I have browsed around and have found mixed partial answers so I decided to give my exact situation and see if you guys could help me (really appreciated).
My data has 3 columns:

  • Before treatment (numeric)
  • After treatment (numeric)
  • Active (factor: Y/N, up to 3 actives in other cases)

I measure the response before and after treatment (which makes me think of paired t-test), but ultimately I want to know if the presence of a particular active made an impact vs placebo (or "no active" group), which makes me think of ANOVA.
I've read something about two-way repeated measures ANOVA but it doesnt quite apply since I cannot use the same subject for active vs. no active. Each subject has a before/after treatment but they are either in the active or no active groups.
So at this point I am a bit confused, should I run ANOVA on the (before-after) differences or is there some sort of "paired data" test for this kind of situation.

  • $\begingroup$ Do you know t-test is a generalised version of ANOVA? $\endgroup$ – SmallChess Oct 26 '15 at 13:23
  • 1
    $\begingroup$ @StudentT It's the other way around; the t-test is a special form of ANOVA. It's anova that's the more general. $\endgroup$ – Glen_b Oct 26 '15 at 14:38

I can see that you have two binary factor variables, one of which is obviously paired. In this case, the assumption of independence in ANOVA 2-way is violated, and therefore you shouldn't use it. You should try the ANOVA 2-way mixed model.

There is a nice explanation:


  • $\begingroup$ Thank you for the reference @Student T, it offered a very nice explanation indeed and I was able to produce a nice R output that helped me understand the effect of my variables. $\endgroup$ – JanaC Oct 28 '15 at 12:51

Your Answer

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

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