External conditions affecting experimental bias

The experimenter bias effect was re-examined under more rigorous controls and a more elaborate experimental design than has typically been found in previous studies of the same topic. Thirty student Es administered a person-perception task to 120 student Ss. The Es were equally divided into three expectancy-inducing conditions including a no-expectancy control group; half of the Es administered the task under monitored conditions and the remaining half did so under unmonitored conditions. Each E administered 10 photos to each of 4 subjects; subject order and trial blocks within subjects were incorporated as main effects along the replicated axis of the design. Extra control and quasi-control measures were taken in the form of postexperimental E and S questionnaires, the use of confederate 5th subjects with preprogrammed responses in the case of non-monitored Es with expectancy outcomes, and observation of the videotapes generated from the 'monitored' portion of the study by experimental psychology students and by the author. Results indicated that the bias effect does not appear to be mediated by data fabrication or intentional or unintentional misjudging or misrecording of responses. As predicted, monitoring the Es with audio-video equipment had significant effects on the outcome of the experiment. Nonmonitored Es with outcome expectancies elicited ratings in the expected direction to a greater extent than did monitored Es with expectancy outcomes (p< .01). Further analysis of the monitoring effect indicated that a reverse bias phenomenum may partially account for differences obtained between monitored and non-monitored Es; with later contacted subjects, excessively rewarded Es obtained responses in the unexpected direction to a greater extent than did monitored control Es. Additional findings indicated that monitored Es seem to behave differently from non-monitored Es regardless of outcome orientation. Also, a portion of the results suggested that further research of whether Ss are more task-conscious than they are experimenter-conscious would be appropriate for further understanding of the experimenter bias effect.