arsemode

·5 anni fa·discuss

The correct way to do this is to adjust for baseline covariates, something that has been done in medicine and other fields for a long time. For A/B-tests specifically, the procedure goes by the name CUPED: https://exp-platform.com/cuped/