Regression-Discontinuity Designs in Medical Settings: An Alternative to Quasi-Experimental Designs inTesting Treatment Effectiveness

Abstract

The regression-discontinuity design (RD) is a powerful methodological alternative to an experimental design when one is not feasible. RD is frequently a better alternative than quasi-experimental design in minimizing threats to internal validity. The traditional approach in RD is to test for slope or intercept differences between the control and the treatment groups. However, an alternative approach in RD is presented that tests for slope differences, intercept differences, and permits examination of change in individuals. This alternative approach illustrates how RD can be used to enhance evidence-based practices and interventions in a clinical setting when analyzing a single group of subjects.