Poster #5 - Generalized Propensity Score Estimation for Continuous Treatments With Skew T Distribution Regression Model

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

Abstract

Generalized propensity score (GPS) was used to remove selection bias when there is no random selection with continuous treatment to estimate treatment effects. This method based on weak unconfoundedness assumption and using normal linear regression to estimate the conditional Gaussian density as GPS for each value of treatment given the covariates to remove selection bias. However, in some situation, the continuous treatment variable was highly skewed from normal distribution. Therefore, using skew t distribution could be an alternate way to achieve more accurate estimations.

Authors

• Zeyuan Jing, University of Florida

• Walter L. Leite, University of Florida

• Huan Kuang, University of Florida