Frey and Detterman have a new article out in Psychological Science that shows that, despite ETS being mum on the fact, the SAT is in large part a measure of g, although more so before the 1994 recentering than after.
In Study 1, we used the National Longitudinal Survey of Youth 1979. Measures of g were extracted from the Armed Services Vocational Aptitude Battery and correlated with SAT scores of 917 participants. The resulting correlation was .82 (.86 corrected for nonlinearity). Study 2 investigated the correlation between revised and recentered SAT scores and scores on the Raven’s Advanced Progressive Matrices among 104 undergraduates. The resulting correlation was .483 (.72 corrected for restricted range). These studies indicate that the SAT is mainly a test of g.
What struck me most about this study is that it has very similar findings to a national achievement test I am analyzing now (can’t give too may specifics, as the stuff isn’t published yet). But here is the pattern of loadings.1
So, I guess the question is: can a reliable general achievement test be developed for the population-at-large that does not, in large part, measure g?
1. As a side note, Frey & Detterman, and I used two different methods for extracting g and determining its magnitude in an achievement test. Since I had raw data, I used a hierarchal orthogonal rotation of the subtest scores, while F&D used something akin to Jensen’s method of correlated vectors. Still, we come up very similar findings—g explains between 50-60% of the covariance in our respective achievement tests.
Posted by A. Beaujean at 03:07 PM