Skip to content

MCMCglmm query

1 message · James Higham

#
Dear Jarrod,

Thanks very much for your fast and detailed reply.

Apologies for the delay in response; I wanted to make sure I could get
these changes working before responding.

The answer to the sign question makes perfect sense - thanks a lot.

Thanks also for the tips on offsetting in MCMCglmm and the other
suggestions and corrections. I've played around with it and including
the logged offset as a covariate and setting the co-efficient to 1 with
very low variance seems to work well for both the regular poisson and
zero-inflated poisson models. Likewise, the suggested syntax for
applying the offset only to the non-zi part of the model seems to be
working well.

I believe that the variables are zero-inflated based on comparison of
predicted zero numbers under the poisson and those observed in the
dataset, but I'll take another look using the MCMCglmm checks.

Thanks again so much for your help - it's very kind of you to take the
time to ensure that so many people are able to use your package
successfully for their projects.

All the best

James


On Wed, Aug 24, 2011 at 7:40 PM, Jarrod Hadfield <j.hadfield at ed.ac.uk>
wrote: