Simple graphical rules for assessing selection bias in general-population and selected-sample treatment effects
Open Access
- 20 June 2024
- journal article
- research article
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 194 (1) , 267-277
Abstract
When analyzing a selected sample from a general population, selection bias can arise relative to the causal average treatment effect (ATE) for the general population, and also relative to the ATE for the selected sample itself. In this paper, we provide simple graphical rules that indicate (1) whether a selected-sample analysis will be unbiased for each ATE and (2) whether adjusting for certain covariates could eliminate selection bias. The rules can easily be checked in a standard single-world intervention graph. When the treatment could affect selection, a third estimand of potential scientific interest is the “net treatment difference”—namely the net change in outcomes that would occur for the selected sample if all members of the general population were treated versus not treated, including any effects of the treatment on which individuals are in the selected sample. We provide graphical rules for this estimand as well. We decompose bias in a selected-sample analysis relative to the general-population ATE into (1) “internal bias” relative to the net treatment difference and (2) “net-external bias,” a discrepancy between the net treatment difference and the general-population ATE. Each bias can be assessed unambiguously via a distinct graphical rule, providing new conceptual insight into the mechanisms by which certain causal structures produce selection bias.Keywords
All Related Versions
Funding Information
- National Institutes of Health (P30DK116074, P30CA124435, UL1TR003142, R01 LM013866)
This publication has 31 references indexed in Scilit:
- The M-Value: A Simple Sensitivity Analysis for Bias Due to Missing Data in Treatment Effect EstimatesAmerican Journal of Epidemiology, 2022
- Evaluation of the dose-effect association between the number of doses and duration since the last dose of COVID-19 vaccine, and its efficacy in preventing the disease and reducing disease severity: A single centre, cross-sectional analytical study from IndiaDiabetes & Metabolic Syndrome: Clinical Research & Reviews, 2021
- Graphical Models for Processing Missing DataJournal of the American Statistical Association, 2021
- Selection Mechanisms and Their Consequences: Understanding and Addressing Selection BiasCurrent Epidemiology Reports, 2020
- Target Validity and the Hierarchy of Study DesignsAmerican Journal of Epidemiology, 2018
- Sickness Absence and Psychosocial Job Quality: An Analysis From a Longitudinal Survey of Working Australians, 2005–2012American Journal of Epidemiology, 2015
- On the Causal Interpretation of Race in Regressions Adjusting for Confounding and Mediating VariablesEpidemiology, 2014
- Identification and estimation of survivor average causal effectsStatistics in Medicine, 2014
- The obesity paradox: Understanding the effect of obesity on mortality among individuals with cardiovascular diseasePreventive Medicine, 2014
- A Structural Approach to Selection BiasEpidemiology, 2004