2 ms·
In physics, we use 2-sigma (95%) limits all the time. 5-sigma (99.9999%) is generally required only for a claim of detection. If you're just feeling your way a
by yes_procrast 13y ago
In physics, we use 2-sigma (95%) limits all the time. 5-sigma (99.9999%) is generally required only for a claim of detection.
If you're just feeling your way around in the dark, 2-sigma is a useful way to work, so we use that to guide exploration.
Why 2-sigma? Well, it's twice as big as 1-sigma.
Experiment didn't go well, but you need a more-impressive result? Use a 90% confidence interval instead of 95%.
- minimaxir 13y ago1-sigma is 68% confidence. 90% confidence intervals assume 1.64 sigma.
- yes_procrast 13y agoAgreed, I'll add a line break to denote that they're separate thoughts. 1-sigma is indeed 68%.