Regression Discontinuity Designs
young caine: Master, may we speak further on the forces of destiny? master PO: Speak.
caine: As we stand with two roads before us, how shall we know whether the left road or the right road will lead us to our destiny?
master PO: You spoke of chance, Grasshopper. As if such a thing were certain to exist. In the matter you speak of, destiny, there is no such thing as chance.
Kung Fu, Season 3, Episode 62
Human behavior is constrained by rules. The State of California limits elementary
school class size to 32 students; 33 is one too many. The Social Security Administration won’t pay you a penny in retirement benefits until you’ve reached age 62. Potential armed forces recruits with test scores in the lower deciles are ineligible for American military service. Although many of these rules seem arbitrary, with little grounding in science or experience, we say: bring ’em on! For rules that constrain the role of chance in human affairs often generate interesting experiments. Masters of ’metrics exploit these experiments with a tool called the regression discontinuity (RD) design. RD doesn’t work for all causal questions, but it works for many. And when it does, the results have almost the same causal force as those from a randomized trial.