Hypothesis Testing and the Perception of Diagnosticity

Abstract

Two different views of social hypothesis testing have been put forth. (1) A poorly calibrated process dominated by biases toward hypothesis confirmation. (2) Rational decision making where a diagnosing strategy prevails. By recognizing the similarity between the standard hypothesis-testing situation and the 2 × 2 covariation paradigm, we conclude that neither of these views is necessarily correct. In particular, we demonstrate that use of a diagnosing strategy is moderated by the presentation format such that sequentially presented information is much less likely to induce a diagnosing strategy than is a tabular presentation. In addition, when a diagnosing strategy is separated from a simpler (and often indistinguishable) hypothesis-testing rule, this simpler rule is strongly preferred. Thus, the apparent dominance of a diagnosing strategy in social hypothesis testing may be due, in part, to a favorable presentation format and to the use of a nondiagnosing strategy that is generally indistinguishable from the use of diagnosticity. Results indicate the advantages of a stronger integration between the hypothesis testing and the covariation literatures.