0

Can we carry out a post hoc test like (Tukey, using mult-comp) for "glmer" models for the random effects? My response variable is binomial and there are three levels in the random effect variable. Since the random effect has a standard deviation (0.3), I plan to run a post hoc test to check if there is within level differences for the effect of the random effect variable on the response variable. Also, can we carry post hoc tests for the nested random effects? I want to know post hoc for the random effect and not the fixed effect.

Harshad
  • 61
  • 6
  • It sounds like your random effect should be a fixed effect. With only 3 levels, you will not get a sensible estimate of its variance. – Robert Long Feb 10 '19 at 12:27
  • Okay. For another random effect I get the value of variance as 52.04 and 7.20 as Std Dev, does that indicate this effect is not random and in fact fixed? Is there any criterion which could demarcate random effects from mixed effects? – Harshad Feb 10 '19 at 14:02
  • 1
    Impossible to say if it should be fixed or random from that information. How many levels of it are there, and what is the variable itself ? For info about fixed vs random effects, just search this site - there are lots of posts about that. For example [this](https://stats.stackexchange.com/questions/4700/what-is-the-difference-between-fixed-effect-random-effect-and-mixed-effect-mode) – Robert Long Feb 10 '19 at 14:06

0 Answers0