Abstract
Forty-nine healthy adults and 45 brain-injured adults were tested using one of three versions of the Category Test. Membership based on brain-injury versus non-brain-injury produced a significant main effect on the total error score for all subtests, except Subtest 2, of the Category Test. No significant differences between versions of the Category Test on total error score or an interaction between Category Test version and group membership was found. Analysis of individual subtest error scores indicated significant differences between versions on Subtest 3 and Subtest 6, with the traditional version fairing least well in differentiating brain damage. These data support the contention that the Category Test is robust to changes in the form of administration.