no between-subjects main effect for mixed within- and between-subject repeated measures ANOVA (ranova)

Hi, thank in advance for any help anyone can provide. I'm working with a 2 (between) x 7 (within) factor ANOVA model. The output of ranova does not seem to provide any stats for the main effect of the between subject factor (group), which I need. See below.

t = table(grp,d(:,1),d(:,2),d(:,3),d(:,4),d(:,5),d(:,6),d(:,7),...
    'VariableNames',{'group','step1','step2','step3','step4','step5','step6','step7'});
Meas = dataset([1 2 3 4 5 6 7]','VarNames',{'Measurements'});
rm = fitrm(t,'step1-step7~group','WithinDesign',Meas);
raresults = ranova(rm);

raresults = 

                                   SumSq      DF      MeanSq        F        pValue     pValueGG    pValueHF    pValueLB
                                  ________    ___    _________    ______    ________    ________    ________    ________

      (Intercept):Measurements     0.04229      6    0.0070484    2.9598    0.007825     0.01301    0.010245    0.090513
      group:Measurements          0.032007      6    0.0053345    2.2401    0.038962    0.051168    0.045013     0.13971
      Error(Measurements)          0.85729    360    0.0023814

Am I missing something, or shouldn't there be a third F test in there for the main effect of group? How do I specify the model to get that test?

[Jared Van Snellenberg]

Seems that in this particular case I can just average over Measurements (within each subject) and run a simple between-group t to get what I'm after, but it still seems to me there should be a way to get this out of a single model run....

Answer 1

[Mike Seay]

As far as I know, there is no way to get the between-subjects main effect out of a single run of ranova(). However, if you want this information without too much work, just run anova() on the same model and check the row corresponding to 'group'. In other words, add something like this:

aresults = anova(rm);
aresults

The result should be the same or similar as with a single model-run from elsewhere, such as anova_rm() on the File Exchange or other statistical software.

Answer 2

[Dario Martelli]

Hi,

Try to use:

    raresults = ranova(rm,'WithinModel','Measurements');

It should work.