An experimental model for the study of conformity to overt group norms
Oshry, Barry Irving
MetadataShow full item record
A survey of the research is the area of conformity to overt group norms reveals that the following basic generalization seems adequate to cover many of the findings in this area: accuracy of performance is a function of accuracy of feedback. Previous experimental models used in these studies, however, have not differentiated between the degree to which "conformity" is a function of the particular source to which this feedback is attributed and the degree to which it is simply a function of the statistical quality of this feedback. The purpose of the present study was to establish an experimental model in which variations in the statistical aspects of feedback, particularly in the precision and bias of the feedback, could be manipulated independently of attributed source and other group factors, and the effects of such variations measured. The psychophysical method of absolute judgments was selected as the task feature for this experimental model. In this task, the subject is required to make a series of absolute judgments of the texture of photographs using a thirteen point response scale, and for this study each subject made 507 responses daily for seven consecutive days. After each response, the subjects were informed of either the correct value of the stimulus or an arbitrary value designated as the average of the group's responses. Thus the precision and bias of this feedback information could be strictly controlled and measured. It was also possible to compute measures of the conformity between the distributions of these feedback values and those of the subjects' responses. Twenty-five subjects were randomly assigned to five groups for each of which the precision and bias of feedback and the attributed source of these feedback were varied. The measure of over-all precision of feedback (amount of information transmitted) was not found to be an accurate predictor of the over-all precision of performance. A close relationship, however, was found between the degree of scale compression of feedback and the degree of response scale compression in the performance of the subjects. It is suggested that scale compression of the feedback functions in determining the frame of reference for performance. This frame of reference is accepted fairly uniformly by all subjects in the feedback groups. The variability in the overall precision of performance, however, indicates that there still are marked individual differences in the consistency with which subjects operate within this frame of reference. A single variation in attributed source of feedback was also used. Two groups received the same perfectly precise and completely unbiased feedback. One group was informed that this feedback represented the average group judgment of their fellow group members, while the other was informed that this feedback represented correct stimulus values. No systematic differences in performance, with any of the measures of performance used, were found as a result of this variation in instructions. The two major findings with regard to the conformity measures are: 1. When the feedback is perfectly precise and completely unbiased, there is a uniform, progressive increase in the degree of conformity to this feedback and, as a result, in the precision of the performance. 2. When the precision of feedback was below the general precision of performance, subjects showed two opposed tendencies. Sometimes they disregarded the feedback and improved the precision of their performance; while at other times they tended to conform to the feedback at the expense of accuracy in performance. There was not sufficient consistency within subjects to allow prediction of the mode taken at any session. One group received as feedback the actual numerical average of the group's responses. Since the experimental task involved a systematic judgment bias, the feedback was also biased. As a result, this group showed an effect which has been labelled the "snowball" effect. Since the conformity to feedback increased the subjects' bias, this in turn increased feedback bias on successive trials, which would again further increase subjects' bias, and so forth. As a result, this group showed the greatest degree of scale compression and loss of accuracy in performance. It is suggested that an average-group-response feedback condition represents an ideal situation in which to study variations in the attributed source of feedback since the rate of "snowballing" would represent a measure of the degree to which feedback information is being attended.
Thesis (Ph.D.)--Boston University