Evaluating Score and Decision Consistency Across Claims in a Validation Argument

This study utilizes an argument-based approach to validation to examine the implications of reliability in order to further differentiate the concepts of score and decision consistency. In a methodological example, the framework of generalizability theory was used to estimate appropriate indices of score consistency and evaluations of the likelihood of decision errors based on the design of a performance assessment and its intended use. The study illustrates how generalizability theory can be applied to address various claims about consistency when decisions are based on two or more cut scores, and results underscore the importance of considering score and decision consistency separately.