Inferences about actions performed in constraining contexts: Correspondence bias or correspondent inference?

Abstract

Prior studies of the correspondence bias reveal a tendency for people to base inferences on behaviors they observe, even when these behaviors are highly constrained by the social context in which they occur. Three processes may combine to sustain this effect: (a) an insufficient adjustment of initial estimates caused by the fundamental attribution error (FAE); (b) the reliance on intuitive estimates of the prevalence of traits, attitudes, and other dispositional characteristics in the general population; and (c) the assumption that few situations are so coercive that they negate all freedom of choice. These processes were differentiated in a modification of the Jones-Harris paradigm. Participants estimated a particular attitude's prevalence in the general population before reading an essay written under either high or low choice conditions and taking a probable or improbable position. Inferences were consistent with prior probabilities when the essay expressed a highly probable opinion and consistent with behavior when the essay expressed an unlikely opinion. These results suggest that perceivers make inferences by estimating the probability that the observed behavior reflects a dispositional characteristic and then revising their prior estimates of the probability of that characteristic accordingly.