A0460
Title: A Bayesian spatial model for survey-based ordinal data
Authors: Ana Corberan-Vallet - University of Valencia (Spain) [presenting]
Miguel Angel Beltran-Sanchez - University of Valencia (Spain)
Miguel Angel Martinez-Beneito - University of Valencia (Spain)
Abstract: Health surveys allow exploring health indicators that are of great value from a public health point of view. These indicators are usually coded as ordinal variables and depend on covariates associated with individuals. A Bayesian individual-level model is proposed for small-area estimation of survey-based health indicators. A categorical likelihood is used to describe the ordinal data. At the second level of the model hierarchy, the cumulative probabilities of the different categories are modeled, taking into account possible covariate effects as well as spatial dependence among areas. Post-stratification of the results allows extrapolating the results to any administrative area division, even for small areas. Finally, a multivariate extension of the model is presented that allows for the joint study of the sets of response variables that are likely to be correlated.
