TWANG, the Toolkit for Weighting and Analysis of Nonequivalent Groups, developed by the statistics group of the RAND Corporation, contains a set of functions to support Rubin causal modeling of observational data through the estimation and evaluation of propensity score weights by applying gradient boosting.