Estimating causal effects with optimization-based methods: A review and empirical comparison.

Area of Science:

• Causal Inference
• Operations Research
• Statistical Modeling

Background:

• Estimating causal effects requires balancing covariates between treated and control groups, especially without randomized experiments.
• Existing methods for covariate balance vary, with recent focus on optimization-based approaches.
• A comprehensive comparison of these optimization methods and their potential is lacking.

Purpose of the Study:

• To provide an overview of causal inference literature.
• To detail optimization-based causal inference methods.
• To compare prevailing optimization-based methods and identify future research opportunities.

Main Methods:

• Literature review of causal inference.
• Detailed description of optimization-based causal inference techniques.
• Comparative analysis of existing optimization-based methods.

Main Results:

• Optimization-based methods show empirical improvements in covariate balance and causal effect estimation.
• A thorough comparison among optimization-based methods is needed.
• Opportunities exist for operational research contributions to causal inference.

Conclusions:

• Optimization-based methods offer a promising avenue for improving causal inference.
• Further research is needed to compare these methods and explore new optimization-driven approaches.
• Collaboration between operational researchers and applied researchers can advance causal inference tools.