Methods for Evaluating Composite Reliability, Classification Consistency, and Classification Accuracy for Mixed-Format Licensure Tests.

The purpose of this study was to propose extensions of reliability estimation methods that could be used to determine the conditions under which single scoring for constructed-response () items is as effective as double scoring in mixed-format licensure tests. Multivariate generalizability theory methods traditionally used to estimate overall composite score reliability were extended with simulations so that classification consistency and classification accuracy estimates could also be obtained. Composite score reliabilities, classification consistencies, and accuracies were estimated based on the double and single scoring of the items of three licensure tests. Composite score reliabilities, classification consistencies, and accuracies were also estimated in decision studies considering varied testing situations such as different numbers of items and different section weights.