Sensitivity analysis and power for instrumental variable studies.

Abstract excerpt

In observational studies to estimate treatment effects, unmeasured confounding is often a concern. The instrumental variable (IV) method can control for unmeasured confounding when there is a valid IV. To be a valid IV, a variable needs to be independent of unmeasured confounders and only affect the outcome through affecting the treatment. When applying the IV method, there is often concern that a putative IV is...

Topics

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

• Share to Bias
• Share to Biometry
• Share to Confounding Factors, Epidemiologic