I'm designing an experiment which has to factors (A,B) with respectively 4 and 5 levels.

Since this is quite a large amount, I figured I use a factorial design. However, it is unclear whether the treatment will work at all, at any level or factor.

So I designed the following experiment:

All participants perform a test that is a negative control (no additions/treatment) and then those same participants perform a test with a randomly assigned level of each factor.

This allows me to test if the treatment at any level has any effect (T-test) and then check which combination of factor/level specifically is effective (ANOVA).

I've been looking for a proper resource on this, I assume this is a factorial design, but the addition of the control over all participants before the random treatment makes it hard to find the proper title/resource on this design.

Can anyone point me in a proper direction?


1 Answer 1


It is a repeated-measures factorial design, albeit of a very simple sort. You have one "within-subject" factor (pre-test and post-test (after treatment) and two "between-subjects" factors (A and B).

But these data don't need any sort of repeated-measures analysis. Simply take the differences (I'd do it as $$y_{i,j, k} = w_{i, j, k} - x_{i,j,k}, $$ where $y$ is the score to analyze, $w$ is the post-test score and $x$ is the pre-test score.

The $y_{i,j,k}$ can be analyzed as a two-way treatment structure in a completely randomized design structure.

  • $\begingroup$ thanks. I gather that i is the ith participant and j and k the respective levels? $\endgroup$ Commented Jul 11, 2014 at 19:54
  • 1
    $\begingroup$ You're welcome. It doesn't matter how you assign the indices as long as you do it consistently. $\endgroup$
    – Dennis
    Commented Jul 11, 2014 at 20:12

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.