Admixture studies discussed by Shuey (1966)

I summarized the 16 admixture studies discussed by Shuey in “The Testing of Negro Intelligence (1966)” and updated my discussion of these:

“To support his claim, Nisbett sidesteps 7 studies that showed a large relation between indexes of white admixture and IQ — Feguson (1919), Peterson and Lanier (1929), Young (1929), Tanser (1939), Tanser (1941), Codwell (1947); Grinder et al (1964) –and 4 studies that showed a moderate to small relation –Davenport (1928), Klineberg (1928), Peterson and Lanier (1929), Bruce (1940). He draws attention, instead, to the low correlations found between indexes of admixture and IQ in 3 studies – Herskovits (1926), Peterson and Lanier (1929), and Klineberg (1928) — and argues that the low correlations stand as evidence against the genetic hypothesis….

In short, the findings of the three studies that Nisbett highlights provide no evidence against the genetic hypothesis; rather, they are consistent with it. Now, turning to the whole set, of the 16 studies predating 1965, 13 showed a positive relation between indexes of white admixture and IQ (7 large, 6 moderate to slight), 2 samples show no such relation, and 1 was equivocal. Of the 2 studies that showed no relation, in one — Kock and Simmons (1926) – light colored blacks nonetheless outscored darker colored backs. In all, the findings are not inconsistent with a partial genetic hypothesis.”

This entry was posted in Uncategorized. Bookmark the permalink.

13 Responses to Admixture studies discussed by Shuey (1966)

  1. Chuck says:

    That doesn’t establish that between group differences have a genetic basis; the paper just provides more support for what we already know: within populations IQ is highly heritable. If you want the hypothesis that between group differences are heritable to be tested, you have to push for such testing.

  2. nikcrit says:

    I have a quick question, though it’s off-topic:

    I’m wondering if it would be valid to surmise an ‘aggregate IQ’ for a given public school district; the sampling sources would likely be a representative sample of kids who outright took IQ tests (WAIS); or, a much larger sample of seniors who took SAT’s and others, and then used studies that assessed and translated their avg. score levels into avg. IQ levels, etc.

    I’m wondering if anything pre-existing comes to mind; i haven’t begun to research this even casually.

  3. JL says:

    Chuck, have you read this study? It uses Add Health data to show that the IQ gap between blacks with white mothers (BWM) and whites is 3 points while the b-w gap in the same data is 12.7 points, i.e. the white-BWM gap is only 24% of the b-w gap. The gaps between whites and BWMs in GPA, math, and science are from 32-46% of the respective b-w gaps. The white-BWM gaps are generally not statistically significant, so the authors claim that the b-w gap is environmental and caused by white mothers providing their children with a better childhood environment. However, the direction of the gaps is consistent, so the non-significance is probably due to a low sample size for BWMs (127 for IQ, 87 for school outcomes). IQ was tested in teenagers (Wave 1). (There were also no significant differences in college graduation rate and income between whites and BWMs, although the sample sizes there were even smaller, and apparently the findings hold only for men, not women.)

    In general, the paper is suffused with the sociological fallacy, with the authors suggesting that the shared environment is what causes individual differences, the results of behavior genetics be damned. The only allusion to genes in the entire paper is this bold footnote on p. 18: “As noted above, the fact that mixed race students look similar to whites on many outcomes suggests that race itself, i.e. genes, are not driving the observed racial differences. See Fryer and Levitt (2006) for more formal evidence that genes are not the cause of racial outcome gaps.”

    However, it is interesting that biracial blacks with white mothers perform somewhat better than one might expect. I think some other studies have suggested the same. How would you explain it? I obviously don’t buy the home environment explanation. If it’s some environmental factor, I’d say it’s the prenatal environment. Other explanations might be hybrid vigor (not likely), assortative mating (white mothers of biracial children are below average, but black fathers of biracial children are probably above average), or even a sampling error.

    • Chuck says:

      Ya, I read it. Similar results were found in an analysis of the Longitudinal Collaborative Study data by Willerman et al. (1974). (“Intellectual development of children from interracial matings: Performance in infancy and at 4 years.”) And environmentalists make much of that study. But such results were not found in the NSLY data analyzed by Gullickson (2004). (“Amalgamations, New and Old: The Stratification of America’s Mixed Black/White Population.”)

