Table of contents

The course is connected to the following study programs

Teaching language

English

Prerequisites

Participants should be familiar with coding in R. The course is most useful for students with some background in linear modelling techniques, for example including advanced methods such as generalised and mixed models.

Learning outcomes

Upon successful completion of the course, the student should:

Knowledge and understanding

  • Demonstrate an advanced understanding of Bayesian statistics, including key concepts like prior distributions, likelihoods, and posterior distributions.
  • Explain the construction and interpretation of probabilistic models used to describe real-world phenomena, with emphasis on generative models.
  • Understand the theoretical basis for model comparison, including criteria like WAIC (Watanabe-Akaike Information Criterion), DIC (Deviance Information Criterion), and posterior predictive checks.
  • Explain and justify the use of multilevel (hierarchical) models in Bayesian inference, especially in complex data scenarios with varying levels of aggregation.

Competence and skills

  • Develop and implement Bayesian models in R or Stan, to address real-world data problems.
  • Perform and interpret posterior inference through Monte Carlo methods such as Markov Chain Monte Carlo (MCMC) sampling, including diagnostics for convergence.
  • Apply data simulation techniques to evaluate and validate statistical models, making predictions and generating credible intervals from posterior distributions.
  • Design and estimate hierarchical and structured Bayesian models, including the incorporation of random effects, interaction terms, or non-linearities.
  • Critically assess and refine models using posterior predictive checks, model comparison tools, and sensitivity analysis to ensure robustness of statistical conclusions.

Judgement and approach

  • Demonstrate the ability to critically assess and select appropriate prior distributions, balancing subjective knowledge with empirical evidence.
  • Develop a thoughtful, principled approach to model selection, balancing complexity and parsimony, and effectively communicating the rationale behind modelling decisions.
  • Assess the ethical implications of statistical models, particularly in the context of uncertainty, data limitations, and the consequences of incorrect inferences.

Interpret statistical results while effectively communicating the uncertainty and limitations of inferences to both expert and non-expert audiences.

Contents

The course uses the book McElreath (2020) Statistical rethinking: A Bayesian course with examples in R and Stan. Chapman and Hall/CRC with accompanying videos and exercises as a scaffold to help students learn about Bayesian statistical philosophy and the practice of building models, starting with simple models and proceeding to more complex ones.

Teaching and learning methods

The course is built around a series of seminars in which participants work through and present solutions to problems and challenges. There is a heavy emphasis on self-study and completing practical exercises, as well as presentations and discussions of own work.

Examination requirements

At least 70% attendance of seminars is required.

Examinations

Report and seminar. More detailed information will be given at the start of the course.

Graded passed / fail.

Student evaluation

The course will be evaluated in accordance with the quality system for education, chapter 4.1.
Last updated from FS (Felles studentsystem) Oct. 15, 2026 8:17:02 AM