Statistical Power Calculator
Tells you how likely a study of a given size is to detect an effect, or how large it needs to be.
Power
— %
Share sends a link that reopens these numbers. Copy details pastes the full breakdown as a list.
Needed per group for the target —
Needed in total —
Reading this —
The formula
power = P(|T| > t crit), T noncentral t df = 2n − 2, noncentrality = d × √(n ÷ 2)
This assumes two independent, equal-sized groups with normal data and equal standard deviations. The noncentral t distribution is integrated numerically rather than approximated by the normal.
What to watch out for
A power of 80% is a common target, not a rule. Other designs, such as paired tests or unequal groups, need their own calculation.
Frequently Asked Questions
How many per group to detect d = 0.5 with 80% power?
64 per group at a 5% two-sided significance level.
And a small effect of d = 0.2?
394 per group. Required size grows with the inverse square of the effect.