An optimization approach for making causal inferences

Authors

  • Wendy K. Tam Cho,

    Corresponding author
    • University of Illinois at Urbana-Champaign, Urbana, IL, USA
    Search for more papers by this author
    • wendycho@illinois.edu
    • Professor in the Departments of Political Science and Statistics and Senior Research Scientist at the National Center for Supercomputing Applications.
  • Jason J. Sauppe,

    1. University of Illinois at Urbana-Champaign, Urbana, IL, USA
    Search for more papers by this author
    • Graduate student in the Department of Computer Science.
  • Alexander G. Nikolaev,

    1. University at Buffalo (SUNY), Buffalo, NY, USA
    Search for more papers by this author
    • Assistant Professor in the Department of Industrial and Systems Engineering.
  • Sheldon H. Jacobson,

    1. University of Illinois at Urbana-Champaign, Urbana, IL, USA
    Search for more papers by this author
    • Professor in the Department of Computer Science.
  • Edward C. Sewell

    1. Southern Illinois University at Edwardsville, Edwardsville, IL, USA
    Search for more papers by this author
    • Professor in the Department of Mathematics and Statistics.

Abstract

To make causal inferences from observational data, researchers have often turned to matching methods. These methods are variably successful. We address issues with matching methods by redefining the matching problem as a subset selection problem. Given a set of covariates, we seek to find two subsets, a control group and a treatment group, so that we obtain optimal balance, or, in other words, the minimum discrepancy between the distributions of these covariates in the control and treatment groups. Our formulation captures the key elements of the Rubin causal model and translates nicely into a discrete optimization framework.

Ancillary