Accuracy and its determinants in distributional assessment

Rating accuracy is proposed to consist of 2 components: internal and extern al accuracy. Internal accuracy is identified as the appropriate basis for e valuating the impact of rating methodology on rating accuracy. Improved ind exes of accuracy were used to study the internal accuracy of distributional assessment (DA). DA achieved moderate to high internal accuracy levels in reproducing the entire profiles, means, and standard deviations of true dis tributions. It also proved to be significantly more internally accurate tha n a conventional Likert-type rating scale in estimating distributional mean s. DA proved to be particularly accurate when ratings were elicited in the form of raw count estimates of frequencies obtained under short recall dela y conditions. Correlational findings suggested that different cognitive pro cesses and/or memory sources may underlie assessments obtained via distribu tional and conventional rating methods.