[Overview on the generalized propensity scoring estimator for continuous treatment].

Area of Science:

• Epidemiology
• Biostatistics
• Observational Studies

Background:

• Propensity score (PS) methods are widely used for causal inference in observational studies to control for measured confounding.
• Traditional PS methods are primarily designed for dichotomous treatments.
• Recent advancements have introduced generalized propensity scoring for continuous treatments.

Purpose of the Study:

• To introduce and review existing estimators for propensity score methods in causal inference.
• To discuss generalized propensity scoring estimators for continuous treatments.
• To present estimators from both model-based and balance-based perspectives.

Main Methods:

• Review of existing propensity score estimation methods.
• Discussion of generalized propensity scoring for continuous treatments.
• Categorization of estimators into model-based and balance-based approaches.

Main Results:

• The paper provides an overview of various propensity score estimation techniques.
• It highlights the development of generalized propensity scoring for continuous treatment variables.
• Existing estimators are presented through distinct theoretical frameworks.

Conclusions:

• The study consolidates knowledge on propensity score methods for causal inference.
• It emphasizes the utility of generalized propensity scoring for complex treatment scenarios.
• Understanding different estimation perspectives is crucial for appropriate method selection.