Let me come back to this problem. I use JAMOVI with my students and JAMOVI uses the ‘afex’ package for anova. With repeated measures, if the variances are not homogeneous (sphericity test), Maxwell, Delaney, and Kelley (2017) consider (pp; 638-643) : “In general, when sphericity fails to hold, some pooled variance intervals will be too narrow and others will be too wide, neither of which is desirable. Using separate variance intervals immediately solves this problem, and thus is our recommendation except in cases where there is strong support for the sphericity assumption. “.
As a result, it would be convenient to have this option in this pckage (so that it is integrated into JAMOVI)
Best,
Hervé
