Semiparametric Bayesian Analysis for Additive and Non-Additive Measurement Errors with an Application to the Nhanes III Study

• Activity Number: 321
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Committee on Applied Statisticians
• Author(s): Jingang Miao*+ and Samiran Sinha and Suojin Wang
• Companies: Texas A&M and Texas A&M and Texas A&M
• Keywords: Measurement error ; Logistic regression model ; NHANES III data ; Interaction ; Regression calibration ; SIMEX

Abstract

In nutritional epidemiological studies, nutrient intakes are usually measured via food frequency questionnaires. The measured nutrient intakes can involve substantial amount of noise. While additive measurement error is a reasonable assumption for the main effect of the nutrient intake, this assumption is not tenable for the interaction effect of two nutrients measured with errors. Although there are a number of methods for handling additive measurement errors in covariates in the logistic regression model, relatively less attention has been paid to handling nonadditive measurement errors. Therefore, we propose a semiparametric Bayesian method for handling both additive and nonadditive measurement errors in a logistic regression model. Mimicking the scenario of the Third National Health and Nutrition Examination Survey data (NHANES III), the proposed method is also designed to handle partially missing values for the error-prone surrogate variables. Through simulation studies we assess some operating characteristics of the proposed method, compare it with simulation extrapolation and the regression calibration method, and apply it to the NHANES III data.