Cristian Echeverria
2004-Aug-06 16:25 UTC
[R] Generalised least square with binomial errors ?
My dataset consists of maps of deforestation in Chile (in GIS), and I want to analyse whether distance from roads, soil type, altitude etc are important determinants of deforestation rate. I had hoped to do this using GLS models, to handle spatial autocorrelation, but I have only been able to find code which handles normally distributed data, whereas I have binary data (deforested Y or N). Please, has anyone written code for GLS modelling with a binary response variable? Thanks Cristian Echeverria Graduate Student Conservation & Community Ecology Group Department of Plant Sciences University of Cambridge Downing Street Downing Site Cambridge CB2 3EA UK [[alternative HTML version deleted]]