Statistical Modelling 4 (2004), 127143
A spatial model with ordinal responses for grazing impact data
Mark J. Brewer
Biomathematics and Statistics Scotland,
and Macaulay Institute,
Craigiebuckler, Aberdeen AB15 8QH,
Scotland.
eMail: m.brewer@bioss.ac.uk
David A. Elston and Matthew E.A. Hodgson
Biomathematics and Statistics Scotland
Aberdeen,
Scotland.
Anneke M. Stolte, Andrew J. Nolan and David J. Henderson
Macaulay Institute, Craigiebuckler,
Aberdeen,
Scotland.
Abstract:
We propose a model for use with ordinal spatial data arising
from field assessments of the grazing and trampling impact
by animals on vegetation, and study the predictive performance
of the model on partial surveys. We employ a mixed effects
model, including a term for spatial correlation, which
assumes a continuous underlying scale of grazing impact,
and where the classification into discrete
categories is made via cut-points. We analyse two classes
of data: full census data and sample data drawn from the
full census. In the latter case, we show that the estimation
of nonsampled data improves as the spatial information
included within the model increases.
Keywords:
assessment of model predictions; Bayesian spatial modelling;
grazing animals; Markov chain Monte Carlo; ordinal response
Downloads:
Color maps
Data and
WinBUGS code in zipped archive
See http://www.mrc-bsu.cam.ac.uk/bugs/
for information on WinBUGS.
back