Balance-sample size frontier

Causal inference with observational data and Matching (statistics).

On one hand, if we use Propensity score matching, it throws away too much information. If we use Exact matching (or Coarsened exact matching), the Curse of dimensionality kicks in and one should throw away too many unmatched samples. This trade-off defines the frontier in the balance vs. sample size space.

See King2016balance.