Semiparametric Double-Robust Estimation for Continuous Treatment Effects: A Case Study of the Effects of Area Deprivation on Child Pedestrian Road Casualties

• Activity Number: 187
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Social Statistics Section
• Author(s): Daniel Graham*+ and Emma McCoy and David Stephens
• Companies: Imperial College London and Imperial College London and McGill University
• Keywords: causal ; double-robust ; semiparametric ; pedestrian ; casualty

Abstract

This paper constructs a semiparametric double robust (DR) estimator for continuous treatment effects. An outcome regression (OR) model is augmented with a set of inverse propensity score (PS) covariates to provide separate bias correction estimating equations for distinct strata of the treatment. These are used to obtain consistent DR points estimates of average treatment effects at various doses. A penalized spline regression is then fitted to these point estimates to derive a semiparametric approximation to the continuous dose-response. The bootstrap is used for inference. Analytical results and simulations show that this DR model can provide a good approximation to linear or nonlinear dose-response functions under various sources of misspecification of the OR or PS models. The estimator is applied in a longitudinal case study of the effect of area deprivation on the incidence of child pedestrian casualties (CPCs) in British cities.