Topic: Adjusted Associations and Confounding

Using Proxy Variables for Unmeasured Common Causes

Using Subject-Matter Knowledge to Select Covariates

Tracing Paths Through a Causal Diagram

The Backdoor Criterion and Candidate Adjustment Sets

Suppression and Sign Reversal After Adjustment

Selection Bias from Restricting the Analysis Sample

Sensitivity Checks for Covariate Choices

Residual Confounding from Imperfect Measurement

Reporting an Adjusted Association With Appropriate Limits

Predicting the Direction of Omitted-Variable Bias