UBC Theses and Dissertations
Psychometrics & post-data analysis : a software implementation for binary logistic regression in jamovi Friesen, Lucas
Differential item functioning (DIF) analysis is an invaluable tool for providing evidence that a test or measure is functioning as intended, and has been used extensively to show evidence that the items of tests are (not) favouring one group of test takers over another. Beyond evidence (for) against differential performance of items, DIF analysis provides extremely useful validation evidence in psychosocial measures because its application need not be limited to simple cases like, for example, detecting DIF based on a binary gender classification. However, the technical knowledge requirements of DIF analysis software can be prohibitively high in some research settings. The jamovi suite “Psychometric Post-Data Analysis” has been developed with a module for performing DIF analysis on dichotomously scored items using a generalized linear model framework (GLIM) that incorporates the GLIM family of methods for detecting DIF in a user- friendly interface. In addition, the idea of Type-M errors (Gelman & Carlin, 2014) was adapted and expanded to the DIF context, and then incorporated into the module. This tool is especially useful in the context of low-powered research settings. The goal of developing this module is to encourage the use of the GLIM framework for DIF analysis in the broader psychosocial measure validation praxis by lowering the technical barrier to assessing DIF and increasing the interpretability of DIF analysis results via post-data analytic techniques.
Item Citations and Data
Attribution-NonCommercial-NoDerivatives 4.0 International