Open nataliemastick opened 5 years ago
I don't think any of the built-in tests can test a single random effect in a glmm (random()
is only implemented for lmm's at the moment). Best option might be a user-specified test function:
# null model with no random effect
mod0 <- glm(eggcount~year+month+age.class, data=eggdata, family=poisson)
# user specified test function
f <- function(m) anova(m, mod0)$Pr[2]
doTest(mod, f)
Hi there, I have been having a reoccurring error with simr. I want to test a random effect in a powerSim. My model looks like this: mod<-glmer(eggcount~year+month+age.class+(1|family:ID), data=eggdata, family=poisson)
All of the help files to show how to do this suggest to test it first in doTest. The suggested code was doTest(mod, random()). I would then follow it with powerSim(mod, test=random()). However, I haven't gotten past the doTest stage, as every time I try it I get the following error: Error in exactRLRT(.) : Invalid < m > specified.
Though when I try to specify my "m", the reduced model with just the random effect, I get an error that there are unused arguments. I have tried inputting my "m" in many different ways, and it is always an unused argument.
Can anyone help me get around this issue?