Two experiments were conducted to examine self-fulfilling expectancy influence in a simulated supervisor-worker interaction. The first experiment led supervisor subjects to expect either high or low compliance from their workers. All workers completed a set of five tasks twice, once when instructed to work fast and once when instructed to work accurately. Speed and accuracy of worker performance was measured, as was the frequency of supervisors' repetition of work instructions. Results indicated that expectancy influence was found only when workers were told to work fast on the first completion of the tasks, and that high-compliance expectation supervisors repeated work instructions more frequently than low-compliance expectation supervisors. A second experiment failed to replicate the successful expectancy effects found in the first. However, results from the second experiment did confirm and extend earlier findings with respect to the cueing of expectations by supervisors, and the apparent awareness of these cues by workers. These two studies were interpreted to indicate that expectancy influence is fragile and difficult to demonstrate. They focus attention on the recipient of an expectancy communication, and they point toward overt mediation of expectations. (Author)