I'm trying to decide if something is a confounder or not. I know that a confounder must be a risk factor for the outcome and it must also be related to the exposure...what I'm not sure about is whether it has to affect the exposure and the outcome in the same way. For example, if I was investigating the effect of heroin use on risk of developing CVS disease and I wanted to explore socioeconomic status as a confounder. If the research indicates that HIGH SES is associated with heroin use, but LOW SES is associated with CVS disease, then is SES still a confounder?
Asked
Active
Viewed 56 times
1 Answers
1
Yes, it can affect them in different ways. Draw the causal graph, and check the d-separation criterion. That's what makes it a confounder.
Neil G
- 15,219