6533b828fe1ef96bd128830c

RESEARCH PRODUCT

IRT Modeling of Decomposed Student Learning Patterns in Higher Education Economics

subject

description

Researchers have spent decades arguing how to measure improvements in learning within formal settings, where achievements are intended and presupposed, in a reliable and valid way. Repeated measures are required to investigate improvements of any kind. Students usually take a multiple-choice test at least twice with the difference between the two measurements indicating how much they have learned. Walstad and Wagner (J Econ Educ 47:121–131, 2016) presented a new approach to gathering more information about different learning patterns by decomposing these difference measures. They describe the patterns of positive learning (PL) and negative learning (NL), i.e., the development from not knowing to knowing (PL) or from knowing to not knowing (NL). Following this approach, in this paper we present a new way to analyze PL and NL in economics based on probabilistic test theory. Using Item Response Theory (IRT), we demonstrate that the models of PL and NL fit the data well; however, there are several items that show remarkably high values for PL or NL. In particular, these items differ in their representation format and in their cognitive taxonomy level.