Division D Preconference Workshop 2: Propensity Score Matching: Methods and Applications

Session Type: Workshop

Abstract

In education research, it is not always feasible or ethical to conduct experimental studies for comparing treatment effects. Non-experimental studies, such as quasi-experimental studies and observational studies, are often used instead. Unfortunately, such non-experimental studies pose threats to the validity of treatment effect estimation due to selection bias. Propensity score matching has been increasingly utilized in such studies as a means of reducing selection bias so as to approximate the characteristics of experimental studies. This course will introduce methods of propensity score matching and discuss when and how to apply propensity score matching in non-experimental studies. Through lectures on propensity score matching methods and hands-on activities for the use of R packages on real data, this course will benefit graduate students improving the quality of non-experimental studies for comparing treatment effects. Instructions for installing R packages and downloading example data will be provided to participants in advance through a course website, and handouts will be also available in the class. No prior knowledge of propensity score matching is required. However, an understanding of basic research design and statistics is desirable. Participants are encouraged to bring their own laptops for hands-on activities during which participants are also welcome to work on their own data.

Sub Unit

• Division D - Measurement and Research Methodologies

Chairs

• Brian F. French, Washington State University

• Hong Jiao, University of Maryland-College Park

Presenter

• Wei Pan, Duke University