Robust Inference in Fuzzy Regression Discontinuity with Multiple Forcing Variables
Rong Ma, Ang Sun and Zhaoguo Zhan
Coles Working Paper Series, FALL16-08, November 2016
Overview Regression Discontinuity (RD) identifies causal effects by exploiting the discontinuity in treatment assignment. Since the late 1990s, it has become a standard tool of applied researchers. We propose a novel test on the causal effect in RD designs since the commonly used t-test potentially suffers from size distortion. Our test is particularly appealing when RD designs are associated with multiple forcing variables for treatment assignment and/or non-substantial magnitude of discontinuity in the probability of receiving treatment. For illustration, we use the proposed test to study whether the awareness of hypertension decreases fat intake. Based on the outcome of the test, we cannot reject that fat intake remains unaffected by hypertension awareness, nor can we reject that fat intake decreases by the amount reported in the existing literature.
28 | Coles Working Paper
Made with FlippingBook - Online catalogs