MULTIVARIATE SIMULATION WITH DATASIM - THE MIHAL AND BARRETT STUDY

A general-purpose data simulator, DATASIM, allows instructors and stud ents to easily generate simulated data for experimental, crossbreak (f requency table), and multivariate research designs. Students can selec t any of a variety of data analysis problems from the built-in library and then obtain problem descriptions and ''individualized'' data sets to analyze and interpret. DATASIM can also be used to generate, plot, and analyze interesting data sets for discussion in class, to demonst rate principles of sampling theory and research design, and to conduct sampling experiments (Monte Carlo research). Finally, the electronic blackboard capability of the program allows instructors to convenientl y display multiple sources of information, such as outlines, graphs, s ample data sets, ANOVAS, and the like. The present paper focuses on th e techniques and methods required to perform multivariate simulation. To this end, DATAsim is used to simulate a study by Mihal and Barrett (1976), which predicted auto accidents from measures of perceptual sty le, reaction time, and selective attention.