Sunday, October 24, 2010: 9:00 AM
Ice Palace (Sheraton Centre Toronto Hotel)
Course Type: Half Day
Course Level: Intermediate

Format Requirements: The course will use a traditional lecture format. Participants should have familiarity with conventional statistical methods such as linear and logistic regression models. In additional, participants should have experience with the use of observational data for estimating the effects of treatments, exposures, and interventions.

Background: This course will introduce participants to the concept of the propensity score and its use for estimating causal treatment effects using observational data. A conceptual background for propensity-score analysis will be presented. Participants will learn how to: estimate a propensity score model; assess the specification of this model; and estimate treatment effects using the propensity score.

Description and Objectives: The course will introduce the concept of the potential outcomes framework for estimating causal effects.  We will briefly review the design and analysis of randomized controlled trials (RCTs). Participants will be introduced to the propensity score and how it can be used to design observational studies with many of the characteristics of RCTs.  We will discuss how to estimate the propensity score using observational studies.  Methods for assessing whether the propensity score model has been correctly specified will be discussed and illustrated using sample data.  The four different propensity score methods: propensity-score matching, stratification on the propensity score, inverse probability of treatment weighting, and covariate adjustment using the propensity score will be discussed and compared.  Sensitivity analyses for propensity score analyses will be discussed. Finally, the use of propensity score methods will be compared with the use of conventional regression adjustment for estimating treatment effects.

The objectives of the course are:

  • To understand the concept of the propensity score.
  • To learn how the propensity score allows one to design an observational study to mimic some of the characteristics of a randomized experiment.
  • To learn different propensity score methods for estimating causal treatment effects using observational data.
Course Director:
Peter C. Austin, Phd