Impact of nonrandom selection mechanisms on the causal effect estimation for two-sample Mendelian randomization methods.

Abstract excerpt

Nonrandom selection in one-sample Mendelian Randomization (MR) results in biased estimates and inflated type I error rates only when the selection effects are sufficiently large. In two-sample MR, the different selection mechanisms in two samples may more seriously affect the causal effect estimation. Firstly, we propose sufficient conditions for causal effect invariance under different selection mechanisms using...

Topics

Share this publication in a Topic to start or enrich a Post.

• Share to Bias
• Share to Causality
• Share to Genetic Pleiotropy