Seminar in Psychometrics
IRTorch: Item Response Theory with Python
Date and time: Monday, April 28, 2025 (16:00 PM CET)
Place ICS CAS room 318, Pod Vodárenskou věží 2, Prague 8, also on Zoom.
Item Response Theory (IRT) is a statistical framework used for evaluating test items and assessing test-taker abilities through the analysis of their responses. This talk introduces IRTorch, a Python package for modeling test data using IRT. The focus will be on the practical aspects of IRT, with an emphasis on flexible semi-parametric IRT models. A challenge in IRT is that the latent trait scale is arbitrarily defined, typically by assumption during model fitting. To address this, IRTorch implements various novel scale transformations to make the latent trait scales of fitted IRT models more meaningful, comparable and interpretable. Through theory and practical examples, I will demonstrate how IRTorch can be used for model estimation, model evaluation, and scale transformation.
References:Wallmark, J., Josefsson, M., & Wiberg, M. (2024). Introducing Flexible Monotone Multiple Choice Item Response Theory Models and Bit Scales. arXiv preprint arXiv:2410.01480.
Wallmark, J. (2024). IRTorch: Item response theory with Python. https://github.com/joakimwallmark/irtorch
Joakim Wallmark
https://www.umu.se/en/staff/joakim-wallmark/
Dr. Joakim Wallmark is a postdoctoral researcher at the Department of Statistics at Umeå University. With a strong background in programming, statistics, and machine learning, he completed his PhD in statistics in February, focusing primarily on psychometrics and item response theory (IRT). Before entering academia, he worked as a software engineer—a role that continues to influence his practical, data-driven approach. Currently, his research spans both psychometrics and healthcare, where he is developing and applying machine learning methods to analyze Swedish register data.