I have a question for users of MCMCglmm that have experience implementing
the zero-inflated poisson model.
I find that the documentation, and previous questions, do not offer a lot
of clear guidance on specifying and interpreting the zipoisson model. In
particular, I see a lot of zero-inflated poisson examples that use the
at.level(trait, x):variableName syntax.
Specifically, the MCMCglmm course notes, available on the package's CRAN
page, uses the following example on page 102:
m5d.1 <- MCMCglmm(art ~ trait - 1 + at.level(trait, 1):fem +
at.level(trait, 1):mar + at.level(trait, 1):kid5 + at.level(trait, 1):phd
+ at.level(trait, 1):ment, rcov = ~idh(trait):units,
data = bioChemists, prior = prior.m5d.1, family = "zipoisson",
verbose = FALSE)
I have been unable to find an answer to the following questions and would
appreciate any guidance:
1) Does at.level(trait, 1) index the poisson latent variable, or is that
at.level(trait,2)? More generally, how does one find out what levels, and
what values, are indexed with "trait"?
2) Why is the example specified only using the trait 1? That is, why not
estimate the model in the following fashion:
m5d.1 <- MCMCglmm(art ~ trait + trait:fem + trait:mar + trait:kid5 +
trait:phd + trait:ment, rcov = ~idh(trait):units,
data = bioChemists, prior = prior.m5d.1, family = "zipoisson",
verbose = FALSE)
3) How do we properly interpret the results from the two models, say using
summary(m5d.1) ?
I would appreciate any pointers to relevant documentation/examples.
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
