Say I have a linear model:
fullmodel <-glm(outcome ~ map * pressor + group + x + y + z, data=data, family="binomial")
And then want to construct a plot of a interaction terms analysis:
graphedmodel <- emmip(fullmodel, pressor~map | group, at=list(map=seq(60,105,by=5), type = "r", CIs=TRUE)
In graphedmodel, what are the values of all the other variables for x, y, z? Are they assumed to take their mean values? Are they simply excluded from the analysis?
Not troubleshooting; the model works. I want to know how the model behaves.
It is plotting estimated marginal means, and that means that any excluded factors are averaged-over, and any excluded numeric covariates are set to their means -- unless directed otherwise by optional specifications such as at or cov.reduce. In the question, I'd guess x, y, and z are numeric, so are set at their means.