Long established and well-honed assessment practices did not evolve to deal with
assessments that are complex, in the sense of interactive tasks, multidimensional
proficiencies, and complex responses to evaluate. But progress is being made on many
fronts to extend practice, as seen in the National Board of Examiners’ computer-based
simulation tasks (Clyman, Melnick, & Clauser, 1999), Adams, Wilson, and Wang’s (1997)
structured multidimensional IRT models, and White and Frederiksen’s (1998) guided selfevaluation
in extended inquiry tasks. This work succeeds because even when it differs from
traditional testing on the surface, each innovation is grounded in the same principles of
evidentiary reasoning that underlie the best assessments of the past.