PAST RESEARCH HAS BEEN MOSTLY DEVOTED TO DEVELOPING MATHEMATICAL MODELS, SUCH AS LINEAR AND CURVILINEAR REGRESSION EQUATIONS, AS VEHICLES FOR PREDICTION. HOWEVER, THESE METHODS REQUIRE INFORMATION ON SUCH VARIABLES AS DEMOGRAPHIC POPULATION CHARACTERISTICS AND OFFENSE TYPES. METHODS SIMILAR TO REGRESSION ANALYSIS HAVE BEEN DEVELOPED WHICH PRODUCE PROMISING RESULTS WITH DATA CONTAINING UNSPECIFIED INTERACTIONS AND HETEROGENEITY. THESE TECHNIQUES, LABELED ASSOCIATION ANALYSIS AND PREDICTIVE ATTRIBUTE ANALYSIS, MAY OFFER A MEANS FOR IMPROVING THE EFFICIENCY AND POWER OF REGRESSION-BASED PREDICTION TABLES, FOR PROVIDING A MEANINGFUL OFFENDER TYPOLOGY, AND FOR COMPARISONS OF TREATMENT EFFECTS UPON DIFFERENT OFFENDER TYPES. CAUSAL MODELS OF PREDICTION, PARTICULARLY REGRESSION, MAY BE USED TO MAKE INFERENCES ABOUT DELINQUENT BEHAVIOR AND ECONOMIC FACTORS RELATING TO CRIME RATE. TIME-SERIES ANALYSIS MAY BE APPLIED TO FORECASTING TRENDS IN CRIMINALITY OR FOR TESTING FOR SEASONALITY IN REPORTED CRIME DATA. A NUMBER OF STUDIES HAVE BEEN CONDUCTED TO EXAMINE PARAMETERS OF SOCIAL, ECONOMIC, AND POLITICAL INFLUENCE IN CRIME DATA. ORDINARY LEAST-SQUARES AND TWO-STAGE LEAST-SQUARES METHODS HAVE BEEN USED TO EVALUATE RELATIONSHIPS BETWEEN POLICE DEMAND AND CRIME SUPPLY AND FOR COST-EFFECTIVENESS/BENEFIT ANALYSES. A MOVING MULTIPLE REGRESSION MODEL WAS DEVELOPED TO ELUCIDATE THE RELATIONSHIPS BETWEEN METEOROLOGICAL VARIABLES AND POLICE MANPOWER NEEDS. SUCH A MODEL COULD PROVIDE INFORMATION FOR MANPOWER ALLOCATION AND DEPLOYMENT DECISIONS. IN A STUDY OF SEASONAL CORRELATES OF CRIME PATTERNS IN MIAMI, A RANDOMIZED COMPLETE-BLOCK DESIGN ANALYSIS OF VARIANCE WAS USED. SEASONAL INDICES WERE DERIVED FROM AN ANALYSIS OF DATA COVERING THREE TIME-BLOCKS FROM 1949 TO 1970. IT WAS FOUND THAT CRIMES AGAINST PERSONS (MURDER, RAPE, AGGRAVATED ASSAULT) LACKED SEASONALITY, WHILE CRIMES AGAINST PROPERTY (ROBBERY, BURGLARY, LARCENY, AUTO THEFT) SHOWED SIGNIFICANT SEASONALITY. TABULAR DATA AND REFERENCE NOTES ARE PROVIDED. (JAP)