What is it about?
Shows how, in situations where one has data on many treatments that a priori might be supposed similar, treating the main effect of treatment as random can permit the calculation of shrunk estimates to increase robustness of predictions.
Featured Image
Why is it important?
Where it is practical to so, as say in a laboratory screening many products, this approach can be used to improve reproducibility of results.
Perspectives
This is the application of an old idea, which goes back to work by Churchill Eisenhart in 1947, to the more modern topic of network meta-analysis.
Professor Stephen J Senn
Consultant Statistician
Read the Original
This page is a summary of: Random main effects of treatment: A case study with a network meta-analysis, Biometrical Journal, January 2019, Wiley,
DOI: 10.1002/bimj.201700265.
You can read the full text:
Contributors
The following have contributed to this page
