[Paper Review] Choosing the Causal Estimand for Propensity Score Analysis of Observational Studies

Noah Greifer, Elizabeth A. Stuart|arXiv (Cornell University)|Jun 19, 2021

Advanced Causal Inference Techniques31 citations

TL;DR

The paper guides researchers on selecting and interpreting the causal estimand (ATT, ATU, ATE, ATO) for propensity score methods in observational studies and discusses implications for regression and IV analyses.

ABSTRACT

Matching and weighting methods for observational studies involve the choice of an estimand, the causal effect with reference to a specific target population. Commonly used estimands include the average treatment effect in the treated (ATT), the average treatment effect in the untreated (ATU), the average treatment effect in the population (ATE), and the average treatment effect in the overlap (i.e., equipoise population; ATO). Each estimand has its own assumptions, interpretation, and statistical methods that can be used to estimate it. This article provides guidance on selecting and interpreting an estimand to help medical researchers correctly implement statistical methods used to estimate causal effects in observational studies and to help audiences correctly interpret the results and limitations of these studies. The interpretations of the estimands resulting from regression and instrumental variable analyses are also discussed. Choosing an estimand carefully is essential for making valid inferences from the analysis of observational data and ensuring results are replicable and useful for practitioners.

Motivation & Objective

• Explain why choosing a causal estimand matters for propensity score analysis in observational studies.
• Describe common estimands (ATT, ATU, ATE, ATO) and their target populations.
• Discuss how estimand choice affects assumptions, interpretation, and statistical methods.
• Relate estimand interpretations to regression and instrumental variable analyses.
• Provide guidance to ensure valid inferences and replicability for practitioners.

Proposed method

• Review and synthesize the definitions and interpretations of key estimands in propensity score analysis.
• Outline the assumptions and conditions under which each estimand is identifiable.
• Discuss the implications of estimand choice for statistical methods used to estimate causal effects.
• Offer practical guidance for medical researchers on implementing methods and interpreting results.

Experimental results

Research questions

• RQ1 Which causal estimand is most appropriate given the target population and study design in propensity score analyses of observational data?
• RQ2 What are the assumptions and limitations associated with ATT, ATU, ATE, and ATO estimands?
• RQ3 How do estimand choices influence the interpretation of results and comparability with regression or instrumental variable analyses?
• RQ4 What guidance can improve validity, replicability, and utility of findings for practitioners?

Key findings

• Estimand choice is essential for valid inferences and correct interpretation of propensity score analyses.
• Each estimand (ATT, ATU, ATE, ATO) corresponds to a specific target population and has distinct assumptions and methods.
• The article discusses how interpretations align or differ from regression and instrumental variable analyses.
• Guidance is provided to help researchers correctly implement methods and to aid audiences in understanding study limitations.