causal

Low power produces both false negatives and inflated false positives

Operating in an underpowered environment creates two problems simultaneously: a very high chance of false negatives (missing real signals), and a high risk that any statistically significant result is a false positive or a grossly inflated effect size—the smaller the power, the greater the average inflation of detected effects.

causalpending

Speaker

John Ioannidis

Evidence Quote

“the smaller the power, the greater the average inflation of the effect that you would see, when you do detect it. So, two major problems. With low power: lots of false negatives. Second problem: lots of false positives and gross exaggeration of the effect sizes.”

Source

John Ioannidis on Statistical Significance, Economics, and Replication— EconTalk
Created: 6/13/2026, 12:20:10 AM

Your notes

Loading notes...