Bayesian Posterior Predictive Checks for Complex Models
نویسندگان
چکیده
In sociological research, it is often difficult to compare nonnested models and to evaluate the fit of models in which outcome variables are not normally distributed. In this article, the authors demonstrate the utility of Bayesian posterior predictive distributions specifically, as well as a Bayesian approach to modeling more generally, in tackling these issues. First, they review the Bayesian approach to statistics and computation. Second, they discuss the evaluation of model fit in a bivariate probit model. Third, they discuss comparing fixedand random-effects hierarchical linear models. Both examples highlight the use of Bayesian posterior predictive distributions beyond these particular cases.
منابع مشابه
Comment: Bayesian Checking of the Second Level of Hierarchical Models: Cross-Validated Posterior Predictive Checks Using Discrepancy Measures
We compliment Bayarri and Castellanos (BC) on producing an interesting and insightful paper on model checking applied to the second level of hierarchical models. Distributions of test statistics (functions of the observed data not involving parameters) for judging appropriateness of hierarchical models typically involve nuisance (i.e., unknown) parameters. BC (2007) focus on ways to remove the ...
متن کاملPosterior Predictive Checks for Model Assessment in Occupancy Modeling
Though methodology and motivation for the use of posterior predictive checks has been introduced in some textbooks (e.g. Gelman et al. 2004), the technique is not commonplace in, for example, applied ecological studies using Bayesian techniques. In many studies using Bayesian techniques, the only semblance of model checking is the use of DIC for model selection, but this does not provide practi...
متن کاملPosterior predictive checks can and should be Bayesian: comment on Gelman and Shalizi, 'Philosophy and the practice of Bayesian statistics'.
Bayesian inference is conditional on the space of models assumed by the analyst. The posterior distribution indicates only which of the available parameter values are less bad than the others, without indicating whether the best available parameter values really fit the data well. A posterior predictive check is important to assess whether the posterior predictions of the least bad parameters a...
متن کاملPosterior Predictive Outlier Detection Using Sample Reweighting
In a Bayesian model, we de ne an outlier as an observation which is \surprising" relative to its predictive distribution, under the model, given the remainder of the data. Hence \outlyingness" can be measured by the posterior predictive p-value of any interesting scalar summary of the (possibly multivariate) observation. For this calculation, we exclude the case of interest from the data, analo...
متن کاملBayesian Model Selection
Traditionally, Bayes factors, posterior odds and posterior model probabilities are used in Bayesian model selection. This approach has, however, the problem that the conclusions are often too sensitive to prior specifications. Another approach is to use discrepancy measures in model comparison and posterior predictive checks in the assesment of model adequecy. These approaches are briefly summa...
متن کامل