      What to make of it? As for the difference between biracials with white mothers and whites (N=127), a .2 SD difference was found in this study. The same data was analyzed by Harris and Thomas (2002) and a total biracial white gap of .33SD was found (N=128). (“The educational cost of being Multiracial: evidence from a National survey of Adolescence.”). (see also:
      Fryer et al., 2008 in “The plight of mixed race adolescents” who found a .3 gap.) Harris and Thomas (and Fryer) used self-classification and the authors of this paper used mothers race and interviewer classification. So the magnitude of the found gap seems to depend somewhat on the method of classification. As, Rowe (2002) found only a .17 SD gap based on interview assessment of race (N-116), it’s
      possible that the interview identification method minimized the gap somewhat. Unfortunately, the authors of this paper don’t report what the gap is, based on their methodology, between biracials with Black mothers and whites and between all biracials and whites, so it’s difficult to tell.

      As for the difference between biracials and whites, let’s use .25 as our magnitude for the sake of argument since this is the average difference found between BWMs and whites across the three nationally representative studies that we have data for (National Longitudinal Survey of Youth, Longitudinal Survey of Adolescent Health, and Longitudinal Collaborative Study). (For reference the difference in the studies between all biracials and whites is .38). This difference is equivalent to a Black-White difference of .5SD – assuming no assortative mating. (My guess is that there is such mating, but it would be interesting to look into this by comparing inter-racial Blacks-White marriages by social class). Now consider that the genetic hypothesis as formulated by Jensen (1969, 1973) 1998) proposes a .5 to.75 SD Blacks-White gap (as opposed to the 1sd gap commonly proposed on HBD blogs). The found gap is still at the lower end of those values. The N is only 300 across the three studies, so it’s difficult to draw firm conclusions. Other data exists — such as ECLS. Head Start studies — but, unfortunately neither of us have access to it. (NAEP is to begin using multiracial classification, so that will provide a large data source in the very near future.)

      The .25 is nonethless uncomfortably low. Separating out the biracials with black mothers might be questionable though. Just as the authors of this paper see Black mother rearing as
      a partial cause of the large b-W gap, we could see white mother rearing as the cause of the low biraicial-white gap. If
      mother rearing can artificially depress scores — presumably at younger ages — it can artificially raise scores. (What’s the g-loadedness of the Peabody Picture Vocabulary Test? That should tell us how malleable scores are.)

      Now, the final issue is the explanability of the difference. I concur with your point about the sociologist fallacy. As for the statement “See Fryer and Levitt (2006) for more formal evidence that genes are not the cause of racial outcome gaps,” in a more recent paper Fryer discusses — perhaps because I emailed him about the issue — how his 2006 data does not falsify the genetic hypothesis, (See: “Racial Inequality in the 21st Century: The Declining Significance of Discrimination.”)

      You know JL, I appreciate this comment. It lets me know that I’m not the only one digging through the research to find data that either falsifies or supports the genetic hypothesis — though I’m not doing it so much anymore. I trust you also write environmentalists academics to encourage them to test the hypothesis or correct their misinformation.
      What’s unfortunate about HBD is that there is virtually no one in academia investigating it. (Take a look how feeble Jensen is. His present state well characterizes intellectual HBD: This leaves it up to people outside of academia to write those in and push them to test the genetic hypothesis, which can be tested through simple structural equation modeling as Jensen (1998) and Rowe (2005) discussed if not genotypic analyzes. But most HBDers are uninterested in doing this.

      • Bill says:

        Shouldn’t you consider the confidence interval? If the white-biracial IQ difference is, say, 0.25 +/- 0.25, then there is no problem since 0.5 is in the confidence interval. The fact that 0.2 is not “significantly” different from zero sure seems to make it likely that 0.5 is in the CI.

      • JL says:

        Arcidiacono et al. argue that white mothers provide superior parenting to their kids even
        when comparing white high schools grads to black college grads. I’m skeptical that this
        is the reason for the small gap, because behavior genetics would suggest that the shared
        environment has little effect at age 16 (testing age in Add Health). Scarr’s adoption
        study showed that at 17 there was no such effect.

        I’m not sure if the Add Health BWMs
        were 16 on average at the time of testing. Their mothers are somewhat younger than other
        Add Health mothers. Also, if I’m not mistaken, most of those tested in Wave I were retested in Wave III, so it might be possible to check if the gap remains unchanged with age. Add Health has also collected information on skin color differences–do you know if someone has calculated the correlation between IQ and skin color in those data?

