方法对比

	Sibling Fixed-Effects Design	E-Value Sensitivity Analysis
领域	Social Epidemiology	Social Epidemiology
方法族	Regression model	Process / pipeline
起源年份	2013	2017
提出者	Brian D'Onofrio, Benjamin Lahey, Eric Turkheimer & Paul Lichtenstein; Thomas Frisell et al.	Tyler J. VanderWeele & Peng Ding
类型	Within-family fixed-effects design for confounding control	Assumption-free sensitivity analysis for unmeasured confounding
开创性文献	Frisell, T., Oberg, S., Kuja-Halkola, R., & Sjolander, A. (2012). Sibling Comparison Designs: Bias From Non-Shared Confounders and Measurement Error. Epidemiology, 23(5), 713-720.	VanderWeele, T. J., & Ding, P. (2017). Sensitivity analysis in observational research: introducing the E-value. Annals of Internal Medicine, 167(4), 268-274.
别名	Sibling Comparison Design, Within-Family Fixed Effects, Discordant Sibling Design, Discordant Twin Design	E-Value, E-Value for Unmeasured Confounding, VanderWeele-Ding E-Value, Bias Factor Sensitivity Analysis
相关	4	3
摘要	The sibling fixed-effects, or sibling-comparison, design controls for everything that siblings share by construction. Genes (on average half), parents, household income, neighborhood, schooling, and family culture are differenced out when you compare brothers or sisters who differ in an exposure, so the residual within-family association is purged of all confounders common to the family. D'Onofrio, Lahey, Turkheimer, and Lichtenstein championed these family-based quasi-experiments as a way to integrate genetic and social-science research by rigorously testing competing causal hypotheses. Frisell and colleagues, however, gave the design its essential warning label: precisely because shared confounding is removed, the within-family estimate is unusually vulnerable to the confounders siblings do not share and to attenuation from measurement error. The design is powerful but double-edged.	The E-value, introduced by Tyler VanderWeele and Peng Ding in 2017, is a simple, assumption-free way to quantify how robust an observational association is to unmeasured confounding. It answers a single, sharply posed question: how strong would an unmeasured confounder have to be — in its association with both the exposure and the outcome — to fully explain away the observed effect? The larger the E-value, the more powerful a hidden confounder would need to be, and so the more robust the finding. The method rests on the bounding factor derived by Ding and VanderWeele in their 2016 'Sensitivity analysis without assumptions,' which holds regardless of the distribution or number of unmeasured confounders. Because it requires only the point estimate and confidence limit on the risk-ratio scale and no untestable bias parameters, the E-value has become a routine reporting standard in observational epidemiology, including social epidemiology where unmeasured confounding is pervasive.