An examination of the Flynn effect in the National Intelligence Test in Estonia.

Shiu, William.
Access rights
Worldwide access.
Access changed 5/21/14.
Journal Title
Journal ISSN
Volume Title

This study examined the Flynn Effect (FE; i.e., the rise in IQ scores over time) in Estonia from Scale B of the National Intelligence Test using both classical test theory (CTT) and item response theory (IRT) methods. Secondary data from two cohorts (1934, n = 890 and 2006, n = 913) of students were analyzed, using both classical test theory (CTT) and item response theory (IRT) methods. CTT analysis compared the summed score for each subtest between the cohorts. IRT analysis examined item invariance across the time period and then, for each subtest, linked the latent variable scores between the two cohorts using the invariant items. IRT analyses revealed that each subtest displayed invariance on over 50% of the items (i.e., partial measurement invariance). Additionally, results from the current study found positive score gains but also revealed reverse FEs. CTT analysis showed three subtests had a FE (Vocabulary [.74], Analogies [1.09], and Comparisons [1.71]), while two subtests had a reverse FE (Computation [-.33] and Information [-.03]). The IRT analysis found that four subtests had a positive FE (Information [.44], Vocabulary [.79], Analogies [1.02], and Comparisons [1.51]), with only the Computation (-.10) subtest displaying a negative FE. The results con rm previous research that the FE continues in Estonia. Using CTT methods, Must, te Nijenhuis, Must, and van Vianen (2009) found positive gains on the Estonian NIT subtests Computation (.15), Information (.94), Vocabulary (.65), Analogies (1.81), and Comparison (2.34). An implication of the current study shows the viability of IRT to supplement CTT when analyzing the FE. The IRT procedures demonstrated in the current study provides a counter argument that the rise in IQ scores is a psychometric artifact, at least in the domains of Comprehension-Knowledge, Fluid Reasoning, and Visual Processing. As this study was unable to examine causative factors involved in the FE, future studies should examine if the score gains might be attributed to some environmental cause (e.g., nutrition, education) or biological cause (e.g., heterosis).

Item response theory., Classical test theory., Intelligence., Differential item functioning., IQ scores., Estonia., Invariance., National Intelligence Test., Flynn effect.