Effects of immediate knowledge of results and adaptive testing on ability test performance

Abstract

This study investigated the effects of immediate knowledge of results and adaptive testing on performance on a computer-administered test of verbal ability. Examinees were administered either a 50-item conventional test or an adaptive test of verbal ability; half the subjects in each group received immediate knowledge of results (KR) concerning the correctness/incorrectness of each item response, while the other half did not. Subjects within high- and low-ability subgroups were assigned randomly to one of the four resulting experimental conditions. The dependent variable was maximum likelihood ability estimates derived from item response patterns. Results indicated that for the high-ability group, mean test scores under KR conditions were significantly higher than were those under no-KR conditions on both the conventional and adaptive tests. Within the low-ability group, mean test scores were higher under KR conditions than under no-KR conditions, but the difference was statistically significant only within the conventional testing strategy. Low-ability examinees achieved higher average test scores on the adaptive test than on the conventional test, while high-ability examinees performed equally well on the adaptive and conventional tests.