| Session Title: Using Regression Discontinuity Designs for Program Evaluation |
| Demonstration Session 626 to be held in Lone Star E on Friday, Nov 12, 3:35 PM to 4:20 PM |
| Sponsored by the Quantitative Methods: Theory and Design TIG |
| Presenter(s): |
| Joseph Stevens, University of Oregon, stevensj@uoregon.edu |
| Keith Zvoch, University of Oregon, kzvoch@uoregon.edu |
| Drew Braun, Bethel School District, dbraun@bethel.k12.or.us |
| Abstract: The proposed demonstration describes the use, analysis, and interpretation of Regression Discontinuity (RD) designs in applied program evaluations. RD designs are often recommended in quasi-experimental situations (Shadish, Cook, & Campbell, 2002) as an alternative to randomized experiments. When properly implemented, RD designs provide causal inferences that are as strong as those obtained from randomized experiments. In practice, however, there are a number of challenges in the implementation, analysis, and interpretation of RD designs. This presentation will define the features of RD designs, demonstrate variations of RD designs, and show how data are analyzed and interpreted. Analyses and examples will be based on literacy data from 1,449 elementary students from a Northwest school district. We will also discuss several challenges in the use of RD designs (e.g., compliance, attrition) as well as important issues in analysis of RD results (e.g., functional form of the pre-post relationship, interactions). |