Computer-Based and Paper-Based Measurement of Recognition Performance

摘要

Eighty-three subjects were administered computer-based and paper-based tests to assess recognition of aircraft silhouettes in order to determine the relative reliabilities and validates of these two measurement modes. Estimates of internal consistencies, equivalences, and discriminative and predictive validates were computed. It was established that (a) computer-based and paper-based measures of recognition test score were not significantly different in reliability or internal consistency, (b) the paper-based measure of average degree of confidence in recognition judgments was more reliable or internally consistent than the computer-based measure, (c) computer-based and paper-based measures of average degree of confidence were more equivalent than the measures of recognition test score, (d) according to two sets of criteria, the discriminant coefficients and F-ratios and corresponding means, the discriminative validities of computer-based and paper-based measures were about the same for distinguishing groups above or below average curriculum grade, (e) according to another set of criteria, the pooled within-groups correlations between the discriminant function and computer-based and paper-based measures, the former had superior discriminative validity than the latter, and (f) statistics associated with the canonical correlation suggested the predictive validity of computer-based measures approximates that of paper-based measures. (kr)