CALIBRATION OF PSYCHOPHYSICAL JUDGMENTS - MAGNITUDE ESTIMATES

Calibration has been widely studied in judgments under uncertainty. Most of these studies show a clear overconfidence effect. In psychophysical judgments calibration has been studied in discrimination tasks and a clear underconfidence effect has been found (Bjorkman, Juslin y Winnan, 1993). Magnitude estimation judgments have not been calibrated since a quantitative response cannot be directly considered as right or wrong. This work presents a way to measure the proportion of correct responses, at an ordinal level of measurement, using two variants of the magnitude estimation technique. Calculating calibration we found that in both magnitude estimation tasks subjects show an underconfidence effect and no relationship is found between this measure and the goodness of fit measure obtained from Stevens' function, its slope, intercept and mean reaction time.