Interactions Between Learner Assessment and Content Requirement: A Verification Approach

Abstract

A practical constraint in the design and development of algorithms and tools for personalized learning is the need to implement adaptive algorithms, oftentimes within complex software environments, without the benefit of a priori large-scale user testing. The lack of such testing makes it difficult to ensure that lessons and guidance from design recommendations and prior studies in other domains has been effectively applied in the training application. This paper summarizes efforts toward a testbed to support verification of adaptive training designs. The testbed operationalizes evidence-based guidance from the research literature and simulated students to enable exploration of design space prior to large-scale implementation. The paper motivates the approach with a specific design question, which is to examine trade-offs between the use of behavioral markers to assess proficiency and the resulting training-content requirements to take advantage of the information that such markers provide.