What is it about?

Shows that various so-called Bayesian prior distributions for the random treatment by trial interaction in meta-analysis are not really Bayesian

Featured Image

Why is it important?

Considers the fact that the random effect variance must be related to the main effect of treatment and tentatively suggests a prior distribution to deal with it. Proposes a graphical representation of the problem.

Read the Original

This page is a summary of: Trying to be precise about vagueness, Statistics in Medicine, August 2006, Wiley,
DOI: 10.1002/sim.2639.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page