Premium
An empirical pseudo‐Bayes approach to determining a weighted‐summation for use in generalized linear models
Author(s) -
Sabo Roy T.
Publication year - 2011
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.1104
Subject(s) - bayes' theorem , mathematics , generalized linear model , statistics , linear model , econometrics , bayesian probability
Studies have consistently shown that exposure to certain pesticides can be linked to adverse health conditions in both children and adults. These effects are complicated by exposure to multiple pesticides in the form of a chemical mixture. In this manuscript a methodology is presented that simultaneously summarizes the exposure profile into a representative value, yet maintains information on the individual constituents. An empirical Pseudo‐Bayes (EB) approach is used to create a weighted‐summation of the chemical concentrations, where Monte‐Carlo methods are used to derive the weights from prior distributions based on information from the univariate relationships between each chemical and some response. This weighted summation is used to estimate a relationship between the set of chemicals and the response. A simulation study is used to compare the EB approach with two traditional approaches used in chemical exposure risk assessment, and an example is provided using pesticide data and metabolic hormone concentrations from the National Health and Nutrition Examination Survey. Copyright © 2011 John Wiley & Sons, Ltd.