Date of Award

Spring 1997

Project Type


Program or Major


Degree Name

Doctor of Philosophy

First Advisor

Victor Benassi


Some psychologists have claimed that people are not good at judging covariation (e.g., Smedslund, 1963; Jenkins & Ward, 1965). This claim, however, has been based on the results of experiments that may not have been optimal for promoting judgmental accuracy (Allan & Jenkins, 1980). Other psychologists have claimed that people are relatively good judges of covariation (e.g., Wasserman, Chatlosh, & Neunaber, 1983; Wasserman & Shaklee, 1984). Common to most of this research is an experimental paradigm in which participants do not ever receive feedback concerning the accuracy of their judgments.

The two experiments in this dissertation were designed to promote accuracy in the judgment of contingency by providing (a) accurate feedback concerning participants' judgments and (b) practice with judging many contingency problems. The results of these experiments indicate that people become better judges of contingency with feedback and practice, but do not improve with practice alone. This is true for positive contingencies (Experiment 1) and negative contingencies (Experiment 2). Judgmental accuracy was greatest for extreme contingency problems ($\Delta$P less than $-$.75 and greater than.75).

Self-efficacy has been shown to account for performance in a variety of domains above and beyond ability. Experiment 1 addressed the relation between self-efficacy, feedback, and judgmental accuracy. Mean self-efficacy increased over the course of the experiment for participants in the feedback condition, but decreased for participants in the no feedback condition. Participants with high self-efficacy in the feedback condition showed relatively accurate judgments of contingency, but participants with high self-efficacy in the no feedback condition showed relatively inaccurate judgments of contingency.

Experiment 1 also addressed whether judgmental accuracy on one contingency task transferred to judgmental accuracy on a different task. The results indicate that there was no transfer in accuracy from one task to another.