Fig. 3: DAG and plots of RDDs. | Nature Computational Science

Fig. 3: DAG and plots of RDDs.

From: Quantifying causality in data science with quasi-experiments

Fig. 3

a, Graphical depiction of how RDDs use a threshold T in a continuous running variable R to effectively randomize treatment X (crossed red line) to estimate causal effects Y. b, Scatterplot (blue dots) and fitted local regressions (red lines) within the bandwidth (shaded red regions) of outcome Y on running variable R for simulated linear data to the left and right of the cutoff (dashed line). c, Scatterplot (blue dots) and fitted local regressions (red lines) within the bandwidth (shaded red regions) of outcome Y on running variable R for simulated nonlinear data to the left and right of the cutoff (dashed line).

Back to article page