Abstract: Collision bias is the most treacherous error in statistics: it can be subtle, it is easy to induce it by accident, and the error it causes can be bigger than the effect you are trying to measure. It is the cause of Berkson’s paradox, the low birthweight paradox, and the obesity paradox, among other famous historical errors. And it might be the cause of your next blunder! Although it is best known in epidemiology, it appears in other fields of science, engineering, and business.
In this talk, I will present examples of collision bias and show how it can be caused by a biased sampling process or induced by inappropriate statistical controls; and I will introduce causal diagrams as a tool for representing causal hypotheses and diagnosing collision bias.
So, don’t tell anyone, but this talk is part of my stealth book tour!
- It started in 2019, when I presented a talk at PyData NYC based on Chapter 2: Relay Races and Revolving Doors.
- In 2022, I presented another talk at PyData NYC, based on Chapter 12: Chasing the Overton Window.
- Today’s talk is based on Chapter 7: Causation, Collision, and Confusion.
- In July I’m presenting a talk at SciPy based on Chapter 9: The Long Tail of Disaster.
And, if things go according to plan, I’ll present Chapter 1 at a book event at the Needham Public Library on December 7.
More chapters coming soon!