Expectation-based and data-based illusory correlation: The effects of confirming versus disconfirming evidence

The present study (n = 154) examines the effects of expectations and stimulus information on the perception of illusory correlation. There have been few studies attempting to integrate expectation-based and data- (dixtinctiveness-) based processes. These studies suggest that data-based illusory correlation can be overruled by prior expectations, but it is not clear whether this is a consequence of a confirmation bias. In the present study, where participants were not exposed to the specific stimulus information, expectation was manipulated by stating that group B behaved more negatively than group A. Moreover, participants were provided with information contained in a statement-rating task that allowed for the confirmation and disconfirmation of the prior expectations. Participants rated the desirability of these behaviours and also performed the standard illusory correlation tasks. Based on self-categorization theory and Alloy and Tabachnik (1984), we predicted that in the absence of prior expectations, completing the rating task before the illusory correlation tasks would produce stronger illusory correlation than the reverse order. However, in the presence of prior expectations we expected the rating task to undermine illusory correlation, because the information obtained in this task tends to disconfirm prior expectations. Results support the predicted interaction between task order and expectation. We discuss some implications for research on confirmation bias.