Examining validity in computerized dynamic assessment

Abstract Computerized dynamic assessment (CDA) posits itself as a new type of assessment that includes mediation in the assessment process. Proponents of dynamic assessment (DA) in general and CDA in particular argue that the goals of DA are in congruence with the concept of validity that underscores the social consequences of test use and the integration of learning and assessment (Sternberg & Grigorenko, 2002; Poehner, 2008; Shabani, 2012;). However, empirical research on CDA falls short in supporting such an argument. Empirical studies on CDA are riddled with ill-defined constructs and insufficient supporting evidence in regard to the aspects of validity postulated by Messick (1989, 1990, 1996). Due to the scarcity of research on CDA, this paper explores the potentials and the viability of this intervention-based assessment in computer assisted language testing context in light of its conformity with Messick’s unitary view of validity. The paper begins with a discussion of the theoretical foundations and models of DA. It then proceeds to discuss the differences between DA and non-dynamic assessment (NDA) measures before critically appraising the empirical studies on CDA. The critical review of the findings in CDA literature aims at shedding light on some drawbacks in the design of CDA research and the compatibility of the concept of construct validity in CDA with Messick’s (1989) unitary concept of validity. The review of CDA concludes with some recommendations for rectifying gaps to establish CDA in a more prominent position in computerized language testing.