ABSTRACT
This chapter treats general issues related to the use of goodness-of-fit measures, residual diagnostics and the use of posterior output to yield risk exceedance probabilities. Goodness-of fit-criteria vary depending on the properties of the criteria and the nature of the model. In conventional generalized linear modeling with fixed effects, the deviance is an important measure. Gelfand and Ghosh proposed a loss function based approach to model adequacy which employs the predictive distribution. The approach essentially compares the observed data to predicted data from the fitted model. The analysis of residuals and summary functions of residuals forms a fundamental part of the assessment of model goodness-of-fit in any area of statistical application. Exceedance probabilities are important when assessing the localized spatial behavior of the model and the assessment of unusual clustering or aggregation of disease.
