On the assessment of cognitive style: a commentary on Peterson, Deary, and Austin

The reliability of Ridingʹs popular Cognitive Styles Analysis test (CSA) was examined by comparing performance on the original CSA test and a new parallel version. Fifty participants completed both test versions twice, with the second sitting approximately a week later. Reliability was measured using parallel forms, test–re-test, and split-half analysis. Correlations of the verbal–imagery (VI) and wholist–analytic (WA) ratios from both test versions (original and parallel) were low (Range r=0.07 to r=0.36, Mean r=0.24). When the CSA and parallel form data were combined however, split-half analysis of the WA style ratio was stable (Mean r=0. 69) but the VI style ratio remained unreliable (Mean r=0.36).