I have data on patients before and after an intervention, half have the treatment, half did not. A paired t-test seems a simple way to determine if there is a significant difference after treatment in either group. However there are other variables that I'd like to adjust for, such as age. Could I use linear regression, with the final measurement as the dependent variable, and the initial measurement, treatment group, and age as independent variables, and then see if the 'group' dummy variable is significant?

I see in some papers that ANCOVA is used for this application, but it seems to be mathematically identical to multiple regression. Is there any point to using it?

Lastly, one outcome variable is continuous, another is a scale from 1-10. Is linear regression appropriate for ordered categorical data like this, or is there a method that covers the gap between logistic and linear regression that lets me add in possible confounding variables (such as age)?

Thanks in advance!


1 Answer 1


Regression is equivalent to ANCOVA, you are right about that. So, if there is a point to using one there is a point to using the other. The typical format of the output varies, that's all.

For a dependent variable that is ordered, a good starting place is ordinal logistic regression.

Whether to use regression with the initial time point as a covariate or to use something like a multilevel model is another question; it has been discussed here in the past.

  • $\begingroup$ Thanks for the quick answer. By multilevel model you mean a mixed effects model? You're right I should have added that to the list, in this case the confounding variables would be included as fixed effects? However if I think they may also be predictive and want information about that, then linear regression is probably preferred? $\endgroup$
    – Huazhong
    Commented Feb 15, 2014 at 7:09
  • $\begingroup$ Yes that's what I meant; but I am not sure why you think that you don't get information about fixed effects in a mixed effects model. $\endgroup$
    – Peter Flom
    Commented Feb 15, 2014 at 12:32
  • $\begingroup$ The answer to why I think I don't get information about fixed effects in a mixed effects model is "gross ignorance"; I've never used that model type. I'll try both, it will be a good learning opportunity. Thanks again. $\endgroup$
    – Huazhong
    Commented Feb 16, 2014 at 0:14

Your Answer

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

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