The Futility of Propensity Score Methods in a Statewide Study of the International Baccalaureate

• In Event: Methodological and Conceptual Issues in Longitudinal Analyses

Abstract

This paper uses 15 years of longitudinal student and school-level data from 1995 through 2009 from the Florida K-20 Education Data Warehouse to compare over 10,000 International Baccalaureate (IB) students to non-IB students. We use propensity score methods to adjust for preexisting differences between IB and non-IB students beginning in 3rd grade. Results revealed that the selection bias associated with participation in IB is very large, and that mechanisms for dealing with selection bias using propensity scores may not be sufficient. In other words, comparing IB and non-IB students in this statewide context is like comparing apples and oranges, and using propensity score methods to adjust for these differences require strong assumptions and extrapolation into regions with very thin data.

Authors

• Henry May, University of Delaware

• Awilda Rodriguez, University of Michigan - Ann Arbor

• Philip M. Sirinides, University of Pennsylvania

• Laura W. Perna, University of Pennsylvania

• April Yee, University of Pennsylvania

• Tafaya Ransom, University of Pennsylvania