Fig. 1
From: Preventing false discovery of heterogeneous treatment effect subgroups in randomized trials
Plot of maximum absolute standardized difference (ASD) within node for each method (x-axis) versus absolute bias (absolute value of estimated treatment effect minus true treatment effect) in each identified node for LR, RF, gradient RF, and mCART in simulation 1A. All identified subgroups to the left of the vertical dashed line of 0.2 have an acceptable balance. ASD absolute standardized difference, LR logistic regression, mCART matching plus classification and regression trees, RF random forest