I would like to run a multiple linear regression between my protein of interest (continuous) and a separate protein (continuous). I would like to adjust for a covariate/confounder which has three levels (k=3), which is physician's diagnosis (control, mild cognitive impairment, Alzheimer's disease). From what I understand, in regression analyses, categorical variables with more than two levels must be dummy-coded (k-1). In this case, since my categorical covariate has 3 levels, I would need two dummy variables. The omitted/reference group would perhaps be individuals with a diagnosis of "control". Therefore, I would just enter those two dummy variables into the regression to control for them. Is this a valid approach to adjust for this covariate/confounder?

Thank you


1 Answer 1


Yes, this is correct. Note that most statistical programs will do this for you when you specify a categorical predictor variable and you should not do it yourself (except perhaps to verify that you get the same answer).

  • 2
    $\begingroup$ The caution "you should not do it yourself" will be particularly important if the model includes an interaction term between diagnosis and the continuous predictor (as it probably should, based on the description). +1 $\endgroup$
    – EdM
    Mar 10, 2022 at 20:06
  • 1
    $\begingroup$ @EdM, yes, or if they want to do an F-test for the predictor or compute marginal means, etc. $\endgroup$
    – Noah
    Mar 10, 2022 at 20:09
  • $\begingroup$ Great, thank you @Noah $\endgroup$
    – MQ99
    Mar 11, 2022 at 2:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.