Sunday, March 14, 2010

Richard Lynn's controversial study (3/5). Method

Data have been assembled for 12 Italian regions for mean IQ, average per capita income in euros for 1970 and 2003 given by the Italian Statistical Office (2008), percentages of the populations that were literate in 1880 [...], statures of military conscripts born in 1855, 1910, 1927 and 1980 [...], infant mortality 1955–57 and 1999–2002 [...], years of education in 1951, 1971 and 2001 [...] and latitude taken as the approximate geographical mid-point of the regions.
The regional IQs have been calculated from the 2006 PISA (Program for International Student Assessment) study of reading comprehension, mathematical ability, and science understanding administered to 15 year olds in 52 countries (OECD, 2007). Scores on these tests are used as a proxy for IQs, adopting the procedure that Rindermann [...] has used for nations. The PISA reading test is a measure of verbal comprehension and the mathematics test is a measure of "quantitative reasoning", and both of these are major components of general intelligence [...], while science understanding is highly correlated with general intelligence [...].
More generally, numerous studies have shown that tests of educational attainment are highly correlated with intelligence at around 0.5 to 0.7 [...], and sometimes more highly, e.g. at 0.81 in a recent study of 70.000+ English children whose IQs were measured at the age of 11 years and educational achievement was measured at the age of 16 years [...]. It has been shown that there is a strong genetic correlation between cognitive ability measured by tests of intelligence and educational, i.e. the same genes determine ability measured in both kinds of test [...]. These are designated "generalist genes" by Kovas, Harlaar, Petrill and Plomin [...] because they determine many expressions of cognitive ability including IQs, math, reading, science, etc. The terms "intelligence" and "IQ" are used in this paper in the sense of the sum of all cognitive abilities or global IQ, as measured by intelligence tests such as the Wechslers and the Binets. The PISA tests measure some mix of g (Spearman's g, the general factor present in all cognitive abilities), gf (fluid intelligence or reasoning ability) and gc (comprehension/knowledge) but it is not considered possible to quantify the contributions of these three factors to the PISA scores.
To calculate IQs for the Italian regions the scores on reading comprehension, mathematic ability, and science understanding have been averaged and these averages have been expressed in standard deviation unit deviations from the British PISA mean (502, SD=99). This gives scores for the Italian regions expressed in standard deviation units in relation to the British mean. These figures are then converted to conventional IQs by multiplying them by 15. Thus, the regional Italian IQs are expressed in relation to British mean IQ of 100, SD 15. Data are missing for Tuscany, Umbria, Lazio and Calabria.