About Ranova: what is intercept:WithinVariable means and why it varies across different model?

Matlab 2014a offers the 'fitrm' and 'ranova' for repeated measure anova. I try to use this for computing large amount of anova tests automatically and output the anova result into a form. Then I found that in the result table, the term (intercept):WithinVariable as well as other result are not consistent. This is the data: >>t = table(group(:,3),W1,W2,'VariableNames',{'X1','Y1','Y2'}); >>Meas = table([1 2]','VariableNames',{'Measurements'}); if I fit it in this way: >>rm = fitrm(t,'Y1-Y2~X1','WithinDesign',Meas); >>ranovatbl3 = ranova(rm); then

but if I fit the model in the following way; >>[ranovatbl5,A,C,D] = ranova(rm,'WithinModel','Measurements'); then

You can see the F/P values are different from what they used to be. This also happens when I try >> rm = fitrm(t,'Y1-Y2~1','WithinDesign',Meas,'WithinModel','orthogonalcontrasts'); >>[ranovatbl4,A,C,D] = ranova(rm,'WithinModel','Measurements'); Or when I use >>[h,p]=ttest(W1,W2), or when I use SPSS. The main effect of my within variable(as well as others) is different in all cases above.

I'm wondering how this happens? (see the attached files for datas)

Thank you!

[CDev]

Hi there,

Did you figure out this problem? I am in a similar position now of not fully understanding how to use fitrm and anova to carry out a repeated measures anova with no between subject factors.