[R-sig-ME] GLMM for underdispersed count data: Conway-Maxwell-Poisson and Ordinal

Simone Santoro santoro at ebd.csic.es
Fri Dec 9 19:40:50 CET 2016


Thank you all very much your hints. They have been really really helpful 
for me. Below you may find a reproducible code to see how three 
approaches fit a simulated data set (clmm::ordinal, glmmTMB::glmmTMB, 
fitme:spaMM). Results seem to me qualitatively similar but with 
clmm:ordinal I cannot use the three crossed random effects because I get 
an error like this:

Error: no. random effects (=135) >= no. observations (=100)

dati<- data.frame(fledges= rpois(100,10), habitatF= 
as.factor(rbinom(100,1,0.5)), areaPatchFath= rnorm(100), poligF01= 
as.factor(rbinom(100,1,0.5)),StdLayingDate= rnorm(100), ageFath1= 
rpois(100,3), ageMoth1= rpois(100,3), year= as.factor(rpois(100,200)), 
ringMoth= as.factor(rpois(100,200)), ringFath= as.factor(rpois(100,200)))

system.time(Fitclm<- clmm(as.factor(fledges) ~ 
# this way it works...
system.time(Fitclm1<- clmm(as.factor(fledges) ~ 

system.time(FitglmmTMB<- glmmTMB(as.factor(fledges) ~ 

system.time(FitglmmTMB<- glmmTMB(as.factor(fledges) ~ 

# This lasts much more (3-4')
system.time(Fitfitme<- fitme(fledges ~ 
= "ML"))

El 08/12/2016 a las 4:32, Ben Bolker escribió:
>     One reference that uses ordinal regression in a similar situation
> (litter size of Florida panthers) is
> http://link.springer.com/article/10.1007/s00442-011-2083-0  ("Does
> genetic introgression improve female reproductive performance? A test on
> the endangered Florida panther")
>    Not sure about the number-of-random-effects error: a reproducible
> example would probably be needed (smaller is better!)
>    Ben Bolker
> On 16-12-06 08:41 AM, Simone Santoro wrote:
>> Dear all,
>> I am trying to find an appropriate GLMM (with temporal and individual
>> crossed random effects) to model underdispersed count data (clutch
>> size). I have found several possible ways of doing that. A good
>> distribution for data like this would seem to be the
>> Conway-Maxwell-Poisson but I have not found a way of using it within a
>> GLMM in R (I have asked here
>> <http://stats.stackexchange.com/questions/249738/how-to-define-the-nu-parameter-of-conway-maxwell-poisson-in-spamm-package>  
>> and here
>> <http://stats.stackexchange.com/questions/249798/conway-maxwell-poisson-with-crossed-random-effects-in-r>).
>> I have seen that Ben Bolker suggested (here
>> <https://stat.ethz.ch/pipermail/r-sig-mixed-models/2014q1/021945.html>and
>> here
>> <http://stats.stackexchange.com/questions/92156/how-to-handle-underdispersion-in-glmm-binomial-outcome-variable>)
>> to use an ordinal model in cases like this(e.g. _ordinal:clmm_). I have
>> tried this solution and the results I obtain makes (biological) sense to
>> me. However, I wonder why but I cannot put all the three crossed random
>> effects I have in the clmm model (_Error: no. random effects (=1254) >=
>> no. observations (=854)_) whereas it is not a problem for the glmer
>> model (the no. of levels of each single random effect does not exceed 854)*.
>> Beyond that, and that's what I would like to ask you, *I cannot find a
>> reference to justify I used the ordinal model* to deal with
>> underdispersed count data (referee will ask it for sure).
>> Best,
>> Simone
>> * FMglmer<- glmer(fledges ~ habitatF * (areaPatchFath + poligF01 +
>> StdLayingDate + ageFath1 + ageMoth1) + (1|year) + (1|ringMoth) +
>> (1|ringFath), data = datiDRS)
>>      FMclmm<- glmer(as.factor(fledges)~ habitatF * (areaPatchFath +
>> poligF01 + StdLayingDate + ageFath1 + ageMoth1) + (1|year) +
>> (1|ringMoth) + (1|ringFath), data = datiDRS)
>> 	[[alternative HTML version deleted]]
>> _______________________________________________
>> R-sig-mixed-models at r-project.org  mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

Simone Santoro
Department of Ethology and Biodiversity Conservation
Doñana Biological Station
Calle Américo Vespucio s/n
41092 Seville - Spain
Phone no. +34 954 466 700 (ext. 1213)

	[[alternative HTML version deleted]]

More information about the R-sig-mixed-models mailing list