Propensity Score Matching

Meeting Schedule

May 2008

"Using Propensity Scores and Full Matching to Examine the Relationship between Adolescent Drug Use and Adult Outcomes"
Dr. Elizabeth Stuart, Departments of Mental Health and Biostatistics, Johns Hopkins Bloomberg School of Public Health
Friday May 9, 2008
2:00pm - 3:30pm
Journalism 243

April 2008

"Cohabitation, Divorce, and the Trial Marriage Hypothesis"
Dr. Felix Elwert, Department of Sociology, University of Wisconsin
Friday April 25, 2008
12:00pm - 1:30pm
Journalism 243

February 2008

"Ballot Initiatives and State Outcomes"
Dr. Luke Keele, Department of Political Science, OSU
Friday February 29, 2008
12:00pm - 1:30pm
Journalism 243

January 2008

"Differences in Governance Practices between U.S. and Foreign Firms: Measurement, Causes, and Consequences"
Dr. Isil Eril, Fisher College of Business, OSU
Friday January 25, 2008
10:30am - 12:00pm
Journalism 243

November 2007

"Tackling Causality in Family Structure Research"
Dr. Claire Kamp Dush, Department of Human Development and Family Science, OSU
Friday November 16, 2007
10:30am - 12:00pm
Journalism 243

April 2007

"Propensity Score Matching, a Distance-Based Measure of Migration, and the Wages of Young Men"
Dr. Xianghong Li, Department of Economics, York University
Friday April 5, 2007
10:30am - 12:00pm
Journalism 243

March 2007

Dr. Bo Lu, Department of Biostatistics, The Ohio State University
Friday March 2, 2007
10:30am - 12:00pm

Description

Selection biases associated with the analysis of observational data are well known. With biases, it is always difficult to establish causal relations. Observational data, even longitudinal data, often do not allow us to make causal statements easily. To address such causality difficulties, we explore substantive interests using the method of propensity score matching. Initially, we seek to address research questions that explore the effects of out-of-wedlock childbearing on union formation patterns, the effects of intermarriage on marital stability, the effects of transitions in marital status on health, and the effects of migration on earnings. Given available data, causal relationships within these questions cannot be readily established without accounting for selection biases. Propensity score matching allows researchers to designate certain characteristics, such as race, income, region, and/or educational attainment, to act as matching criteria. These matching criteria serve as independent variables or controls that are used to generate the propensity score. Based on the criteria, a counterfactual group can be created. A matching variable divides respondents into control and treatment groups while the background control variables generate the propensity score that is used to match respondents into pairs. Propensity score matching, indeed, provides an important alternative for addressing selection bias. Future research needs to explore potential limitations such as sample size, the extent of overlap between treatment and control groups, and hidden bias.

Faculty Advisors

Dr. Zhenchao Qian, Department of Sociology

Dr. Bo Lu, Department of Biostatistics

Dr. Kristi Williams, Department of Sociology

Dr. Patricia Reagan, Department of Economics

Contact Information

Matthew Painter, Ph.D. Student, Department of Sociology

E-Mail: painter.63@sociology.osu.edu