A tutorial on item response modeling with multiple groups using TAM
Journal article › Research › Peer reviewed
Publication data
| By | Purya Baghaei, Alexander Robitzsch |
| Original language | English |
| Published in | Educational Methods & Psychometrics, 3, Article 14 |
| Pages | 1-14 |
| ISSN | 2943-873X |
| DOI/Link | https://doi.org/10.61186/emp.2025.1 |
| Publication status | Published – 01.2025 |
| Keywords | partial invariance, RMSD, Multiple-group item response theory (MGIRT), Haberman linking method, Stocking-Lord linking method |
Multiple-group item response theory (MGIRT) is the standard psychometric model for the analysis of large-scale assessments in education. In this tutorial, a short nontechnical introduction to MGIRT is first provided. Second, the R package TAM and the included relevant functions to estimate MGIRT are introduced and applied to a small data set from the science assessment of TIMSS 2019 for Grade 8. Specifically, the following techniques are illustrated: the estimation of population parameters, analysis of differential item functioning (DIF; also referred to as non-invariance), and linking under full non-invariance. The hands-on training in this walkthrough enables researchers to estimate MGIRT with TAM and interpret the outputs confidently.
