The Case Against Manually Weighting Covariates When Matching for Causal Inference

• In Event: New Insights in Causal Inference: Interference, Heterogeneity, and Balancing

Brief Overview

We present arguments against the manual weighting of covariates, for matching methods of causal inference (King, Lucas, and Nielsen 2017), and explain why weighting via machine-learning optimization is preferable.

Authors

• Idriss Bennis, Keck Graduate Institute

• Alexis J Diamond, Minerva Schools at KGI

• Mikulas Plesak, Minerva Schools at KGI