A1171
Title: A bootstrap-based goodness of fit test for binary spatial models
Authors: Eva Biswas - Iowa State University (United States) [presenting]
Daniel Nordman - Iowa State University (United States)
Mark Kaiser - Iowa State University (United States)
Andee Kaplan - Colorado State University (United States)
Abstract: Binary spatial observations frequently occur in environmental and ecological studies, where Markov random field (MRF) models are commonly applied. Despite their widespread use and long-standing history, appropriate model diagnostics for spatial binary data in MRF models have remained a challenging issue. A complicating factor is the difficulty in assessing neighborhood specifications for binary data. To address this, a formal goodness-of-fit test is proposed for diagnosing MRF models for spatial binary values. The test statistic involves a type of conditional Moran's I based on the fitted conditional probabilities, which is capable of detecting departures in model form, including neighborhood structure. The application of the spatial test is illustrated using a dataset on the breeding pattern of grasshopper sparrows across Iowa.
