UPDATING THE TRYON MULTIVARIATE ANALYSIS COMPUTER SYSTEM

THE PRIME LONG-TERM OBJECTIVE OF THIS PROJECT IS TO UPDATE, ENHANCE, AND DISTRIBUTE TWO COMPUTER PROGRAMS UTILIZING THE UNIQUE MULTIVARIATE ANALYSIS TECHNIQUES DEVELOPED BY THE PSYCHOLOGIST ROBERT TRYON WITH NIH SUPPORT DURING THE 1960S. ONE LONG-TERM GOAL IS TO PRODUCE AN ENHANCED ALL-FORTRAN VERSION OF A MAIN BCTRY82 SYSTEM, PORTABLE TO MOST LARGE COMPUTERS. A SECOND LONG-TERM GOAL IS TO PRODUCE VERSIONS OF A SMALLER TRYCLUST PROGRAM, FOR USE IN BOTH MICROCOMPUTER AND SUPERCOMPUTER ENVIRONMENTS. TRYON'S TECHNIQUES DEVELOPMENT OF CLINICAL TAXONOMIES, PRODUCED BY CLUSTERING CASES IN A MULTIDIMENSIONAL SPACE DEFINED BY RELIABLE ASSESS-MENT SCALES; MONTE CARLO PREDICTION OF IMPORTANT OUTCOME OR CRITERION VARIABLES FROM TAXONOMIC GROUP MEMBERSHIP. PHASE I TRYCLUST WORK WILL INVOLVE EXPANDING ITS CURRENT CAPABILITIES AND TESTING THE EXPANDED PROGRAM ON MICROCOMPUTERS AND ON A VARIETY OF LARGE MACHINES. PHASE I BCTRY82 WORK WILL FOCUS ON CONVERTING THE PRESENT CODE TO FORTRAN V, REDIMENSIONING THE PROGRAM TO HANDLE VERY LARGE DATASETS, AND ADDING A NEW STATISTICAL PROGRAM, AS A FEASIBILITY DEMONSTRATION. PHASE I TASKS ARE STRAIGHTFORWARD AND SHOULD BE SUCCESSFUL. PHASE II TASKS ARE A LOGICAL EXTENSION OF PHASE I WORK AND SHOULD LIKEWISE BE SUCCESSFUL. THESE PROGRAMS WILL FILL A SPECIAL NICHE IN MULTIVARIATE APPLICATIONS SOFTWARE, AND THEIR COMMERCIAL DISTRIBUTION IN MICROCOMPUTER