Causal Effect Estimation Allowing Covariate Measurement Error Using Propensity Score

• Activity Number: 392
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Yi Huang*+ and Karen Bandeen-Roche
• Companies: University of Maryland Baltimore County and Johns Hopkins Bloomberg School of Public Health
• Keywords: propensity score subclassification ; finite mixture ; causal inference ; nondifferential measurement error ; average causal effect ; balance criteria

Abstract

The underlying true covariates are often measured with error in biomedical and policy studies. The naive approach is to ignore the error and use the observed covariates in current propensity score framework for ACE estimation. However, after extending the existing causal assumptions to incorporate errors-in-covariates, we showed that the naive approach typically produces biased results. In this talk, we demonstrate the newly developed finite mixture model for ACE estimation reflecting the uncertainty from measurement error in the joint likelihood. This method will estimate the propensity score subgroup membership and subgroup-specific treatment effect jointly. Its performance will be evaluated by multiple simulations studies and one application using the recent data from Infant Feeding Practice Study II.