Validity of short forms of the category test

Abstract

The criterion - related validity of two short forms of the Category Test was examined. As was the case in earlier studies, both short forms yielded high validity coefficients and high agreement with respct to Halstead's cut - off score for impairment. The 120 - item form was the stronger predictor, accounting for 90.8% of the variance in the original Category Test, which the 108 - item form accounted for 79.5%. However, both short forms were seriously limited with respect to how closely, the estimated scores approximated the actual scores; even the best short form prediction equations produced estimated scores within a 10 - point range of the actual scores only 44% and 42% of the time. Since the main clinical purpose for using a short form would be to obtain an estimate of the actual score, neither short form can be recommended for clinical use. This study illustrates the need to consider the specific purpose for which a predictor is to be used when investigating the crierion - related validity of a measure.