SMLP 2026 Recap Discussion
Workshop 26 - Winter Semester 2026-27 (Draft)
Overview
This session is a recap / discussion of the Summer School on Statistical Methods for Linguistics and Psychology 2026 (SMLP 2026).
Many SFB members attended the school. We will discuss which ideas and tools are most useful for ongoing CRC projects.
Reference Material
Materials
- Presentation Slides — Interactive Reveal.js slide deck
- Presentation Source — Quarto source file for the slides
- References — BibTeX bibliography of core papers and textbooks
Related CRC 1252 Bayesian Workshops
The methods discussed from SMLP 2026 build directly upon topics covered in our CRC workshop series:
- Workshop 11: Bayesian Mixed Effects Models with brms — Setting domain-specific priors, prior and posterior predictive checks, and hierarchical modeling
- Workshop 13: Bayesian Model Comparison Using Cross-Validation — Out-of-sample predictive accuracy using LOO-CV and WAIC
- Workshop 14: Bayesian Hypothesis Testing Using ROPE — Practical equivalence boundaries and the HDI+ROPE decision framework
- Workshop 15: Bayes Factors — Evidence ratios, Savage-Dickey density ratios, and bridge sampling
- Workshop 18: Sample Size and Power Analysis — Simulation-based power analysis and Bayesian sequential designs
- brms-ws Repository — Interactive tutorial notebooks and Docker environment for local practice
Reference Materials & Textbooks
- SMLP 2026 Website
- An Introduction to Bayesian Data Analysis for Cognitive Science (Nicenboim, Schad, & Vasishth)
- Frequentist Statistics for Cognitive Science (Vasishth, Veríssimo, & Vrazitulis)
- Embrace Uncertainty: Mixed-effects models with Julia (Kliegl, Alday, & Bates)
- Data Simulation Workshops & faux (Lisa DeBruine)
- PsyTeachR Open Materials (University of Glasgow)
- Bodo Winter's recent workshops
Session Details
Date: 16 September 2026 Time: 14:00 - 15:30 Location: House of Prominence, Attic, Luxemburger Str. 299, Cologne
Feedback from the Session
We shared the following suggestions for future sessions and follow-up materials:
- More guidance on reporting would have been helpful — e.g. how to report Bayesian analyses (priors, Bayes factors, model comparisons) in papers and pre-registrations.
- More motivation for why e.g. priors matter, including:
- the fact that the Bayes factor depends on the prior, so results are not prior-independent;
- the distinction between informative vs. default (weakly informative) priors and when each is appropriate.
- Moving from literature to your own data — translating priors and effect sizes reported in the literature into priors for one's own analyses.