Bayesian-type count data models with varying coefficients: estimation and testing in the presence of overdispersion (Q2739983)

The paper deals with Bayesian-type count data models with varying coefficients, in particular with Poisson and Negbin models. Using the Kalman filter procedure the varying coefficients are estimated. All hyperparameters are estimated as maximum likelihood estimators on the base of EM-type algorithms. The problems which arise when a Poisson model is used in the presence of overdispersion are discussed. A bootstrapping procedure is proposed to get clear of the necessity of the varying-coefficients approach against the fixed-coefficients hypothesis. Simulation studies and real data applications illustrate the proposed methodology.