THE THEME OF THIS PROGRAM WAS THE USE OF PROBABILISTIC MODELS AS MEASUREMENT TOOLS FOR EXPERIMENTAL DESIGN IN TEST CONSTRUCTION AND FOR ANALYSIS OF TEST DATA. SOME CENTRAL ISSUES IN THIS FIELD ARE DISCUSSED AND RESOLVED IN THE FIRST FOUR CHAPTERS OF THE REPORT UNDER THE FOLLOWING TITLES--(1) CONTRIBUTIONS TO ITEM SELECTION, (2) CLUSTERING PROCEDURES, (3) ASYMPTOTIC EFFICIENCES OF SOME TESTS OF INDEPENDENCE USED ON ITEM SELECTION, AND (4) THE SCORING OF MULTIPLE-CHOICE QUESTIONNAIRES. THE REMAINING CHAPTERS ARE STUDIES OF A NUMBER OF METHODOLOGICAL ISSUES WHICH AROSE IN THE CONSIDERATION OF THE ABOVE CENTRAL ISSUES, AS FOLLOWS--(1) THE USE AND INTERPRETATION OF PRINCIPAL COMPONENT ANALYSIS IN APPLIED RESEARCH, (2) TESTING FOR THE EQUALITY OF CORRELATION COEFFICIENTS FOR VARIOUS MULTIVARIATE MODELS, (3) ASYMPTOTIC DISTRIBUTION OF FUNCTIONS OF A CORRELATION MATRIX, (4) ORDER STATISTICS IN TWO DIMENSIONS, (5) A DECOMPOSITION OF DICHOTOMOUS RANDOM VARIABLES INTO INDEPENDENT COMPONENTS OF VARIATION, AND (6) SOME APPLICATIONS OF FINITE BOOLEAN ALGEBRAS TO DATA ANALYSIS. (LP)