Regression discontinuity design is the quasi-experimental design in which participants in, for instance, an intervention study, are assigned to the treatment and control groups on the basis of a cutoff value on the pre-intervention measure which affects the outcome, rather than by the randomization. If the treatment has an effect a discontinuity in the outcomes would be predictable at the cutoff. A weakness is that the extrapolation of counterfactual outcomes for treated in absence of treatment is needed, based on the regression model estimated for the non-treated. See the Figure for an illustration where treatment is provided to those below a cutoff on a pretest.