#284: Reinventing Measurement Models for the Learning Sciences: Reconceptualizing Psychometrics to Accommodate Context as Intrinsic to Learning

This paper introduces an approach to measurement that positions context and situativity as an intrinsic part of the measurement model in a manner compatible with the learning sciences’ view of learning as a transactive process within a broader sociocultural ecosystem. This approach relies on a mixed methods approach to measurement in which qualitative data are integrated into a moderated nonlinear factor analytic (MNLFA) framework. Using data from a longitudinal student of Ph.D. students in the biological sciences, we demonstrate that sensemaking context elicited through interview data could be incorporated directly into the model to enhance the comparability of scale scores collected across multiple timepoints. On multiple items that demonstrated item parameter drift (IPD) without moderation, the inclusion of a relevant idiographic contextual variable permitted the fitting of a moderated model that eliminated IPD.