A0416
Title: Estimation of the average treatment effect with error-prone confounders
Authors: Li-Pang Chen - National Chengchi University (Taiwan) [presenting]
Abstract: Estimation of the average treatment effect (ATE) is one of the questions in causal inference, which is usually used to measure the causal effect of a treatment on the outcome of interest. The inverse probability weight method based on the propensity score is a commonly used strategy to estimate ATE. In applications, the challenges of the estimation include high-dimensional covariates and measurement errors in datasets. Ignoring those features may eventually induce an unreliable estimator of ATE. The primary consideration is a dataset where covariates and treatments are possibly subject to measurement error, and potential outcomes have a nonlinear relationship with covariates. To tackle these challenges and derive a precise estimator of ATE, the FATE method is developed, which refers to feature screening, adaptive lasso, treatment adjustment, and error correction for covariates. The feature screening procedure is based on error-eliminated data and is valid to handle exponentially distributed outcomes. In addition, provided that misclassified treatment and measurement error in covariates are corrected, the reliable estimator of propensity score is derived with collinearity taken into account, and thus, the estimator of ATE with measurement error correction is derived. Theoretical results are established. Numerical studies also reveal that the proposed FATE method has a satisfactory performance and is better than its competitive methods.
