This might be a kind of basic question in confounding in logistic regression, but would still appreciate some help.
I coded (in SAS) a logistic regression model were my predictor (religiosity) is known to be a protective factor of the outcome (substance use).
As expected, the odds ratios show that the more religious you are the less your risk of substance use. I am now running a similiar model adjusting for potential confounders. not having a psychiatric disorder is also a protective factor against substance use and therefore may be a confounder. I included this variable in my model.
My question referes to the reference level that I write in my code for this variable in order to make sure that I am in fact controlling for it. currently my psychiatric disorder variable (confounder) is coded "0=no", "1= yes". Does it matter which reference level I provide in my SAS code when I examine for confounding. I used "0" as the reference level, but I am trying to understand if I should have used "1".
Any help would be appreciated.
Thank you