Spearman’s Hypothesis vindicated

Meng Hu pointed me to the following paper:

te Nijenhuis, J., & van der Flier, H., Is the Flynn effect on g ?: A meta-analysis, Intelligence (2013),

A psychometric meta-analytic correlation between the Flynn Effect and g of -0.38 was found. Recall that in 2000, James Flynn argued that the Flynn Effect was a Jensen Effect (1). And that as late as 2008 the dishonest (2) Richard Nisbett was citing Flynn’s dated findings.

The results were not surprising at all. What was was to stumble on Joep Dragt’s incredible master thesis: “Causes of group differences studied with the method of correlated vectors: A psychometric meta-analysis of Spearman’s hypothesis.” The following differences were examined: the US Black-White difference, the US Hispanic-White differences, the US American Indian-White difference, the Dutch Native-Immigrant differences, the South African White-Asian Indian-African differences, and the Serbian Majority-Roma differences. Quote:

In this study we collected the complete empirical literature and conducted a meta-analysis. The findings clearly show that the true correlation between mean group differences and g loadings is strong: a correlation of .71 based on the Wechsler tests as a reference for the restriction of range correction and a correlation of .91 when the Dutch GATB was taken as a reference. Probably the GATB is a better reference, as its variance in g loadings is closer to the variance in g loadings from a theoretically optimal test battery, measuring all broad abilities of Carroll’s (1993) model. Also, the correlations between group differences and g loadings do not differ by group; some out comes are even virtually identical….

….Recent psychometric meta-analyses have clearly shown that g loadings correlate highly with measures of heritability. te Nijenhuis and Grimen (2007) show that g loadings of subtests correlate perfectly with these subtests’ heritability coefficients. Moreover, te Nijenhuis and Franssen (2010) show that inbreeding depression correlates .85 with g loadings. This strongly suggests that g loadings and heritability coefficients may be interchangeable. This in turn suggests that the high correlation between g loadings and group differences could imply that mean group differences have a substantial genetic component. However, this is not necessarily the case, as the score patterns of biological factors, such as better nutrition and better health care for pregnant women, may mimic the score pattern of the heritability coefficient. At the present, these effects are impossible to disentangle, as all the available research is correlational and not experimental….

…IQ tests are important instruments for selection and placement in work and educational settings, and there are large differences in mean IQ scores between groups. The present study makes a strong empirical contribution to the important discussion about the nature of group differences in intelligence. It strongly suggests that IQ batteries are not culturally biased, apart from a small effect for language bias when there is a substantial number of subtest with a strong verbal component in one or more specific tests. This provides strong support for the validity of these tests because group differences in intelligence are not attributable to cultural bias. It appears that with the exception of language-biased subtests, one can confidently use IQ batteries in work and educational settings….

Huh…there were also several interesting references:

(1) te Nijenhuis, J., & Franssen, D. B. (2010). What is the significance of test-score differences? Five psychometric meta-analyses on g loadings and IQ scores: The relation of inbreeding, visual impairment, schizophrenia, epilepsy, and giftedness with general intelligence. Unpublished manuscript, University of Amsterdam, Amsterdam, NL.

(2) te Nijenhuis, J., & Jongeneel-Grimen, B. (2007).Can people get smarter: Three psychometric meta analyses and three exploratory meta-analyses on g loadings and IQ scores. Unpublished manuscript, University of Amsterdam, Amsterdam, NL.

(3) te Nijenhuis, J., de Pater, I. E., van Bloois, R., & Geutjes, L.L. (2009).Two psychometric meta-analyses and three exploratory meta-analyses on g loadings and IQ scores: The relation of giftedness, mental retardation, alcohol and cocaine abuse, and depression with general intelligence. Unpublished manuscript, University of Amsterdam, Amsterdam, NL.

