When doing ANOVA test for repeated measures I have problems with post-hoc analysis. To do this kind of analysis I use ezANOAVA or aov [1,2,3] or mixed models [4,5] (better when there are missing data and if it is more robust in general). The problem starts when selecting the post-hoc analysis for the significant factors and especially with the models with more than one factor and their interactions in both cases: repeated measures ANOVA and lMER. I know that I can use friedman post hoc [6] for repeated measures with one factor and multcomp package [7] for lmer with another factor.

But the question is: is there a solution in R to do post hoc analysis for interactions? Or do I have to apply paired t.test and forget the subject's variability in order to explain the differences.

Here is my code for these situations, if somebody can improve it, I would be very grateful.

Thanks for your attention and best regards

Javier.

#####################################################################

# The code

#####################################################################

#####################################################################

# If not it is problem to you install new packages write TRUE

I_WANT_TO_INSATALL_NEW_PACKAGES