Get the latest tech news
P-Hacking in Startups
When agile experimentation at startups becomes a p-hacking trap
Statistically, the probability of at least one false positive increases with each additional test, converging toward 1 as the number of comparisons grows: By stating in advance which metric will count as evidence, you fix the false positive rate at its intended level. Even if you’ve accounted for multiple variants and didn't the temptation to shift metrics, one bias remains: impatience.
Or read this on Hacker News
