[R-sig-ME] Is glmmADMB the only package in R which can handle, a zero-inflated Poisson model for repeated measures?

[Highland Statistics Ltd]

------------------------------

Message: 2
Date: Sat, 26 May 2012 17:59:28 +0000
From: "Mueller,John Martin"<john.mueller at louisville.edu>
To: "r-sig-mixed-models at r-project.org"
	<r-sig-mixed-models at r-project.org>
Subject: [R-sig-ME] Is glmmADMB the only package in R which can handle
	a zero-inflated Poisson model for repeated measures?
Message-ID:
	<109E4F6374337D4C9153497DE5F3EEC51CA2A3E7 at EXMBX07.ad.louisville.edu>
Content-Type: text/plain

I am looking into which package to use in R to handle a zero-inflated Poisson model for repeated measures (growth curve analysis through multiple level modeling).

> From what I can tell, the lme4 package can handle the Poisson model and repeated measures but can't handle the zero-inflated Poisson model.  The pscl package can handle the zero-inflate Poisson model but can't handle repeated measures.  And that leaves the glmmADMB package as being the only package in R that can handle a zero-inflated Poisson model for repeated measure.

Is this correct?  Or is there a package in R other than the glmmADMB package that I should be considering?

And if the glmmADMB package is the only package for this purpose, can the glmmADMB handle a 3-level multi-level model (hierarchical linear model)?

The model that I am working on has TIME at level 1, FIRMS at level 2, and LOCATIONS (metropolitan statistical areas) at level 3.  The dependent variable at level 1 is a count variable (number of credit sources used by a firm) which has an over dispersion of zeros at various points in time.  Thus, the reason why I am having to use a zero-inflated Poisson model (or I could use a hurdle model).

Thanks for your help.

- JM

	[[alternative HTML version deleted]]



------------------------------



John,
You can easily do this with R2WinBUGS and then in WinBUGS. See Chapter 4 of our 2012 book
in which we analyse 2-way nested data; R code is include. Extension to your 3-way nested data would be
an extra loop, plus an extra nesting level of the RE. It is also explained how you can calculate an
intraclass correlation for each level.

The choice between ZIP and hurdle depends on your underlying question.

Alain


-- 

Dr. Alain F. Zuur
First author of:

1. Analysing Ecological Data (2007).
Zuur, AF, Ieno, EN and Smith, GM. Springer. 680 p.
URL: www.springer.com/0-387-45967-7


2. Mixed effects models and extensions in ecology with R. (2009).
Zuur, AF, Ieno, EN, Walker, N, Saveliev, AA, and Smith, GM. Springer.
http://www.springer.com/life+sci/ecology/book/978-0-387-87457-9


3. A Beginner's Guide to R (2009).
Zuur, AF, Ieno, EN, Meesters, EHWG. Springer
http://www.springer.com/statistics/computational/book/978-0-387-93836-3


4. Zero Inflated Models and Generalized Linear Mixed Models with R. (2012) Zuur, Saveliev, Ieno.
http://www.highstat.com/book4.htm

Other books: http://www.highstat.com/books.htm


Statistical consultancy, courses, data analysis and software
Highland Statistics Ltd.
6 Laverock road
UK - AB41 6FN Newburgh
Tel: 0044 1358 788177
Email: highstat at highstat.com
URL: www.highstat.com
URL: www.brodgar.com