Of these (2) seems really interesting. One of the analyses concerns g-loadings and heritability coefficients. Presumably the other two look at g-loadings and some clearly environmental influences. If so, such studies would be of particular interest. It’s clear that what I call causal biological influences manifest as g(+) differences. So if you see a g(+) differences it’s plausibly largely causally biological. It’s not as clear — to me — that causal cultural differences necessarily manifest as g(-) differences. To be able to confidently draw this conclusion you would need many more studies looking at the association between g-loading and known cultural causes e.g., schooling, intervention programs, adoption, etc. There is no shortage of data — but, for some reason, this half of the equation has not been well investigates. As a result, seeing a strong g(+) difference doesn’t allow one to rule out culture causes in general. One can only rule out the specific subset of cultural causes that have been well explored and shown to be g(-) e.g., Flynn Effect like causes, Protestant Effect like causes, Learning Potential like causes, Test-Practice like causes, etc. Personally, I would love to see a MCV study of the correlation between unrelated siblings reared together (UST)– since this would establish that cultural causal differences of the shared environmental type are anti-Jensen Effects. Some would argue that such a study would be redundant. After all, MZA correlations are highly g(+); this implies that the UST correlations would be highly g(-), since unrelated siblings reared together are the inverse, from a behavioral genetic perspective, of monozygotic twins reared apart,. But it’s not clear to me that this is the case. Were it, it would be virtually indisputable that race differences in the US are largely genetically conditioned, since differences can’t be environmentally causally biological (as discussed a while ago) and since differences must be due to some combination of shared environment and additive genetics (hence regression to the mean) — yet it’s not virtually indisputable…ergo…a Jensen Effect on the MZA correlation must not necessarily imply an anti-Jensen Effect on the UST correlation. (Or am I missing some important consideration? (3)) Indeed, if someone were to show the latter I would write up paper on the logical induction and try to get it published in the most hereditarian hostile journal of philosophy of biology that I could find.

(1) Flynn, J. R. (2000). IQ gains, WISC subtests and fluid g: g theory and the relevance of Spearman’s hypothesis to race. In G. R. Bock, J. A. Goode, & K. Webb (Eds.), The nature of intelligence: The Novartis Foundation symposium (pp. 202 −227). New York: Wiley

(2) As a taste of Nisbett’s deceit, he claims, recently:

The Bell Curve encouraged the assumption that a significant portion of the 15-point IQ difference between blacks and whites that existed in the early 1990s might be due to genetics. The authors’ treatment of the evidence on this question was biased in the extreme, devoting a great deal of space to the single study that gave significant support to the genetic interpretation, and devoting little space to the considerable amount of direct evidence indicating that the IQ difference is not due to genetics. This evidence stems from the fact that the “black” gene pool in the United States contains a large amount of European genes.Almost all the research indicates no higher IQs for blacks with a significant degree of European heritage than for those with a lesser degree. One of the most telling studies is an adoption study examining the IQs of black and mixed-race children who were adopted early…(Nisbett, R. E. (2013). Schooling Makes You Smarter. AMERICAN EdUCAToR, 11.)

For one, there was no evidence of “extreme bias” in H & M’s book. For another, H & M rightfully discussed the Minnesota Transracial Adoption Study and neglected to discuss the study which Nisbett is fond of citing — Moore (1986) — because the former, but not the latter, had the statistical power to test a genetic hypothesis of the magnitude proposed (i.e., a genotypic difference of 1 SD). Moreover, Nisbett’s claim concerning the direct evidence is false. The majority of studies showed and continue to show a relationship between White Admixture and IQ in the African American population. Many of these studies were reviewed here and in sections O through Q here. (Adding to these studies, Meng Hu and I have found significant correlations between IQ and reported ancestry and IQ and skin color in two large relatively recent nationally representative studies, e.g., here.)
(3) So…

a. either the gap is causal cultural or causal biological.
if the gap is causal biological, it is either causal environmental-biological or causal genetic-biological.
the gap can not be causal environmental-biological (for reasons discussed before).
ergo, the gap must be either causal cultural or causal genetic-biological.
b. either the B/W gap is shared environmental or additive genetic (this follows from the logic of differential regression to the mean, discussed before)
c. either shared environmental influences can produce g(+) gaps or they can not.
the gap is decidedly g(+).
ergo, either the gap is not shared environmental or the gap is shared environmental + shared environmental influences can produce g(+) differences.

this reduces down to…
d. either the gap is additive genetic or the gap is shared environmental and shared environmental influences can produce g(+) gaps
e. either the debate over cause (wholly environmental versus some genetic) is over or it is not
it’s not over.
f. the Jensen effect on the MZA correlation either necessitates an anti-Jensen effect on the URT correlation or not.
if it does, since URT is a direct measure of the influence of shared environment, shared environmental influence must induce an anti-Jensen effect.
ergo, by d, the B/W would obviously have to be genetic.
h. the B/W gap is not obviously genetic. the inductive evidence of this is that the debate is still going on. And that I’m not compelled.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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