Journal of Computerized Adaptive Testing

It was my privilege to present a keynote address to the 2014 Computerized Adaptive Testing Summit, hosted in Princeton, New Jersey, by the International Association for Computerized Adaptive Testing (IACAT).In the address, I described a taxonomy of adaptive testing schemes that Roy Levy, John Behrens, and I wrote about in greater detail in "Variations in Adaptive Testing and Their Online Leverage Points," which was published as chapter XI in David Williams, Mary Hricko, and Scott Howell's (2006) edited volume Online Assessment, Measurement and Evaluation: Emerging Practices.With the kind permission of IGI Global, the publisher, that chapter is reprinted here.The aim of the presentation was to place key ideas of computerized adaptive testing (CAT) as applied in educational and psychological measurement-namely, adaptivity, probabilistic reasoning, and knowledge-based model construction-into a broader conceptual framework.The framework draws on David Schum's (1994) principles of evidentiary reasoning and Glenn Shafer's (1976) notion of frames of discernment in statistics.Traditional every-student paper-and-pencil tests and item response theory (IRT)-based CAT are seen as exemplars in a structured space of potential adaptive assessment paradigms, with organizing concepts that help us better understand not only these familiar forms of assessment but a range of others that, despite wide-ranging surface differences, build on the same underlying principles, especially as we see supporting technologies continue to emerge.I believe that continuing developments in assessment, building on advances in computation, analytic methods, and psychometric models, and forms such as simulation-based tasks, game-based assessment, and collaborative assessment, benefit from the perspectives offered here (for examples of these developments, see Baker et al., 2017;