In this paper, the problem of combining information from different data sources is considered. We focus our attention on spatially misaligned data, where available information (typically counts or rates from administrative sources) refers to spatial units that are different from the ones of interest. A hierarchical Bayesian perspective is considered, as proposed by Mugglin et al. in 2000, to provide a fully model-based approach in an inferential, and not only descriptive, sense. In particular, explanatory covariates are arranged to be modeled according to spatial correlations through a conditionally autoregressive prior structure. In order to assess model performance and its robustness we generate artificial data inspired by a real study and a simulation exercise is then carried out.
BAYESIAN HIERARCHICAL MODELS FOR MISALIGNED DATA: A SIMULATION STUDY
ROLI, GIULIA;RAGGI, MERI
2015
Abstract
In this paper, the problem of combining information from different data sources is considered. We focus our attention on spatially misaligned data, where available information (typically counts or rates from administrative sources) refers to spatial units that are different from the ones of interest. A hierarchical Bayesian perspective is considered, as proposed by Mugglin et al. in 2000, to provide a fully model-based approach in an inferential, and not only descriptive, sense. In particular, explanatory covariates are arranged to be modeled according to spatial correlations through a conditionally autoregressive prior structure. In order to assess model performance and its robustness we generate artificial data inspired by a real study and a simulation exercise is then carried out.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
