Improving sensitivity to change in self-report items: The role of alignment and item formulation in a longitudinal evaluation

Abstract

The capability of scales to detect change, i.e. sensitivity to change, is important in longitudinal studies. We present a quantitative study comparing the effects of alignment to course content and item granularity (generic vs. concrete wording) on the capability of different scales to capture changes in ICT self-efficacy. In an evaluation of a course on teaching with ICT, we used ICT self-report questionnaires for pre-service teachers at the beginning and end of a one-semester in a treatment and a control group (N_intervention = 278 vs. N_control = 395). After propensity score matching on the imputed data, we applied descriptive, frequentist, and Bayesian methods to examine differential scale behavior. Results show more pronounced longitudinal changes for scales on instructional ICT self-efficacy, especially for scales formulated for concrete tasks. Our results suggest that better alignment and specific formulation enhance sensitivity to change in self-reports, and relate to scale construction and selection in educational evaluations.