        The idea that blacks are worse parents than whites even after controlling for education and income is of course very Rushtonian. It suggests that blacks are low-investment parents compared to whites. It’s funny that they would argue this.

        I think the effect of the prenatal environment is underappreciated as a source of IQ variance. You could use surrogate mothers to study it.

        It’s true the anti-hereditarians are now winning arguments by default. Jensen et co. have failed to train up successors for themselves.There must be many psychologists who are closet race realists, but they prefer not to engage in public mud-wrestling about the topic. One is probably James Lee, whose review of Nisbett’s book is a masterpiece of refutation. He’s now involved with the BGI intelligence GWAS project. Proof for race realism may eventually come indirectly through such research.

  4. Kiwiguy says:

    ***What’s unfortunate about HBD is that there is virtually no one in academia investigating it.***

    You have to wonder what Rowe would have achieved if he hadn’t passed away at the age of 53 in early 2003. He seems to have been a particularly capable researcher who was interested in HBD topics.

  5. Chuck says:


    “Arcidiacono et al. argue that white mothers provide superior parenting to their kids even
    when comparing white high schools grads to black college grads. I’m skeptical that this
    is the reason for the small gap, because behavior genetics would suggest that the shared
    environment has little effect at age 16 (testing age in Add Health).”

    In general, behavior genetics does suggest this. But the heritability of the Peabody assessment, specifically, is only about .50 and the correlation between Peabody and verbal IQ is only about .65 (Jensen, the G-factor pg. 467). This leaves open a good deal of room for environmental influence.

    Imagine if the .50 enviromentality is divided evenly between shared and non-shared environmental effect, giving a shared environmentality of .25. To shift the IQ of a population of individuals 1x standard deviations, “only” 2x standard deviations of between family environmentality would be needed, right? (2x times (sqrt .25) =1x). It’s not implausible, then, that superior white mother rearing could then have some impact.

    • JL says:

      Jensen talks about the Peabody achievement tests, which are not the same as the Peabody Picture Vocabulary Test.

      This is what the test publisher says:

      The PPVT-III has an average correlation of .69 with the OWLS Listening Comprehension scale and .74 with the OWLS Oral Expression scale. Its correlations with measures of verbal ability are: .91 (WISC-III VIQ), .89 (KAIT Crystallized IQ), and .81 (K-BIT Vocabulary).

      However, Add Health used an earlier edition, the PPVT-R, more specifically an abbreviated version of the PPVT-R.

      Charles Murray on the PPVT-R:

      The Peabody picture vocabulary test specifically has shown median
      correlations of .62 and .64 with full-scale IQ as measured by the Stanford–Binet and the WISC respectively.

      The PPVT is not ideal for our purposes, because it’s not as g-loaded as tests with more diverse items.

  6. JL says:

    Here’s some information from Bouchard on the heritability of the PPVT in Add Health:

    Guo and Stearns (2002) made use of a sample of participants in the National Longitudinal Study of Adolescent Health who were in grades 7 to 12 and had completed the Peabody Picture Vocabulary Test. They did not provide an overall analysis of the influence of SES but rather report the influence of various components. In the model that includes three factors simultaneously and therefore allows a direct comparison of the influence of each component (Model 4, Table 8) we find the following results.

    1. For three levels of family income (16k but 100k) the heritabilities were .526, .572 and .586.

    2. For three levels of Mother’s education (less than high school, high school graduate, greater than high school) the heritabilities were .557, .527 and .480. These results are NOT in the direction predicted by the authors.

    3. For biological father absent vs. biological father present the heritabilities were .509 vs. .565.

    It seems likely a standard composite measure of SES (typically father occupation, fathers education and family income) would yield a very modest moderation of heritability. These findings do not replicate the much stronger effects reported by Turkheimer et al. (2003). It is not well known but in large scale studies (samples of thousands) father’s education is correlated as well (sometimes higher) with child’s IQ as the mother’s education (Table 6. Bouchard & Segal, 1985). This is of course due, in part, to the genetic correlation between level of parental education and IQ. Consequently information on the influence of father’s education and/or a composite of father and mother education would have been of interest. Guo and Stearns report a heritability of .576 for Blacks and .72 for whites. It should be noted that the sample sizes for some of the comparisons in this study are quite modest.

    (I don’t quite understand how the heritability for whites can be so much higher than for blacks when the sample is mostly white and the heritabilities for each SES group are much lower than .72 or even .576.)

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