A recent study, co-authored by SCAN-Unit Associate Senior Fellow Dr. E. Pronizius, explores whether an observational study can support a cause-and-effect claim despite unmeasured confounders. It examines how strongly these confounders relate to both the factor and the outcome. If their combined influence is below a specific threshold, the causal claim becomes more credible. Dr. Erin Buchanan developed the ViSe tool (an R package and Shiny app) to apply this method, offering visual aids and simple rules to analyze one’s own data or published results.

See for more: 

 

• Buchanan, E. M. (2024). Visualizing Sensitivity. R package version 0.1.3. doi: 10.32614/CRAN.package.ViSe 

 

• Höfler, M., Pronizius, E., & Buchanan, E.M. How Large Must an Associational Mean Difference Be to Support a Causal Effect? Methodology, 20(4), 318-335. https://doi.org/10.5964/meth.14579 

 

Picture copyright: https://pixabay.com/illustrations/statistics-diagram-graphic-bar-7206876/