What statistical power does my study have?
Get the free Statistics toolkit
Power, effect-size, and test-selection templates & worked examples from the Statistical Test Selection Workbook. We’ll email you the download link.
How statistical power works
Power is the probability of detecting a real effect of the size you specify. Four quantities are locked together — sample size, effect size, significance level (alpha), and power — fix any three and the fourth follows. This tool fixes sample size, effect, and alpha, and returns the power.
- Two means: power rises with Cohen’s d and the number per group.
- One-sample / paired mean: a within-subject design needs fewer participants for the same power.
- Two proportions: driven by the gap between the rates and the per-group size.
- Correlation: driven by the size of r and the total sample, via Fisher’s z.
- Aim for 80% or more. Below that, a non-significant result is ambiguous.
A two-sided normal approximation suitable for planning. For complex designs (clustering, repeated measures, survival, non-inferiority) confirm with a formal power analysis or a statistician.
Frequently asked questions
What is statistical power?
The probability your study detects a real effect of the size you specify. Power of 0.80 = an 80% chance of finding it (and 20% chance of missing it, a Type II error).
What power should I aim for?
80% is the conventional minimum; 90% is stronger. Below ~80%, a non-significant result can’t be distinguished from an underpowered one.
Should I compute power after my study (post-hoc)?
Not from the observed effect — “observed power” is circular. Use a meaningful target effect set from prior work, which is what this tool does.
How do I increase power?
Recruit more participants, target a larger (still realistic) effect, use a within-subject/paired design, reduce measurement error, or accept a higher alpha.
Does it store anything?
No. The calculation runs entirely in your browser; nothing is uploaded or saved.