Unequal groups ANOVA (3 way, 2 rm) - not just warning, but no answers at all

[freddybrown1984]

Thanks so much for all the help RE the 3waytworm ANOVA. I finally understood how to correct the matlab path, considering the “Unrecognized function or variable 'spm_uc'” error messagae

Can I confirm, though... will this three way ANOVA not work with groups of different sizes? And there’s no work-around? I can now get all of my desired outputs with the equal sized groups, but not when they differ by n = 1.

Thanks - and again apologies if I've missed a previous instance of this question being answered

Freddy

[0todd0000]

Correct, spm1d currently does not support unbalanced designs. You can work around this by imposing balance using random subsets. Please see relevant discussions here:

• https://github.com/0todd0000/spm1dmatlab/issues?q=unbalanced
• https://github.com/0todd0000/spm1d/issues?q=unbalanced

[freddybrown1984]

Great stuff. Thanks (again) Todd!

I think I've seen some mentions of adding a power analysis into the SPM outputs. Can "observed power" be calculated for my ANOVAs?

[0todd0000]

Power can be calculated for arbitrary designs but for the 1D case there are several necessary considerations. Please see the power1d site for a few descriptions of relevant considerations.

[freddybrown1984]

Brilliant. Thanks!