Solved – Post-Hoc pairwise comparison for slope (interaction term) in R

I have been using scripts from package "Agricolae" but realised today that it compares only between the intercepts.

  1. I expect that since i am testing for the interaction term using ANCOVA, the post-hoc test should do that as well instead of being a more general comparison. Or am i wrong?
  2. Are there are scripts/methods that does pairwise comparison for "slope" and "intercept"??
  3. If Scheffe is not available, what is the next best option for unequal sample sizes?

If you are using R to do the analysis, the emmeans package has an emtrends function that estimates estimated marginal slopes. e.g.,

library(emmeans) emtrends(model, pairwise ~ treatment, var = "dose") 

The package's vignette on interactions, which is viewable here, gives some examples and more discussion.

Similar Posts:

Rate this post

Leave a Comment