Abstract:
|
Analyzing and interpreting interactions have been perennially challenging to statisticians and epidemiologists. In this roundtable we will discuss some provocative questions regarding interaction analysis. (1) Scale Dependence of interaction (2) Statistical versus Biological interaction (3) Consequences of misspecifying the main effect (4) Meta-analysis of interactions (5) Variable selection targeted towards interaction. Finally we will summarize the current state of art for analysis of gene-environment interactions and the handful of robust and replicable findings that exist in the literature. A good pre-reading will be several papers in American Journal of Epidemiology, Volume 186, Issue 7, 2017.
|