-
link to article
- by William Briggs
summary
The probability leakage of model M with respect to evidence E is defined. Probability leakage is a kind of model error. It occurs when M implies that events y, which are impossible given E, have positive probability. Leakage does not imply model falsification. Models with probability leakage cannot be calibrated empirically. Regression models, which are ubiquitous in statistical practice, often evince probability leakage.
Let's start with the truth!

Support the Broken Science Initiative.
Subscribe today →
recent posts
Principles for Feeling Great Now and Learning How To Live To Be 100
I’m one of the millions of people trying to keep the weight off without taking expensive drugs like Mounjaro or Ozempic for life. It wasn’t easy, but I found ways.
Are pharmaceuticals more expensive and less effective at treating hypertension than diet and exercise?