# Generalized linear model with random effects for skewed data

I'd like to use SPSS Generalized Linear Model to analyze a dataset of insects collected from one particular species of vegetables.

I have following variables:

NUMBER (number of insects collected)

SITE COVERED (true/false) - some of the sites were covered with plastic in order to create dry and wet conditions there

STAGE OF DEVELOPMENT (1,2,3) - stage of development of a vegetable in particular site (fresh, mature, old)

SITE NUMBER (1-10)

80% of data points are zeros

I want to check if SITE COVERED and STAGE OF DEVELOPMENT affect NUMBER, but also make sure that SITE NUMBER has no influence on NUMBER.

My model is SITE COVERED + STAGE OF DEVELOPMENT + STAGE OF DEVELOPMENT*SITE COVERED

But how am I supposed to include SITE NUMBER (as a random effect) in order to get a full picture of interactions? As far as I know, zero-inflated Poisson model should work out here. But how to include a random factor in SPSS?

Please correct me if I'm wrong in my suggestions.

P.S. also I'm considering use of R, since it has some packages for zero-inflated models, however, I didn't find allowing easily to include a random factor

What you're asking for is sometimes called a generalized linear mixed model (GLMM). I don't know about SPSS, but in R you can estimate them with any of the nlme, lme4, or MCMCglmm packages. Ben Bolker has written a very nice demonstration of using MCMCglmm to fit a zero-inflated Poisson model, available here.