BAYESIAN PREDICTIVE INFERENCE FOR LONGITUDINAL SAMPLE-SURVEYS

We describe and apply methodology for Bayesian predictive inference th at is appropriate for many multistage, longitudinal sample surveys. Th e Patterns of Care Studies, two-stage cluster samples of cancer patien ts conducted on two occasions, are emphasized. For the application we fit a super-population model to the survey data, and provide posterior inference for the desired finite population parameters. The former ta sk includes the use of normal probability and standardized residual pl ots to assess the quality of the fit of our proposed random effect mod els. The inferential techniques take account of our use of transformed variables, and the variability associated with all nuisance parameter s.