SENSITIVITY AND BIAS IN COVARIATION DETECTION - A DIRECT APPROACH TO A TANGLED ISSUE

Signal detection theory was used to examine the effects of sensitivity and bias in covariation detection. On each trial, participants judged whether a sample of paired data was drawn from a correlated or an unc orrelated population. Average sensitivity was suboptimal compared to a n ideal observer, and performance was at chance levels for population correlations less than rho = .30. The prior likelihood of encountering a sample drawn from a correlated population was also manipulated, res ulting in higher proportions of false positive and false negative erro rs than were expected on the basis of a Bayesian classification rule. These biases did not hinder sensitivity, in contradiction to Ahoy and Tabachnik's (1984) theory of covariation detection, nor did they enhan ce sensitivity, in contrast to Wright and Murphy's (1984) finding that biases can facilitate the detection of covariance. Moreover, although these biases were somewhat more extreme than Bayes' theorem would pre dict, there was a tendency for observers to shift their decision crite ria optimally as a function of the degree of the signal-plus-noise pop ulation correlation. (C) 1997 Academic Press.