The Effect of Propensity Score Matching Method on the Quantity and Quality of Matches

Often program administrators are interested in knowing how students benefit from participation in programs compared to students who do not participate. Such comparisons may be sullied by the fact that participants self-select into programs, resulting in differences between groups prior to programmin...

Full description

Saved in:
Bibliographic Details
Published inThe Journal of experimental education Vol. 85; no. 4; pp. 535 - 558
Main Authors Jacovidis, Jessica N., Foelber, Kelly J., Horst, S. Jeanne
Format Journal Article
LanguageEnglish
Published Washington Routledge 02.10.2017
Taylor & Francis Group, LLC
Taylor & Francis Inc
Subjects
Assessment
College Freshmen
Comparative Analysis
Effect Size
Ethnic Diversity
Evaluation Methods
Gender Differences
higher education
Information Literacy
Likert Scales
Matched Groups
matching
MEASUREMENT, STATISTICS, AND RESEARCH DESIGN
Measures (Individuals)
Probability
propensity score matching
quasi-experimental
Quasi-experimental methods
Quasiexperimental Design
Researchers
Scores
Statistical Analysis
Student Evaluation
Student Participation
Online AccessGet full text

Cover

More Information
Summary:Often program administrators are interested in knowing how students benefit from participation in programs compared to students who do not participate. Such comparisons may be sullied by the fact that participants self-select into programs, resulting in differences between groups prior to programming. By controlling for researcher-identified-self-selection variables, propensity score matching enables researchers to create comparable matched groups. However, when employing propensity score matching, researchers are faced with a plethora of matching options. In the current study, we compared the quantity and quality of matches obtained when applying common matching techniques to real data. The methods produced matches of varying quantity and quality. Moreover, group comparisons on the outcome led to different conclusions depending on the matching method employed.
ISSN:0022-0973
1940-0683
DOI:10.1080/00220973.2016.1250209