0
$\begingroup$

I have a continuous & categorical outcome variables in three groups of patients, where the groups are defined by time periods 2010-2011, 2012-2014, 2015-2016. I found out that age and gender are confounding variables for income, and therefore, I wish to make the age distribution of the 3 groups constant so that any difference in income cannot be attributed to age. Same as gender.

I don't think regression analysis works since we have a categorical variable (the 3 groups), but I'm just clueless. Any advice on how I should go about doing this?

$\endgroup$
  • $\begingroup$ why do you think regression cannot handle a categorical variable? $\endgroup$ – IWS Nov 28 '17 at 9:56
  • $\begingroup$ Apologies, I meant to say that, I am not sure if I could make a "study period" variable as one of the independent variables. Can I? $\endgroup$ – HNSKD Nov 28 '17 at 10:08
  • $\begingroup$ Without further context I'd say you can use time period as an independent variable. Do note that how to add time (as groups or continuous, as separate effect with or without interaction terms), depends on the exact research question. Imagine studying the effect of a certain treatment, where the contents of the treatment changed over time. In this case you might want to add time, and an interaction of time and treatment into your model. And you'd have to think about the continuousness of the time-effect (did the treatment radically change at specific moments, or gradually over time?) $\endgroup$ – IWS Nov 28 '17 at 11:10
  • $\begingroup$ If I add it as groups, should I make it ordered? or is it preferable to leave as nominal variable? $\endgroup$ – HNSKD Dec 4 '17 at 7:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.