Abstract:
|
Tests for interaction are used in clinical trials to find a treatment effect that differs by subgroup. In this paper, some of the statistical issues surrounding interaction tests will be discussed. It will include the subgroup-specific test, quantitative and qualitative interaction, and statistical power. Regarding statistical power on interaction test, we will discuss the misconception that there is always less statistical power for interactions than main effects. We will also discuss the misconception about the interpretability of main effects when there is an interaction.
|