Skip to content
Prev 15353 / 20628 Next

LMER: Visualizing three-way interaction

Dear Klemens,
I also have been asked to remove plot3d plots that I thought were
stupendous, but I guess not everybody likes them!
How about two two-way interaction plots for the regressions of continuous
variable 1 vs 2, with separate panels for the levels of the categorical
variable?
You can get fitted model values and CI's with the lsmeans function, using
the "at" argument to specify covariate values.
If the plot is too bland, you can sprinkle the raw data on top as an extra
layer of points.
Best wishes, Evan


On Wed, Apr 12, 2017 at 5:41 AM, Klemens Kn?ferle <knoeferle at gmail.com>
wrote: