Two approaches for teaching empirical research methods to graduate students in political science courses are evaluated. Specific objectives of the evaluation, in addition to comparing relative teaching effectiveness of the two approaches, were to explore causal dynamics between the teaching environment, user satisfaction, and learning performance. The first approach, EXPER-SIM (Experiment Simulation), is a computer program which transmits simulation models into fortran subroutines and generates data according to a selected model. The second approach, SETUPS, (Supplementary Empirical Teaching Units in Political Science), includes instructional packages covering various aspects of political science. A SETUP typically contains an introduction to theory and literature, data analysis exercises, codebook and data sets, and directions for use in conjunction with statistical software packages. Evaluation of the approaches took place during a one-semester course in which students were randomly assigned to either EXPER-SIM or SETUPS groups. Students in both groups investigated the same analytic problems. Findings from statistical analysis of pre-and posttest scores and from general course ratings by students indicated that the EXPER-SIM students showed greater gains in analytical reasoning ability, positive perceptions of computer utility, ease of computer use, and substantive learning performance. The conclusion is that EXPER-SIM is more effective than SETUPS as a strategy to teach methods of empirical research. (DB)