A Bayesian Specification Analysis: An Application of Leamer's SEARCH to Air Pollution Aggregate Epidemiology
In this paper the authors examined the role that the priors of investigators have played in air pollution aggregate epidemiology. They found that he Chappie and Lave (1982) results were overwhelmingly dominated by what appear to be arbitrary priors for variables of uncertain meaning. Further, the specification uncertainties in the models they examined ware substantial relative to the sampling uncertainties of their data. The authors conclude that the aggregate epidemiology data set that Chappie and Lave employed was uninformative about the association between air pollution and mortality incidence. That is, when they considered the entire set of plausible models, the corresponding range of inferences became too large to be useful.
How to Obtain Report
Download report now
Use link to download or view the report