Post Hoc Analysis in Mixed linear models

Question

I tested whether different version-styles of a loading screen (hourglass vs. progress bar) in different progression patterns (linear, accelerate, decelerate, irregular, binary) affect time estimations within subjects.

By analyzing the data with a binomial linear mixed effects model I have found significant results for the interaction effect of "versionStyle x DisplayDuration x progressionPattern" I would like to run a post hoc analysis to test across which condition and Display duration the time estimation was significantly affected by the version Style"

This is the code I used for my analysis:

(data1 <- glmer(Long ~ DisplayDur * Pattern * Proggression+ (1 + DisplayDur + Pattern+ Progression| Subjects), dat=anadat,family="binomial",control=glmerControl(optimizer="bobyqa")))

Is there some command that I can use in R to do this?

Answer 1

The phia package should be able to give you the tools you need. It has a very nice vignette. The testInteractions function will allow you to test the different permutations of simple effects, and it gives different options for adjusting p values.