Sunday, April 22, 2007
We previously reported that a measure of school achievement built from national test scores has a nearly perfect correlation with national IQ (at least in the range of scores tested). Subsequently, Lynn et al. (in press) published a very similar analysis:
This paper examines the relationship of the national IQs reported by Lynn & Vanhanen (2002, 2006) to national achievement in mathematics and science among 8th graders in 67 countries. The correlation between the two is 0.92 and is interpreted as establishing the validity of the national IQs. The correlation is so high that national IQs and educational achievement appear to be measures of the same construct. National differences in educational achievement are greater than differences in IQ, suggesting an amplifier effect such that national differences in IQs amplify differences in educational achievement. Controlling for national differences in IQ, slight inverse relationships of educational achievement are observed with political freedom, subjective well-being, income inequality, and GDP. However, public expenditure on education (as % of GDP) was not a significant predictor of differences in educational achievement.
The IQ's Corner blog has an interesting note about forthcoming commentary.
On a related note, recall that Templer & Arikawa (2006) reported a near perfect environmental correlation between national skin color and national IQ for old-world countries. An unfortunately confused commentary by Hunt & Sternberg accompanied the publication. They wrote: "We argue that the report by Templer and Arikawa contains misleading conclusions and is based upon faulty collection and analysis of data. The report fails to hold up for quality of data, statistical analysis, and the logic of science." The criticisms by Hunt & Sternberg are based largely on a misreading of Templer & Arikawa's methods, particularly the method for deriving national skin color values.
A paper published in 2000 by Jablonski & Chaplin ("The evolution of human skin coloration") can more directly address these criticisms. Jablonski & Chaplin published a table of skin color reflectance values from many old world populations (Table 6, also see the appendix). I very crudely averaged values from the same country to make a new measure of national skin color. This measure of national skin color correlates with the skin color index of Templer & Arikawa at r=-.91 (the negative is not important here). The reflectance measure of skin color correlates with national IQ at r=.87. The school achievement measure of Lynn et al. correlates r=-.79 with the skin color index of Templer & Arikawa and r=.75 with the skin color reflectance values crudely averaged from Jablonski & Chaplin. Thus, the skin color values derived by Templer & Arikawa are well validated by an external data source and the national IQ-skin color relationship is found to be robust across two measures of national IQ and two measures of national skin color.
Note that there are substantially more missing values in the school achievement and skin reflectance data sets (no imputation of missing values) with missing values skewed towards lower values of national IQ/school achievement and darker skin colors. Also note that the blind averaging use on the skin reflectance data most likely attenuates the correlations.
Templer & Arikawa had two abstracts at the 2006 ISIR conference, which provide additional support for the validity of the measures and their relationships:
Correlations of Skin Color and Continent with IQ
Donald I. Templer & Hiroko Arikawa
The present study determined (1) the correlations between skin color and IQ across the countries of three different continents; and (2) the correlations of both skin color and continent in the three pair combinations with the three continents. The product-moment correlations between IQ and skin color were -.86 across the 48 African countries, -.55 across the 48 Asian countries, and -.63 across the European countries. When the 96 countries of Africa and Asia were combined skin color correlated -.86 and continent correlated .75 with IQ. The respective correlations were -.97 and .89 across the 81 countries of Asia and Europe, and -.71 and .54 across the 81 countries of Europe and Asia. In multiple regression continent yielded minimal increment to skin color in predicting IQ. In an earlier study (Templer & Arikawa, 2006a) skin color correlated more highly with IQ than racial category, but racial category yielded greater increments in multiple regression than did continent in the present study. The present findings, combined with previous research relating skin color and IQ (Templer &amp;amp;amp;amp;amp; Arikawa, 2006a; 2006b), indicate that skin color is a robust correlate of IQ in an international perspective.
Empirical Support for Rushton's K Differential Theory
Donald I. Templer & Hiroko Arikawa
The purpose of the present study was to empirically substantiate Rushton's Differential K Theory that purports that groups of persons with K (in contrast to r) characteristics have a life history and reproductive strategy that includes higher intelligence, less reproduction, less sexual activity, better care of offspring, lower birth rates, greater life expectancy, better impulse control, and greater social organization. The present research intercorrelated national mean IQ, infant mortality, HIV/AIDS rates, birth rates, prevalence rates, and life expectancy in 129 countries in Africa, Asia and Europe. All of the correlations were substantial and in the expected direction. Also supportive of Rushton's theory is that there was only one factor which accounted for 75% of the variance and was labeled "K-r continuum." All five variables were correlated with an economic variable (per capita income) and a biological variable (skin color, which correlated highly with intelligence in previous research). Skin color correlated more highly with all five variables than per capita income so as to support the contention of Rushton that this continuum is biologically based. Factor analysis with all seven variables yielded one factor that accounted for 73% of the variance.
Jason Malloy adds: Templer & Arikawa's research follows Lynn and Rushton in arguing that cold temperatures were a significant force in the evolution of human race differences in intelligence. I have stated some problems I find with this hypothesis here, although it is also largely consistent with the geographic distribution of global populations by IQ. A recent analysis by blogger Audacious Epigone adds yet another revealing data point to this association.
Latitude (and hence colder climate) is associated with IQ not only cross-nationally (.67) but within the US as well. AE found a correlation of .70 between his measure of state IQ and the latitude of the most populous city in each of the 50 states. Furthermore intelligence is associated with latitude equally for both US whites and blacks (.52 and .51).
It's not immediately apparent if and how this association is genetic or environmental. Either way it seems fair to seriously consider that global warming will provide yet another detrimental negative pressure on the intelligence of human populations in the coming decades.