Abstract
Probabilistic unfolding models have been used for attitude measurement (Andrich in Applied Psychological Measurement 12(1):33–51, 1988, Applied Psychological Measurement 19(3):269–290, 1995; Luo in Journal of Mathematical Psychology 42(4):400–417, 1998, Journal of Mathematical Psychology 45(2):224–248, 2001; Roberts et al. in Applied Psychological Measurement 20:231–255, 1996; Applied Psychological Measurement 26(2):192–207, 2002), but they also hold great promise for evaluating rater accuracy. In this study, rater accuracy is viewed as a latent trait of raters that is affected by the essay characteristics. By applying unfolding models to the accuracy data, the directionality of inaccuracy can be differentiated. The results of this study indicated that unfolding models provide useful indices for evaluating rater accuracy.