CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: The E-value: Methodological clarifications Authors:  Arvid Sjolander - Karolinska Institute (Sweden) [presenting]
Abstract: Most epidemiological studies aim to estimate the causal effect of a particular exposure on a particular outcome. Recently, the so-called ``E-value'' was proposed, as a tool to assess the sensitivity of obtained estimates to unmeasured confounding. The E-value has quickly become very popular, and is today often recommended. We will clarify some important methodological points regarding the sensitivity analysis, on which the E-value is based. Specifically, we will derive the feasible range for the underlying sensitivity analysis parameters, and we will discuss when and why the sensitivity analysis can be expected to be conservative.