A big thanks to Prof. Rolf Groenwold from Utrecht for his talk on unmeasured confounding. I especially enjoyed the in-depth discussion of pros and cons of popular methods to ‘handle’ this problem.
Methods critically discussed included:
- Two-stage sampling
- Cross-over designs
- Negative controls
- Instrumental variables
- Sensitivity analysis
Rolf’s reflection on dealing with multiple sources of bias simultaneously and quantifying the impact of measurement error also got everyone thinking.
Perhaps the highlight was the following quote Rolf shared with us from Mr. Donald Rumsfeld, clearly a scholar in articulating types of unmeasured confounding 😉
Photo credit: azquotes.com
We hope you will join us again Oct. 4th for our next meeting on Mendelian Randomization. Sign up is already available on our ‘Calendar’ page.
Hope to see you then! -JR