Statistics Seminar

Su Mo Tu We Th Fr Sa
28 29 1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31 1 2
Date/Time:Monday, 07 Mar 2016 from 4:10 pm to 5:00 pm
Location:Snedecor 3105
Cost:Free
URL:stat.iastate.edu
Contact:Denise Riker
Phone:515-294-1076
Channel:College of Liberal Arts and Sciences
Categories:Lectures
Actions:Download iCal/vCal | Email Reminder
Population causal inference: from local efficiency to global efficiency without resorting to direct approximation, Gary Chan, Associate Professor of Biostatistics and Health Services, University of Washington, Seattle

The estimation of population averaged treatment effects based on observational data is extremely important in practice and has been studied by generations of statisticians under different frameworks. Many recently proposed estimators require modeling of a propensity score function, an outcome regression function or both, and some of them are locally semi-parametric efficient. Non-parametric extensions and globally efficient estimation often requires direct approximation
of unknown functions by smoothing techniques, but their performance can be poor in practical sample sizes. Without explicitly estimating either functions, we consider a wide class calibration weights constructed to attain an exact three-way balance of the moments of observed covariates
among the treated, the control, and the combined group. The wide class includes exponential tilting, empirical likelihood and generalized regression as important special cases, and extends survey calibration estimators to different statistical problems and with important distinctions. Global semiparametric efficiency for the estimation of average treatment effects is established for this general class of calibration estimators. The results show that efficiency can be
achieved by solely balancing the covariate distributions without resorting to direct estimation of propensity score or outcome regression function. We also propose a consistent estimator for the
efficient asymptotic variance, which does not involve additional functional estimation of either the propensity score or the outcome regression functions, and is the first of its kind. The proposed variance estimator outperforms existing estimators that require a direct approximation of the efficient influence function, and accurate inference can be conducted without resorting to resampling. This is a joint work with Phillip Yam and Zheng Zhang