Hypotheses Evaluation Using the Bayes Factor
Workshop "Hypotheses Evaluation Using the Bayes Factor"
Date: Thursday/Friday, 4th and 5th July 2019
Venue: building E1 7, room 0.08 (building of the Cluster of Excellence "Multimodal Computing and Interaction")
- If you have teaching obligations during the workshop, we recommend you to try to postpone the teaching event or to organize a substitution for you - or to take part in this workshop but miss the ca. two hours during which you will have to teach.
- Please bring your own laptop with you in order to work with it during the workshop.
Go to the bottom of https://informative-hypotheses.sites.uu.nl/software/bain/ to download the course materials (some of the materials are referenced below). You can prepare by executing the R tutorial R-hand-on-mini-course.pdf and reading BFTutorial.pdf. The workshop will flow along the following lines (addressing psychologists, that is, concepts, applications, hands-on and no formulas):
4th July 2019, 9.30-17.00 hrs.
- In three steps, interactive with the audience, I will introduce the replication crisis, and discuss possible causes like: publication bias, questionable research practices, and hidden moderators. During the discussion the p-value and Type I and II errors will be re-introduced.
- Introduction to classical and informative (a simple example is m1 > m2 > m3) hypotheses, and the Bayes factor as a tool to evaluate them (what is the Bayes factor, what are posterior model probabilities, what is Bayesian updating).
- Lab meeting. Participants need to have R and RStudio installed on their laptop. Via RStudio (tab tools – install packages) the package bain can be installed. Participants can do the lab-meeting even if they have no knowledge of R. But, you can prepare for R with the R-hand-on-mini-course.pdf that is included with the course materials
5th July 2019, 9.30-17.00 hrs.
- Using figures (no formulae) it will be more thoroughly explained what the Bayes factor is.
- Two applications will be presented in which informative hypotheses and the Bayes factor are used: the evaluation of replication studies and the evaluation of the same set of hypotheses using multiple sources of data input (e.g., a multi-laboratory study in which not all the variables have the same operationalization).
- Interactive with the audience. Discussion meeting. First in small groups the pro's and con's of null-hypotheses significance testing and the Bayesian approach will be discussed. Then three groups will be constructed (the classicists, the Bayesians and the referees), and a structured debate will follow.
- Questions and answers, continuation of the lab-meeting.