Abstract

The purpose of the present paper was to demonstrate a method of estimating the probability of misclcissification when a Scale is used to classify respondents arbitrarily into the high and low groups based upon the scores on the scale. Determination of misclassification probability is made possible because Item Response Theory provides theoretically sound estimation of latent trait. In this study, misclassification was defined by the hypothetical incidents in which the subjects with the actual degree of traits below the median are assigned into the high group and the subjects with the actual degree of traits above the median are assigned into the low group based upon the scores on the scale. The estimated misclassification probability associated with Beck Hopelessness Scale was .06. The interpretive meaning of the misclassification probability as defined in this study was discussed. Finally cut-off pointe on the raw score scale reflecting those probabilities were provided for practical consideration, and important issues for further studies were suggested.