[R-sig-ME] Concern/question regarding discrepancy between glmmADBM summary output and glht comparisons output

[Dalal Hanna]

Hello,

I am new to this mailing list and also to glmms.
I am writing concerning a discrepancy I observed between glmmADMB summary
output and and glht comparisons which made me want to enquire about the
"best" way to proceed for an analysis.

As in this thread
<https://stats.stackexchange.com/questions/230734/post-hoc-output-and-glmm-output-dont-add-up>
 I'm observing different results between a negative binomial glmm summary
model output and glht  post-hoc comparisons.  I have pasted reproducible
code below.
I did a little digging online, and found this
<https://www.researchgate.net/post/Multiple_comparisons_in_GLMMs>post
<https://www.researchgate.net/post/Multiple_comparisons_in_GLMMs> which
lead me to believe this has to do with multiplicity. (When I use an
unadjusted test, as the answer in the post suggests, I find the same results
as in the model summary).
Still, I find myself wondering if in general its better to use the glht
function that corrects for multiple comparisons (and doesn't match up with
the model outputs), or if I should use the glht function that doesn't correct
for multiple comparisons.
Based on some posts (example
<https://stats.stackexchange.com/questions/204741/which-multiple-comparison-method-to-use-for-a-lmer-model-lsmeans-or-glht>)
I've read I suspect that its 'better' to correct for multiple comparisons
to avoid Type 1 error, but I am not sure about this given the mismatch with
the model output and wanted to ask before moving forward.
I have not encountered this problem before as I have only previously used a
model selection approach with linear mixed models, and did not have to
think about p-values and post-hoc analyses for mixed models.

Thank you for your time, I really appreciate any help/clarifications on
this topic.
Dalal Hanna


###Reproducible example####
#Generate dataframe
RecreationalTrails<-c(5, 0, 0, 4, 7, 0, 0, 0, 6, 5, 0, 6, 6, 0, 0, 0, 0, 0,
0, 0, 0, 0,4, 6, 8, 0, 0, 7)
LandUse<-c("Protected", "Agricultural", "Forestry", "Unprotected Forest",
"Protected", "Agricultural", "Forestry", "Unprotected Forest",
           "Protected", "Agricultural", "Forestry", "Unprotected
Forest","Protected", "Agricultural", "Forestry", "Unprotected Forest",
           "Protected", "Agricultural", "Forestry", "Unprotected
Forest","Protected", "Agricultural", "Forestry", "Unprotected Forest",
           "Protected", "Agricultural", "Forestry", "Unprotected Forest")
Parc<-c("Monts Valin", "Monts Valin", "Monts Valin", "Monts Valin", "Fjords
du Saguenay", "Fjords du Saguenay","Fjords du Saguenay",
        "Fjords du Saguenay", "Hautes Gorges",  "Hautes Gorges","Hautes
Gorges","Hautes Gorges",  "Grands Jardins",    "Grands Jardins",
        "Grands Jardins", "Grands Jardins", "Mont Tremblant",  "Mont
Tremblant", "Mont Tremblant", "Mont Tremblant",  "Mauricie",
        "Mauricie",  "Mauricie",  "Mauricie", "Jacques Cartier",
"Jacques Cartier", "Jacques Cartier", "Jacques Cartier")

ESCombinedDataRE<-data.frame(c("LandUse", "Parc", "RecreationalTrails"))
ESCombinedDataRE <- data.frame(LandUse, Parc, RecreationalTrails)
names(ESCombinedDataRE) <- c("LandUse", "Parc", "RecreationalTrails")
ESCombinedDataRE$LandUse <- factor(ESCombinedDataRE$LandUse,
levels=c("Protected", "Unprotected Forest", "Forestry", "Agricultural"))

ESCombinedDataRE
str(ESCombinedDataRE)

#Run model
library(glmmADMB)
R_glmer <- glmmadmb(RecreationalTrails ~ LandUse+ (1|Parc),
data=ESCombinedDataRE, family= "nbinom")

#Validate model
#homogeneity
ResidR<-resid(R_glmer)
qqnorm(ResidR)
qqline(ResidR, col=2)
#Note that I am not certain that this model actually meets its assumptions.
I see here that there is
#likely a problem of homogeneity, but am not sure what alternative model I
could run for this dataset, suggestions most welcome.

#Model results
summary(R_glmer)
#Post-hoc comparisons
summary(glht(R_glmer, linfct=mcp(LandUse="Tukey")))

#Alternative post-hoc comparison without adjusted test
mc_RT <- glht(R_glmer, linfct=mcp(LandUse="Tukey"))
summary(mc_RT, test=adjusted("none"))

	[[alternative HTML version deleted]]