Variables influencing judgments of correlational relations.

Investigated the effects of 3 variables on 72 undergraduates' ability to assess correlational relations. These variables were (a) the form in which the stimulus information was presented (i.e., in continuous or binary form), (b) the actual magnitude of the correlation presented in the stimulus information, and (c) the Ss' expectancy concerning the relation in question. Ss judged the correlation between pairs of variables on the basis of serially presented items describing instances of their co-occurrence. Both direct estimates of correlation and phi coefficients derived from frequency estimates were used as measures of perceived covariation. Results indicate that Ss' estimates of correlation were significantly influenced by all 3 factors. Ss' judgments were sensitive to the difference between high and low correlations in the stimulus information, but they reflected this difference to a greater extent when the information was presented in binary, rather than continuous, form. Also, Ss made higher estimates of correlation when they expected the variables to be correlated than when they did not expect them to be correlated. (31 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)