Mixed model with zero truncated Poisson distributed data
On 01/11/2012 04:58 PM, Helen Ward wrote:
Hello, I am a PhD student currently trying to derive a mixed effects model. I would like to describe the relationship between age and male reproductive success in a population of greater horseshoe bats. My data consists of three columns: MaleID, Age, NumberofPups (at that age). Many of the males appear multiple times in the data set, so I believe I need to derive a mixed model with MaleID as a random variable. The data is Poisson distributed, but zero-truncated. So far I have only succeeded in making a mixed model with a poisson distribution (using glmmPQL in the MASS package), and a zero truncated poisson model (using vglm in the VGAM package), but not a mixed model capable of handling zero truncated Poisson data. It has been suggested that I could just minus 1 from each value in the NumberofPups column to make a more usual Poisson distribution, so I can ignore the zero truncated bit. I have tried this and it changes the results of the model, but is this an acceptable transformation? If not, can anyone advise me on a mixed model that can handle zero truncated Poisson data please? I also intend to post this on the R-sig-mixed-models <https://stat.ethz.ch/mailman/options/r-sig-mixed-models/h.l.ward%40qmul.ac.uk> list, so apologies if you've seen it twice!
If an individual would have had pups, would you have seen it? I'm wondering if you could simply add the zeroes back into the data. Bob
Bob O'Hara Biodiversity and Climate Research Centre Senckenberganlage 25 D-60325 Frankfurt am Main, Germany Tel: +49 69 798 40216 Mobile: +49 1515 888 5440 WWW: http://www.bik-f.de/root/index.php?page_id=219 Blog: http://blogs.nature.com/boboh Journal of Negative Results - EEB: www.jnr-eeb.org