2 STRUCTURAL LAWS FOR INTELLIGENCE-TESTS

This article discusses two structural laws for intelligence tests. After reviewing the relevant definitions and the empirical evidence supporting these laws, it proposes an extension of the laws, namely, that of a cylindrical structure. Supporting data are presented from the American and Israeli versions of the Wechsler WISC-R. The first law of intelligence testing concerns the sign of correlation coefficients (and monotonicity of regression). It gives conditions under which one can expect all correlations between intelligence test items to be positive (or, at worst, zero). The first law is confirmed by the date of the Wechsler tests. The second law concerns the relative sizes of the correlation coefficients among intelligence items, and specifies a two-facet design for the content of the items from which to expect a circular (radex) arrangement of the items according to their intercorrelations. Previous work on these two laws was largely confined to paper and pencil tests. For the 12 subtests of the WISC-R, a further design facet is employed, with the following elements: paper and pencil, manual manipulations, and oral. This provides items whose intercorrelations require a three-dimensional representation linking them to a cylindrical interpretation in the light of the three design facets. The new facet corresponds to the axis of the cylinder, while the other two facets correspond to the circular base of the cylinder. The cylindrical correspondence holds for both the American and the Israeli (Hebrew language) versions of the WISC-R, and for each of the yearly age levels from 6 to 16. © 1991.