Data Analysis: Confounders, Mediators, and Moderators
Data Analysis: Confounders, Mediators, and Moderators
Data Analysis: Confounders, Mediators, and Moderators
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Summary<br />
Confounding is a bias you hope to prevent or control.<br />
Effect modification (interaction) is a more detailed<br />
description of the effect itself.<br />
Confounding is something to avoid.<br />
Effect modification is something to identify <strong>and</strong> report.<br />
<strong>Mediators</strong> clarify the causal pathway.<br />
Don’t control for a factor that is caused by the exposure<br />
of interest.<br />
Don’t control for a factor that is caused by the outcome<br />
of interest.<br />
Draw a conceptual model before analyzing data.<br />
References<br />
Weinberg. Toward a clearer definition of confounding. Am J Epidemiol 1993;137:1-8.<br />
Greenl<strong>and</strong>, Pearl, Robins. Causal diagrams for epidemiologic research. Epidemiology<br />
1999;10:37-48.<br />
Kaufman, Cooper. Commentary: Considerations for use of racial/ethnic classification in<br />
etiologic research. Am J Epidemiol 201;154:291-8.<br />
A Evans, March 2006 <strong>Confounders</strong>, <strong>Mediators</strong>, <strong>Moderators</strong> 4