DigitHelm
Statistics & Probability

Power Analysis Calculator | Sample Size, Effect Size & Statistical Power

Determine the required sample size for a study given desired statistical power (1 − β), significance level α, and effect size. Supports Cohen's d for t-tests and z-tests, Cohen's h for proportion tests, and Cohen's w for chi-square tests. Also computes achieved power for a given sample size.

Instant Results100% FreeAny DeviceNo Sign-up

TEST TYPE

DIRECTION

MODE

PRESETS

Required n (per group)

63

Power = 80%

Critical value z_α

1.960

two-sided

Type I error α

5.0%

False positive rate

Type II error β

20.0%

False negative rate

POWER CURVE (n vs POWER)

nPowerBar
1020.0%
2035.2%
3049.1%
4060.9%
5070.5%
6078.2%
8088.5%
10094.2%
12097.2%
15099.1%
20099.9%

What Is the Power Analysis Calculator | Sample Size, Effect Size & Statistical Power?

Determine the required sample size given a desired power (1 − β), significance level α, and effect size. Supports Cohen's d for t/z-tests, Cohen's h for proportion tests, and Cohen's w for chi-square tests. Also calculates achieved power for a given n.

Formula

n = ((z_{α/2} + z_β) / d)² — one-sample; n = 2×((z_{α/2} + z_β) / d)² — two-sample; power = Φ(|d|√(n/2) − z_{α/2})

How to Use

  1. 1

    Choose the test type: One-Sample Z, Two-Sample T, Proportion Z (Cohen's h), or Chi-Square.

  2. 2

    Select direction: Two-Sided (H₁: μ ≠ μ₀) or One-Sided (H₁: μ > μ₀).

  3. 3

    Choose mode: Find n (enter desired power) or Find Power (enter known n).

  4. 4

    Click Small / Medium / Large to fill in Cohen's effect size convention, or type a custom value.

  5. 5

    Enter α (significance level, commonly 0.05) and the desired power (commonly 0.80 or 0.90).

  6. 6

    Read the required sample size, achieved power, critical value, and β error from the results.

  7. 7

    Review the power curve table to see how power grows with increasing n for your effect size.

Select the test type, direction, and mode (find n or find power), then enter the effect size, α, and power or n.

Example Calculation

Medium effect, two-sample t-test: d = 0.5, α = 0.05, desired power = 0.80. n per group = 2 × ((1.96 + 0.8416) / 0.5)² = 2 × (5.603)² ≈ 2 × 31.4 ≈ 63 total, 32 per group. For 95% power with the same effect: n per group ≈ 53.

Understanding Power Analysis | Sample Size, Effect Size & Statistical Power

Cohen's Effect Size Conventions

Jacob Cohen established widely used benchmarks for effect size magnitudes. These are guidelines, not rules — context determines what constitutes a meaningful effect in your field.

StatisticTestSmallMediumLarge
Cohen's dt-test / z-test0.20.50.8
Cohen's hTwo proportions0.20.50.8
Cohen's wChi-square / goodness of fit0.10.30.5
Cohen's fANOVA / F-test0.10.250.4
Pearson rCorrelation0.10.30.5

Sample Size Quick Reference (Two-Sample t-test, α = 0.05, Two-Sided)

Required n per group for two-sample t-test with equal groups. Values computed from n = 2 × ((z_α/2 + z_β) / d)².

Effect size dPower 70%Power 80%Power 90%Power 95%
0.2 (small)265394527651
0.3118175234290
0.5 (medium)436485105
0.8 (large)17253442
1.011172228
1.5581013

Type I vs Type II Errors

  • Type I error (α, false positive): rejecting H₀ when it is true. Controlled directly by setting α = 0.05 or 0.01.
  • Type II error (β, false negative): failing to reject H₀ when it is false. Power = 1 − β. Industry standard: β ≤ 0.20 (power ≥ 80%).
  • Effect size d = (μ₁ − μ₀)/σ measures how far apart the null and alternative means are in standard deviation units.
  • Cohen's h for proportions: h = 2 arcsin(√p₁) − 2 arcsin(√p₀). The arcsin transformation stabilises the variance.
  • Increasing sample size n reduces β (raises power) without affecting α, the most direct lever for improving study sensitivity.
  • One-sided tests require smaller n than two-sided for the same power, but are only valid when direction is pre-specified.
  • Pilot studies are often used to estimate effect size; if the pilot d is inflated by sampling error, the main study will be underpowered.
  • For clinical trials, regulatory agencies typically require power ≥ 80% at α = 0.05 (two-sided) for primary endpoints.

Frequently Asked Questions

What is statistical power and why does it matter?

Power (1 − β) is the probability of correctly detecting a real effect. A study with 80% power will miss 20% of true effects. Low-powered studies waste resources and produce unreliable results — if they do find an effect, it may be inflated; if they do not, the null result is inconclusive.

What is Cohen's d?

Cohen's d = (μ₁ − μ₀) / σ, the difference between means divided by the pooled standard deviation. It is a dimensionless measure of effect size. d = 0.2 is small (like a 2-point IQ difference), d = 0.5 is medium, and d = 0.8 is large.

Why do small effects require such large samples?

Sample size scales as 1/d². A small effect d = 0.2 requires 25× more participants than a large effect d = 1.0 for the same power. Detecting a 1% improvement in a process requires thousands of observations; detecting a 20% improvement may need only tens.

What is the difference between one-sided and two-sided tests?

A two-sided test asks "Is μ ≠ μ₀?" and splits α between both tails. A one-sided test asks "Is μ > μ₀?" and puts all α in one tail. One-sided tests are more powerful but only valid when direction is pre-specified before seeing the data.

Can I use this calculator for clinical trial planning?

Yes, the z-test approximation is standard for clinical trial sample size estimation. For superiority trials use d and the two-sample formula. Regulatory submissions (FDA, EMA) typically require power ≥ 80% at α = 0.05 two-sided. Consult a biostatistician for complex designs.

You Might Also Like

Explore 360+ Free Calculators

From math and science to finance and everyday life — all free, no account needed.