Poster #7 - Improving the Precision of Bounds for Generalization

• In Event: 47.099 - MTCC Poster Session 11
  In Poster Session: 47.099-1 - Division D Section 2 Statistical Theory and Quantitative Methodologies

Abstract

Policymakers have grown increasingly interested in how the results from an experimental study generalize to a population of interest. However, because many studies do not select samples randomly, generalization requires sampling ignorability which is often controversial in practice. This study considers a bounding framework on the population average treatment effect and assesses the extent to which bounds are tightened using propensity score stratification. Using a simulation study, we consider the effect of the predictive strength of the covariates in the propensity score and outcome model and the distributional overlap in propensity scores on the bound width. We conclude with an illustration based on two case studies.

Authors

• Wendy Chan, The University of Pennsylvania

• Japbir Gill, Orleans Parish School Board