Replication, Consistency of Effect, and Control of the Overall Type 1 Error Rate
View Presentation View Presentation
*Steven Snapinn, Amgen/GBE 

Keywords: Mutiplicity, Replication, Consistency of Effect

Replication is one of the main principles of the scientific method. Requiring that multiple studies each achieve statistical significance will also help control the overall type 1 error rate. In this presentation I will explore the type 1 error rate associated with a requirement that n out of m trials achieve a significant result. I will also explore the concept of consistency of effect, both within trials and between trials, and specifically the “consistency paradox.” This is the concept that a consistent treatment effect is a desirable quality of any drug, but consistency within- and between-studies actually provides evidence against the observed treatment effect being real.