Blacks, IQ, and Color in the NLSY97

[note: JL has pointed out that subjective measures of skin color, as used in this data set, are much more unreliable than I initially conjectured. Apparently there’s also a race by interviewer effect (Hill, 2002). Contrary to what some may conclude, this added unreliability means that the true associations between color and IQ is higher than found. In the same way, the unreliability of self-identified race as an index of true biogeographic-ancestry also means that the true association is higher — i.e., the found correlations need to be corrected upward.]

Continuing with my most recent line of investigation, I looked at the association between skin color and aptitude tests in the NLSY97. You can read a description of that study here. It’s a follow up to the NLSY79 which, if you read the BC, you should be familiar with. The NLSY97 participants took a number of aptitude tests. Out of them, I selected the SAT verbal and qaunt, the Armed Services Vocational Aptitude Battery verbal and quant, and the Peabody Individual Achievement Test quant. The color-score correlation of 0.12 found for the ASVAB is close to the mean of all correlations. The correlations, again, are just those between color and scores within the self-identifying Black population. As I couldn’t find the relevant variables, I was unable to separate out immigrant Blacks from native Blacks; when I figure out how to do that — so as to follow up with a previous investigation, I will update this post. I didn’t feel like converting the ASVAB scores into IQ scores. For reference, the upper and lower 1/5th of the distribution differ by about 0.4 SD. (The scores are percentiles x 1000.)

Using the ASVAB scores as our index of IQ, this study puts the N-weighted correlation at 0.15 (N= 3694). (Ya, I actually look up those dusty old studies.)

Herskovits (1926)/r=0.17/n=115
Klineberg (1928)/r=0.12/n=139
Peterson and Lanier (1929)/r=0.18/n=83
Peterson and Lanier (1929)/r=0.3/n=75
Scarr et al. (1977)/r=0.155/n=288
Lynn (2002)/r=0.17/n=430
NLSY97 (unpublished)/r=0.12/1433
ADD Health (unpublished)/r=0.17/n=1131

And the average Cohen’s d between the upper and lower 4rths of the spectrum is about 0.5 n = >6,000 (Shuey’s pre-60’s data here)

Feguson (1919)/d= about 0.7/n=657
Feguson (1919)/d= about .9 SD/n=667
Kock and Simmons (1926)/d= about 0.15/n=1078
Klineberg (1928)/d= about 0.15/n=200
Young (1929)/ d= about 0.8 and 0.33/n=277
Peterson and Lanier (1929)/d = about 0.66/n=83
Peterson and Lanier (1929)/d= about 0.2 SD/m=83
Bruce (1940)/d=about 0.25/n=72
Codwell (1947)/d= about 0.33/n=480
Lynn (2002)/d= about 0.5/n=430
NLSY97 (unpublished)/d= about 0.4/n=1433
ADD Health (unpublished)/d=about 0.5/n=1131

It’s interesting that both the found correlations and mean differences show a good deal of cross temporal consistency.

I discussed what the predicted correlation for a genetic hypothesis would be elsewhere. It would be a function of a) the reliability of the measures of IQ and skin color (maybe 0.9 and ?), b) the correlation between color and African ancestry in the US Black population (about 0.45), c) and the correlation between IQ and White ancestry in the Black population, given the 1) restriction of range in the distribution of White ancestry in the Black population (? maybe 0.5), 2) the within population heritability of IQ (0.4 to 0.8, age depending), and 3) some proposed between population heritability (0.5 to 1). Contra Nisbett, for any hypothesis it would be below 0.20.

The fact that there is a correlation, of course, does not reduce the environmental-genetic uncertainty much, as the correlation could always be explained environmentally. Though, given that between family differences explain little variance in IQ, such explanations are constrained. (For example, Guo and Stearns (2002) found that between family effects explained only 17% of Black IQ variability in the ADD health data, data in which there was roughly a 1/2 SD difference between lighter and darker Black kids (ages ranging approximately from 12 to 18).) If not between family effects, then what? “Colorism” seems to be an obvious possibility — but it must work through some environmental mechanisms.

Whatever the case, from a Popperian perspective, a genetic hypothesis just survived another of my falsification attempts. As they say, that which does not falsify, makes stronger.

