Propensity-score matching in economic analyses: comparison with regression models, instrumental variables, residual...

Area of Science:

• Health Economics
• Econometrics
• Biostatistics

Background:

• Observational data analysis is crucial in health economics.
• Propensity score matching is a widely used method.
• Existing literature covers estimation but less on conceptual comparisons.

Purpose of the Study:

• To compare propensity score models with alternative treatment effect estimators.
• To provide intuition on the relative merits of different methods.
• To inform methodology choice in health economic evaluations.

Main Methods:

• Conceptual comparison of propensity score models and alternative estimators.
• Review of existing empirical comparisons.
• Consideration of factors influencing method choice (nonlinearity, sample size, data availability, heterogeneity, missing variables).

Main Results:

• Highlights the importance of conceptual understanding beyond empirical comparisons.
• Identifies key factors impacting the choice of estimators in health economic evaluations.
• Discusses potential combinations of propensity score matching with other methods like differences-in-differences.

Conclusions:

• The choice of estimator in health economic evaluations depends on specific study characteristics.
• Propensity score matching offers a valuable framework but requires careful consideration of alternatives.
• Further research is needed for comprehensive comparisons across all methods.