Bayesian Hierarchical Logistic Models for Combining Field and Laboratory Survival Data

Robert L. Wolpert, William J. Warren-Hicks
Duke University

Nov 30 1990

Generalized linear regression models fit to multicollinear data sets can be unreliable for making predictions in data sets free from the multicollinearity. We overcome this problem in a study of brook trout response to acidification by constructing hierarchical Bayesian models to provide a coherent logical structure for synthesizing evidence from field observations and from related laboratory bioassay experiments, despite the great differences between field and laboratory settings and uncertainties about hose differences.


logistic regression, Bayesian hierarchical model, monte carlo integration


PDF icon 1991-06.pdf