, when representative samples of all U.S. white teenagers (not just the college bound segment) took the old verbal SAT as part of norming studies, a score of 700+ was actually over +3 SD from the mean because the true distribution is not perfectly normal, and departure from normality likely gets worse the more you deviate from the mean of your race.

I suspect this problem is largely limited to highly crystallized tests like the SAT and AFQT.  Because these tests are measuring cultural learning in addition to biological ability, there’s no reason to expect them to perfectly fit the Gaussian curve, which is largely limited to biological distributions.  Although it’s sometimes suggested that culture loaded tests are more heritable than culture reduced ones, several studies have found the opposite.  And of course there’s more to biology than just genes.

A second possible explanation was perhaps alluded to by the West Hunter blog: blacks in the US are not perfectly homogeneous.  For example, the brilliant Steve Sailer notes that roughly 10% of U.S. “blacks” are of predominantly non-black ancestry.  So even though African Americans as a whole might have a mean IQ of 85 and an SD of 12, these aggregate statistics obscure the fact that 10% of African Americans are mostly white at the genetic level, and thus probably have a mean around 93 and an SD around 14.  This hyper-hybridized subset might have their own bell curve, which might partly explain the unexpectedly high number of brilliant blacks.

A third possibility is these high scoring blacks are not just hyper-hybridized black Americans but rather the best and brightest from Africa, taking the SAT as foreign students or as children of the most elite immigrants.

A fourth possibility, which I find unlikely, is that a century of statistics are wrong, and that the African American mean and SD are not 85 and 12, but rather 78 and 15+.  Before they died, scholars Jensen and Rushton speculated that virtually all testing of U.S. blacks ever done excluded a large underclass of blacks.  These would probably be the most dark-skinned Negroid featured blacks in the heart of the ghetto, in schools no white psychologist would dare give an IQ test, but if they would, they might find scores of only 70 or even 50, which would drag the overall U.S. black mean down to 78, while expanding the SD enough to explain high IQ blacks at the opposite extreme.

A final possible explanation for the large number of brilliant blacks is that the general U.S. black mean and SD is artificially lowered by bad environment, and only the best and brightest blacks are largely privileged enough to overcome the bad environment and reach their genetic potential.  As an HBDer, I find this unlikely, but possible.