Test-retest reliability of Riding's cognitive styles analysis test

The test–retest reliability of Riding’s cognitive styles analysis (CSA) was investigated over mean intervals of (i) 14 days and (ii) 23 months. The test–retest reliability, measured using the Pearson product-moment correlation, for cohorts of largely third year information systems and engineering students was found to average r=0.34 for the wholist-analytic dimension while the verbaliser–imager dimension varied between −0.19 and 0.36. These values are considerably below the generally accepted value of r=0.8 as a minimum figure for reliability as estimated by test–retest correlation. Consequently it can be concluded that the CSA does not estimate these dimensions with sufficient precision for at least these sample populations.