Fig. 1
From: Graphing and reporting heterogeneous treatment effects through reference classes
Two local kernel weighting schemes for ITE estimation and one global reweighting scheme for CATE estimation, as compared to ATE estimation. The ITE and CATE reweighting schemes represented all reduce the effective sample size of the original data and target an average risk corresponding to the 25th risk quantile. The scale of the y-axis is chosen so that the sum of the weights is equal to the effective sample size, or equivalently that the effective sample size is equal to the area under the curve. The effective sample sizes as a percentage of the original data are shown in the legend