"you can never completely get rid of confounding in observational studies -- it's just not possible"

https://youtu.be/gAK58BIUgdk

Comments