The Structure and Reliability of Riding's Cognitive Style Analysis Test

Elizabeth Peterson   Ian J Deary   Elizabeth J Austin,  Department of Psychology, University of Edinburgh,   UK.

Currently one of the most frequently used and empirically validated measures of cognitive styles is Riding's computerised Cognitive Styles Analysis test (CSA).  The CSA assesses an individual's performance on the verbal-imagery dimension of cognitive style and the wholistic-analytic dimension of cognitive style. The assessment of style is based on the ratios of reaction
times in different conditions. Unlike many other measures of cognitive style, the CSA has undergone substantial empirical investigation. Findings indicate the CSA style dimensions are independent of personality, separate from intelligence and related to observed behaviour and physiological measures.  However, little empirical research has been conducted on the CSA's reliability or examined closely the suitability of the CSA's structure and design. The aim of this study was therefore to examine critically the CSA's reliability, structure and design. Fifty participants completed the CSA and a new parallel version designed for this study, with a second sitting of both tests approximately a week later. Similar error rates and reaction times were found on the CSA and the parallel version. Reliability was measured using parallel forms, test-re-test, and split-half reliability analysis. Correlations between and within the verbal-imagery (VI) and wholistic-analytic (WA) ratios on the original and parallel tests were lower than is conventionally accepted for a psychometric test (range r = .074 to r = .358, Mean r = .24). When the CSA and the parallel versions data was combined, a split half reliability approaching r = .7 (Mean r = .689) was found for the WA ratio however, the VI remained unreliable (Mean r = .357).  Therefore, by using the CSA and the parallel version together, people can be reliably categorised on the WA dimension. The instability of the VI dimension may result from the more ambiguous verbal and imagery stimuli.  Problems associated with the use of ratio measures to assign a cognitive style will be discussed.

Return to Abstracts Index