English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

It is a truth universally acknowledged that an observed association withoutknown mechanism must be in want of a causal estimate. However, causalestimation from observational data often relies on the (untestable) assumptionof `no unobserved confounding'. Violations of this assumption can induce biasin effect estimates. In principle, such bias could invalidate or reverse theconclusions of a study. However, in some cases, we might hope that theinfluence of unobserved confounders is weak relative to a `large' estimatedeffect, so the qualitative conclusions are robust to bias from unobservedconfounding. The purpose of this paper is to develop \emph{Austen plots}, asensitivity analysis tool to aid such judgments by making it easier to reasonabout potential bias induced by unobserved confounding. We formalizeconfounding strength in terms of how strongly the confounder influencestreatment assignment and outcome. For a target level of bias, an Austen plotshows the minimum values of treatment and outcome influence required to inducethat level of bias. Domain experts can then make subjective judgments aboutwhether such strong confounders are plausible. To aid this judgment, the Austenplot additionally displays the estimated influence strength of (groups of) theobserved covariates. Austen plots generalize the classic sensitivity analysisapproach of Imbens [Imb03]. Critically, Austen plots allow any approach formodeling the observed data and producing the initial estimate. We illustratethe tool by assessing biases for several real causal inference problems, usinga variety of machine learning approaches for the initial data analysis. Code isavailable at https://github.com/anishazaveri/austen_plots

Sense and Sensitivity Analysis: Simple Post-Hoc Analysis of Bias Due to Unobserved Confounding