A 2 × 2 × 2 factorial design was used to test the influence of amount of group variance, direction of advocated shift, and judgmental instruction sets on conforming responses The hypothesis predicted an interaction where conformity was maximized, with less variance under empirical validation instructions and with more variance under social validation instructions

Subjects were preselected for neutral opinions on three items, which were subsequently used as experimental items Social influence was induced by making the responses of four other subjects available on each trial These responses were dispersed with either greater or lesser amounts of variance and either agreed or disagreed with the statement Instructions indicating that expert evaluation was available provided empirical validation instructions, while alternate instructions indicating lack of standardized norms provided social validation instructions

Subjects changed opinions significantly more when the group advocated agreement than when it advocated disagreement on the issue The major hypothesis was partially supported by the data Responses on a postexperimental item indicated preferences for less variance under empirical instructions sets but preference for greater variance under social instructions sets The conformity results suggested that the predicted interaction may occur only if the situational context is favorable, as when the group advocates agreement on an issue