[R] Tukey and extracting letters in multcomp

Lauren Moscoe moscoe at wisc.edu
Thu Jan 14 22:34:40 CET 2016


I have the following model:

model <- lmer (Y  ~ gen + (1|env) + (1|gen:env))

I would like to use Tukey's method to identify significant pairwise differences among levels of the factor gen, which has 18 levels. (Env has 3 levels.)

I have been trying to do this using glht and cld in the package multcomp.

This is what I ran, with notes about the output.

tukey.gen <- glht(model, linfct=mcp(gen = "Tukey))


# This resulted in 46 warnings that said, "In RET$pfunction("adjusted", ...) : Completion with error > abseps."

cld(tukey.gen, level=0.05)

# This was my attempt to extract letters to represent Tukey results. I did get letters for each level of gen, but they came alongside the same warning as above, repeated over 50 times.

Why am I getting these warnings? Should I not use the results of cld, given these warnings? How can I avoid the warnings?

I would be very grateful for any advice for how to successfully use Tukey's method for gen in this model, as well as for how to extract letters to denote significant differences. Either suggestions for adjusting my code or for a new approach entirely would be fine.

Thank you,


PhD Candidate

Department of Botany

University of Wisconsin-Madison

	[[alternative HTML version deleted]]

More information about the R-help mailing list