During 1975-1979 this research into the potential of computerized adaptive testing to reduce errors in the measurement of human capabilities used Marine recruits for a live-testing validity comparison of computerized adaptive and conventional tests. The program purposes were to: (1) identify the most useful computer-based adaptive testing strategies; (2) identify testing conditions that maximize the positive rather than negative psychological effects of computerized testing; (3) investigate intra-individual multidimensionality problems in ability testing; (4) examine probabilistic responding and free-response methods for computerized adaptive testing in order to extract maximum information from each test item response; and (5) develop, refine and evaluate new computer administered ability tests for spacial, perceptive, memory, and other abilities not now measurable using paper and pencil testing. Monte Carlo and Bayesian adaptive testing methods were used in these studies. Fifteen major findings, primarily on adaptive testing and test administration conditions, and implications for further research are given. Abstracts of the 16 research reports for studies for this program are given. (BS)