Some issues in modelling biodiversity using spatially modelled covariates

Statistical models have enhanced the understanding of the relationship between biodiversity and the environment. Typically, some sort of regression analysis is performed where physical variables are covariates. It is frequently the situation that the covariates are not observed; they are spatial predictions. This study indicates that this process may bias the statistical distribution and the resulting parameter estimates if the variance of the predictions is ignored.

Document type: 
Posters and banners