18 August 2023
A Gentle Introduction to Bayesian Statistics
This course describes the stages involved in Bayesian analysis: specifying the prior and data models, deriving inference, model checking and refinement. We discuss prior and posterior predictive checking, and selecting a technique for sampling from a probability distribution. Other topics discussed are: approximate measurement invariance (a Bayesian method to assess comparability of data), evaluating hypotheses via the Bayes Factor and information criteria, and combining evidence from multiple studies addressing the same research question. Finally, we propose strategies for reproducibility and reporting standards, outlining the WAMBS-checklist (when to Worry and how to Avoid the Misuse of Bayesian Statistics).
The popularity of Bayesian statistics has increased over the years, however, as of now Bayesian methods are not a part of the statistics curricula in most graduate programs internationally. The Bayesian framework can handle some commonly encountered problems in classical statistics, such as the lack of power in small sample research and convergence issues in complex models. Furthermore, some researchers prefer the Bayesian framework because it provides a way of sequentially updating knowledge with new data instead of requiring that each new study tests the null hypothesis that there is no effect in the population.
During this course, students will be gently introduced to Bayesian statistics using many empirical examples. The instructors will clarify the differences between the philosophies and interpretations in classical and Bayesian frameworks, and will illustrate types of research questions that can be answered only using Bayesian methods. This course will also give students experience with running Bayesian analyses and interpreting results, and will instruct participants on the prevailing “best practices” for Bayesian statistics. Participants will emerge from the course with knowledge about how to apply Bayesian methods to answer their research questions, and with the ability to understand articles that examine and apply Bayesian methods for structural equation modeling.
During the first two days we discuss the stages involved in Bayesian analysis, from specifying the prior and data models, to deriving inference, model checking and refinement. We discuss the importance of prior and posterior predictive checking, and selecting a proper technique for sampling from a probability distribution. We discuss strategies for reproducibility and reporting standards, outlining the WAMBS-checklist (when to Worry and how to Avoid the Misuse of Bayesian Statistics).
During the third day we start with a brief primer on confirmatory factor analyses and how to compare CFA models across many groups (e.g., countries). Often, methods based on maximum likelihood just do not fit very well if many groups are being compared and we introduce a Bayesian alternative: approximate measurement invariance.
During the fourth day, hypothesis evaluation using (Bayesian) model selection will be discussed. We will start with an introduction to informative hypotheses, followed by an introduction to both Bayesian model selection (BMS) and model selection using information criteria (i.e., AIC and its generalization called the GORIC). It will be elaborated how BMS and the GORIC can be used to evaluate null, alternative, and theory-based hypotheses. There will be attention for the interpretation of Bayes factors, posterior model probabilities, and GORIC weights.
On Day 5, we will discuss updating a hypothesis and combining evidence from multiple studies addressing the same research question in which the evaluation of informative hypotheses using model selection plays an important role.
Please note that there is always the possibility that we have to change the course pending COVID19-related developments. The exact details, including a day-to-day program, will be communicated 6 weeks prior to the start of the course.
Dr. Rens van de Schoot
Knowledge of regression analysis and factor analyses is required. No previous knowledge of Bayesian analysis is assumed.
You do not need to know matrix algebra, calculus, or likelihood theory. Since the course offers a gentle introduction there are hardly any formulas used in the lectures. The main focus is on conceptually understanding Bayesian statistics and applying Bayesian methods to your own data set. We assume knowledge of the software package you plan to use (R, Mplus, or JAGS).
Participants from a variety of fields—including psychology, education, human development, public health, prevention science, sociology, marketing, business, biology, medicine, political science, and communication—will benefit from the course.
After engaging in course lectures and discussions as well as completing the hands-on practice activities with real data, participants will:
Explain the differences between ‘classical’ and Bayesian statistics;
Know when to use to Bayesian analyses instead of classical statistics;
Know how to apply Bayesian methods to answer their own research questions;
Know how to apply the WAMBS-checklist (When to worry and how to Avoid the Misuse of Bayesian Statistics);
Critically evaluate applications of Bayesian methods in scientific studies;
Have an idea of how to obtain prior information for their own data;
Know how to assess the comparability of data with both the exact and the approximate approach;
Know how to evaluate (theory-based) hypotheses using model selection;
Know how to combine evidence for a hypothesis from multiple studies, where the evidence within each study is obtained with model selection.
Participants will also complete the course with a foundation for future learning about Bayesian modeling and knowledge about available resources to guide such endeavors.
EUR 720: Course + course materials
EUR 200: Housing fee (optional)