This study continued the search for the optimum individual monetary incentive pay system by investigating the effects of a flat pay system and three individual incentive pay systems (linear, positively accelerating and negatively accelerating) on worker productivity. Four experiments were conducted using a within-subject, multiple-baseline design. Experiment I was a systematic replication of the earlier work of Oah and Dickinson (1992) and Experiments 2, 3 and 4 were systematic replications of each other with different questions being asked about the efficacy of the three incentive pay systems. In each experiment, the subjects worked in groups, ranging in size from four to six members, and were exposed to the flat pay system and to one of the incentive pay systems while engaged in a simple production task. Subjects participated in twenty (Experiment 1 and 2) to twenty-five (Experiment 3 and 4) fifteen-minute sessions. The production task consisted of constructing ''widgets'' from pop beads. The primary dependent variables were the number of correctly made widgets per work session and the cost-per-widget. The four experiments produced mixed results with respect to widget productivity and cost-per-widget; however, there were two consistent findings. A systematic relationship between pay and productivity emerged in that, with all four experiments, the incentive pay systems generated higher levels of productivity than did the flat pay system. Also, the three incentive pay systems differentially affected performance levels and cost-per-widget. Across the four experiments, the negatively accelerating pay system emerged as the most reasonable option for pay system designers. This finding suggests that it is not the size of the incentive which controls performance, but rather the fact that there was a pay-for-performance contingency in place.
