On the Utility of Consumers' Theories in Judgments of Covariation

Abstract

Previous research on people's proficiency in assessing covariation between continuous variables is reviewed, and the impact of prior expectations and characteristics of the evidence on judgment accuracy is discussed. The present study investigates how the availability of a theory (i.e., prior expectations of relatedness) influences assessments of contingency as a function of the diagnosticity of the data. The results indicate that prior expectations have beneficial effects on the accuracy of covariation judgments, even when the relationship implied by the theory is inconsistent with the data. Little evidence is found for a multiplicative combination of theory and data, although diagnostic data lead to more accurate judgments. The study also examines some mechanisms that might underlie these effects and presents evidence that the utility of theories is due to prior expectations that give rise to an active hypothesis-testing approach to the assessment of contingency. The findings suggest that prior expectations do not necessarily have dysfunctional (i.e., biasing) effects on judgments of covariation and that theories, even when they are inconsistent with the data, may facilitate the perception of contingency if they are used as hypotheses to be tested on data.