COMPS Group
The COMputational PSychometrics (COMPS) Group, based at the Institute of Computer Science of the Czech Academy of Sciences, advances statistical and computational methods for analyzing measurement in the social sciences. We develop new psychometric models, efficient estimation algorithms, and open-source software, including ShinyItemAnalysis (SIA), difNLR, and others. Our work combines statistical theory, simulation studies, and real data applications. We collaborate with international researchers to bring novel methods into practice.
We organized a workshop for the NCME 2026 conference
in Los Angeles, CA, USA, on
our recent book
New paper about SIAmodules and SIAtools published in Psychometrika. See more publications!
Invited Lecture by Alina von Davier (Duolingo)
We have celebrated the 50th anniversary of the Institute of Computer Science!
News
- COMPS Group is at IMPS 2026 in Seoul! Click for more...
- Tommaso Giorgi visits COMPS Group from University of Bologna, welcome Tommaso!
- Adéla Hladká received the Wichterle Award. Congratulations, Adéla!
- Professor Adrian Raftery visited COMPS Group at ICS CAS. See more invited visits and talks.
- Jun Wu received an MSCA-CZ grant (2025-27), congratulations, Jun! See more grants and awards.
- The COMPS group served as the local organizer for IMPS 2024. See more events.
- The book Computational Aspects of Psychometric Methods. With R was published at CRC Press.
- Check our new manuscripts.
Our current research
focuses on computational aspects of the following areas of psychometrics:
- Reliability, variance component models
- Latent variable models
- Differential item functioning
- Computerized adaptive testing
- Text mining and NLP for prediction of item difficulty
- Meta-analyses
Our software
offers implementations of our research. We develop the following R packages:
Our research involves simulation studies demonstrating efficacy of proposed algorithms in comparison with existing methods.