Judging interevent contingencies: Being right for the wrong reasons

Abstract

Mathematically inaccurate judgment rules often produce correct covariation judgments; thus, accuracy of covariation judgment alone may be a poor index of the sophistication of a subject’s understanding. We offer a past paradigm of our own (Wasserman & Shaklee, 1984) as an instance in which impressive judgment accuracy may have been the product of simple and inaccurate judgment rules. The present investigation replicates the judgment paradigm of our prior experiment, using a set of 12 covariation problems designed to produce unique judgment patterns by each of four judgment rules. Subjects’ judgment patterns indicated that use of a mathematically accurate rule was quite rare (comparison of conditional probabilities: 3.1% of subjects). The modal judgment pattern conformed to that predicted by a rule in which subjects compare only two cells of a 2×2 contingency table (Strategy a-versus-b: 38.1% of subjects). Distributions of strategy classifications differed among several judgment conditions which varied in the presentation format of event-frequency information.