Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

To mention it, not necessarily directed at you. Larger test groups are required when trying to distinguish between less effective treatments.

If rabies has a 99% death rate and you try your rabies treatment on 10 people, 9 of whom live, that's much stronger evidence than a situation where 50% of people died without treatment versus 48% with treatment (sample size 100). To be confident it's not just random chance, you'll need a really large sample size for the 2nd situation.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: