Marginal effects of ZINB models
On Mon, 10 Dec 2012, Jeremy Goss wrote:
Dear all, I am modeling the incidence of recreational anglers along a stretch of coastline, and with a vary large proportion of zeros (>80%) have chosen to use a zero inflated negative binomial (ZINB) distribution. I am using the same variables for both parts of the model, can anyone help me with R code to compute overall marginal effects of each variable? My model is specified as follows: ZINB <- zeroinfl(Tot.Anglers ~ Location + Season + Daytype + Holiday.not + CPUE + ShoreType + Access + Source.pop + WindSpeed + offset(beat_length), dist="negbin", data=anglers)
We haven't implemented any marginal effects for hurdle/zeroinfl because I rarely find these useful in practice. Also, you probably would need several marginal effects for the same variable because you might want to describe the effect on the zero-inflation, on the count component, and on the mixture of both. But with the building blocks provided by hurdle/zeroinfl you can compute many of the quantities that are potentially of interest "by hand". For hurdle models, there is some discussion of this in the following posting: https://stat.ethz.ch/pipermail/r-help/2012-January/300949.html Best, Z
Many thanks, Jeremy [[alternative HTML version deleted]]
______________________________________________ R-help at 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.