This paper demonstrates the application, analysis, and interpretation of a Hybrid Regression Discontinuity (HRD) design. HRD is a combination of a Regression Discontinuity (RD) design and a Randomized Control Trial (RCT). The HRD was implemented in conjunction with a summer literacy program administered by a school district in the U.S. Pacific Northwest. Students scoring below a lower cutoff of 30 words correctly read per minute (WCPM) in the spring of first grade comprised the RD treatment group, students scoring above an upper cutoff of 50 WCPM comprised the RD control group, and those within the interval between cutoffs were randomly assigned to the RCT treatment or control condition. Across multiple parametric and nonparametric analyses, treatment effects were observed at the upper cutscore and within the randomization interval. Implications for evaluating functional form, statistical power, and generalizing the RD treatment effect beyond the immediate region surrounding the assignment cutscore are discussed.
Zvoch et al. (Tue,) studied this question.