Currently looking at data in which each participant recorded their level of pain from 1-10 in terms of severity for both pre and post operation, for 3 different treatments. Such as the table below, except i have 100 participants. What kind of test(s) should i use here to see if there are differences between the effectiveness of treatments?

I thought about doing the following: Independent t tests to see if there are any differences between the treatments in post pain rating. And then paired sample t tests within each treatment condition to see if those had an effect on pain, but this would inflate type 1 error. Or possibly some type of repeated measures anova?

ID Pre Post Treatment
1 0 4 VR
2 3 7 VR
3 2 3 Music
4 6 8 None
  • $\begingroup$ Do you have relevant covariates (eg type of operation unless all patients underwent the same medical procedure)? $\endgroup$
    – dipetkov
    Aug 25, 2022 at 6:26

2 Answers 2


Because the data are ordinal (1 - 10), an ordinal regression really is best.

However, depending on what you're doing you could get away with OLS. The approach here would be to use the Pre measurement as a covariate. Regress post score on pre score as well as a binary indicator for treatment. This is known as ANCOVA. You can them compare the three treatments against one another, using an appropriate contrast and p value correction.


I think this sounds well suited for a ordinal regression problem where you'd enter the subject ID as a covariate to control for baseline responses, and then the treatment covariate. The treatment estimate should tell you whether the treatment increased or decreased the perceived pain. If you want to use a method that is a little more efficient, use a mixed model formulation of ordinal regression with the subject ID as a random intercept.


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.