USPS: Unsupervised and Supervised methods of Propensity Score Adjustment for Bias

Unsupervised methods: Define Local Treatment Differences (LTDs) and Local Average Outcomes (LAOs) within Clusters of well-matched patients and display their distributions across Clusters. This form of Nonparametric Preprocessing of observational data is also called Local Control because it uses post-hoc blocking to implement nested ANOVA (treatment within cluster.) Supervised methods: Estimate and Validate Propensity Scores and use them to either sub-group or smooth observed patient outcomes over the common support of alternative treatment cohorts.

Downloads: