The Multidimensional Structure of Verbal Comprehension Test Items

Abstract

This article investigates the mutidimensional structure of verbal comprehension test items. Empirical evidence, based upon results of a verbal comprehension test (a subtest of the Lorge-Thorndike Intelligence Test) administered to Israeli sixth graders, is provided to support the theory that item tasks are multivariate-multiordered composites of three faceted components: language, contextual knowledge and cognitive operation. Guttman's "Order" and "Facet" theory, as well as his Smallest Space Analysis (SSA) were used with the data. The SSA yielded a "cylindrical" structure suggesting that (a) language and cognitive-operation components represent additive "key" skills that maintain the cognitive knowledge processing skills, and (b) that a differentiation can be made between skills that are automatically applied and those that are applied through processing operations. Linear and circular properties of the cylindrical manifestation were demonstrated by the serially ordered character of the sub-matrices of the matrix of interitem correlation coefficients.

The cylindrical structure of the Wechsler Intelligence Scale for Children — IV: A retest of the Guttman model of intelligence

“…In ability testing this structure suggests the appearance of more cognitively complex tasks closer to the center of the figure and less complex tasks farther away from the center, with the appearance of verbal, numerical, and figural-spatial test content in separate sectors. The radex model of intelligence was tested through MDS (Marshalek, Lohman, & Snow, 1983;Snow, Kyllonen, & Marshalek, 1984) or SSA (Adler & Guttman, 1982;Beauducel, Brocke, & Liepmann, 2001;Koop, 1985;Guttman, 1965a,b;Peled, 1984;Schlesinger & Guttman 1969;Tziner & Rimmer, 1984, 1991Ziedner & Feitelson, 1991). In all the above studies, encompassing various ability test batteries and various sample characteristics, the findings were more or less similar, indicating a two-dimensional solution where the tests were ordered from the most complex, abstract, inferential tasks near the origin, to the more simple tasks requiring associative learning at the periphery.…”