We are working on a BOLD fMRI study with a visual stimulation block design. Participants get to see 4 visual stimuli categories in blocks.

Each participant belongs to one out of three "experimental" groups.

The variable of interest/DV is the peik height of the hemodynamic response measured by fMRI.

This peik height value for every subject is acquired for a range of brain structures.

Therefore I assume the following RM-ANOVA design:

  • Dependent variable: peik height
  • between subjects: group (3 levels)
  • within subjects/repeated factors: stimulus (4 stimuli), brainstructure (regions of interest selected by literature)

There are no missing values. Data is organized in wide format.

My questions:

  1. Can I use an RM-ANOVA design with two repeated factors in my case?
  2. How many levels may a repeated factor have? Is there a limit based on sample size or something else?
  3. Since there is no specialised post-hoc for RM-factors: may I use the ordinary paired t-test?

1) Yes- you will need to make sure each factor is appropriately put into the model depending on your stats program. FSL will allow you to specify natively, however.

2) Will let others answer who may know better, but there likely is with sample size being an important consideration. However, if you have a typical n in MRI studies (~30), you should be okay with your current model.

3) Paired-T tests are used for post-hoc RM-ANOVA comparisons. However, you will need to account for multiple comparisons using Bonferroni, Bonferroni-Holm, Tukey corrections.

  • $\begingroup$ For 1), linear mixed effects modelling is also a possibility, which easily enables post-hoc comparisons. $\endgroup$ – Isabella Ghement Jan 30 '19 at 15:50
  • 1
    $\begingroup$ Agreed @IsabellaGhement. In my MRI that is what I used but phrased as RMANOVA per the question. $\endgroup$ – Matt Wittbrodt Jan 30 '19 at 23:31

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.