Supporting Valid Interpretations of Learning Progression Level Diagnoses
Assessments associated with learning progressions report student understanding in the form of learning progression level diagnoses. Valid interpretations of such diagnoses are only possible if students consistently express the ideas associated with a single learning progression level across items. Latent class analysis was employed to evaluate whether patterns of expected responses on clusters of diagnostic multiple-choice items afford valid interpretations of learning progression level diagnoses. Results indicated that students with scientifically-accurate understanding of the relationship between force and motion usually reason systematically across items, but most other students do not. Consequently, interpretations of learning progression level diagnoses on two proposed learning progressions would often be invalid. Analyses of this sort would be useful in developing and validating future learning progression.