This entry was posted in Uncategorized. Bookmark the permalink.

6 Responses to Blacks, IQ, and Color in the NLSY97

  1. JL says:

    Nice work. 0.9 sounds awfully high for skin color measurement reliability though.

  2. 猛虎 says:

    Jensen (Educability and group differences) tells us that the highest expected correlation between IQ and skin color would be 0.2 (SQRT (0.16)(0.25)). 0.16 being the reliability of skin color as an index of white ancestry, and 0.25 the reliability of IQ as an index of white ancestry. I don’t know what is the magnitude of the relationship between IQ and ancestry. Have you an idea ?

    • Chuck says:

      Sort of. I have already discussed this issue. Refer, for example, here:

      “a. According to recent analyses, the mean African admixture is 20% and the standard deviation of admixture is 15%. According to Zakharia, et al. (2009):

      “Numerous studies have estimated the rate of European admixture in African Americans; these studies have documented average admixture rates in the range of 10% to 20%, with some regional variation, but also with substantial variation among individuals [1]. For example, the largest study of African Americans to date, based on autosomal short tandem repeat (STR) markers, found an average of 14% European ancestry with a standard deviation of approximately 10%, and a range of near 0 to 65% [1], whereas another study based on ancestry informative markers (AIMs) found an average of 17.7% European ancestry with a standard deviation of 15.0% [2].…
      …These results were confirmed in the estimation of IA by using the program frappe (also in Figure 1). The amount of European ancestry shows considerable variation, with an average (± SD) of 21.9% ± 12.2%, and a range of 0 to 72% (Table 1).”

      Based on this we can calculate an expected IQ-ancestry correlation.

      b. One interpretation of a correlation coefficient is: amount of change in x, change y or, in this case, the amount of change in admixture per change in genetically conditioned test score. 
In this case the genetically conditioned difference between Blacks and White would be 0.75 SD, since we are proposing that 75% of the gap is genetic; the ancestry difference would be 5.3 SD, which is the number of SDs separating Blacks who are 20% White and Whites, given that 1 SD of admixture equals 15% Whiteness ((100-20)/15=5.3). The correlation between test scores and genotypic ancestry, would then be 0.75/5.3 or 0.14.

      c. This would be the correlation for an index that had perfect reliability. Correcting for the unreliability (see 1b), the the correlation between IQ and the index would be 0.085”

    • Chuck says:

      One problem with Jensen’s analysis (in Educability and group differences) is that he didn’t take into account range restriction in ancestry in the African American population. See if you can find this article: Jensen, 1981b. Obstacles, problems, and pitfalls in differential psychology. In S. Scarr (Ed.), Race, social class, and individual differences in IQ.

    • Chuck says:

      “I don’t know what is the magnitude of the relationship between IQ and ancestry. Have you an idea ?”

      If I knew what the r was between IQ and African ancestry in the Black population, would I be blogging about ‘possible’ genetic race differences? Probably not.

      To calculate a hereditarian hypothesis’s prediction you need an estimate of the standard deviation of admixture in the Black population under consideration and an estimate of the correlation between color and African Ancestry. See below for example:

      “To determine this, we need to ascertain the average and standard deviation of admixture in the population. In table 1 of Cheng et al. 2012, we see that the African admixture in 84% and we can deduce that the standard deviation is 8.5, based on the interquartile range. From the above, we can determine that Blacks in this sample differ from Whites by 9.88 SD of African admixture, based on the sample admixture SD, e.g., (84-0)/8.5 SD. Were genetic IQ d to equal 1, then a shift in 1 genetic IQ SD would be associated with a shift in 9.88 SD of African admixture (granting such and such statistical assumptions). The predicted genetic IQ, ancestry correlation would then be 0.1. And 0.1 x 8.5% would give us the predicted % African Admixture difference between sub-samples that deviate by 1 standardized unit in genotypic IQ. Since the extremes in SES should differ by about 2 SD in IQ (at least for education and occupation), a genetic hypothesis would predict a 0.2 SD difference in ancestry between the extreme SES categories or a difference of less than 2%….”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s