Sample Size Calculator

Single proportion

Determines the sample size required to test whether a population proportion differs from a specified hypothesized or reference proportion, given a chosen significance level and statistical power.

Key takeaways

  • What is calculated: Determine the smallest n at which the test of H₀ attains the specified power.
  • Quantities you must supply: Reference proportion; Anticipated proportion; Significance level; Power.
  • Type of calculation: Power-based: it answers how much data is needed to detect a stated effect. Power 0.80 at α = 0.05 is the usual target; confirmatory studies use 0.90.
  • Calculation routes: 3 are available for this design; the one selected determines the formula, the quantities requested and the result.

Methods available for Single proportion

  • Normal approximation (z-based) Closed-form sample size approximating the binomial by a normal distribution. Chosen for simplicity and comparability with published tables when the expected-count conditions are satisfied.
  • Normal approximation with continuity correction Normal approximation adjusted for the discreteness of the binomial distribution. Chosen as a middle course: more conservative than the plain approximation, without the iteration of the exact method.
  • Exact binomial (Clopper-Pearson style) Sample size derived directly from the binomial distribution, with no normal approximation. Chosen because it makes no distributional approximation and guarantees the actual type I error does not exceed the nominal level — which the normal approximation cannot promise.

When to use

Use when the primary objective is to determine whether a population proportion differs from a specified reference or hypothesized proportion with adequate statistical power.

Sample size

Type I error rate.

Probability of detecting the specified effect.

A two-sided alternative splits α between both tails; a one-sided alternative places all of α in one tail, needs a smaller n, and must be pre-specified.

Final n is inflated by 1/(1 − dropout).

Reset to defaults

Awaiting inputs

—

Set the parameters and calculate.

Methodology — Single proportion — normal approximation

Formula used Single proportion — normal approximation

n=[z1−α/2p0q0+z1−βp1q1]2(p1−p0)2
Normal approximation, q = 1 − p
  • z1−α/2(two-sided)in place ofz1−α(one-sided) Two-sided alternative selected: α is split between both tails, so the larger quantile is used and the requirement rises

where —

p₀
Hypothesised reference proportion under H₀. 0 < p₀ < 1
p₁
Anticipated proportion under the alternative. 0 < p₁ < 1, p₁ ≠ p₀
α
Probability of rejecting a true null hypothesis (Type I error). 0.0001 – 0.5; conventionally 0.05, or 0.025 one-sided for regulatory non-inferiority
1 − β
Probability of rejecting the null hypothesis when the specified alternative is true. 0.50 – 0.9999; conventionally 0.80 or 0.90
ℓ
Expected proportion of enrolled units lost before analysis. 0 – 0.95

Formulas for the 3 methods

Calculation methodFormula
Normal approximation (z-based)
n=[z1−α/2p0q0+z1−βp1q1]2(p1−p0)2
Normal approximation with continuity correction
ncc=n4(1+1+4n|p1−p0|)2
Exact binomial (Clopper-Pearson style)
1−β=Pr(X≥c∣n,p1)

How this method works

The binomial sampling distribution is approximated by a normal one. Because binomial variance depends on the proportion itself, the null variance p₀q₀ and the alternative variance p₁q₁ enter separately.

Two-sided test: H₀: p = p₀ against H₁: p ≠ p₀; the α-quantile is z₁₋α/₂.

Selected method — normal approximation. The binomial is approximated by a normal distribution. Because binomial variance depends on the proportion itself, the null variance p₀q₀ and the alternative variance p₁q₁ both appear.

The approximation is unreliable when any of np₀, n(1−p₀), np₁, n(1−p₁) falls below about 5. All four are reported below, and a warning is shown when the condition fails.

Null hypothesis. H₀: p = p₀

Alternative hypothesis. Two-sided H₁: p ≠ p₀. One-sided H₁: p > p₀ or p < p₀.

Calculation procedure

  1. State the reference proportion p0 and the proportion p1 you want to detect.
  2. Take z1−α/2 and z1−β.
  3. Use the null variance under p0 and the alternative variance under p1 rather than one pooled value.
  4. Evaluate n=[z1−α/2p0(1−p0)+z1−βp1(1−p1)]2(p1−p0)2.
  5. Round up, then inflate for dropout. Check that np0 and n(1−p0) both exceed about 5, or switch to the exact method.

Exact or approximate

Approximate.

The formula shown always matches what was computed: when a one-sided test is selected the rendered quantile changes from z1−α/2 to z1−α, and where several methods exist, the formula follows the method selected in the calculator.

Advantages & limitations

Advantages

  • Closed form and reproducible by hand.
  • Adequate for mid-range proportions with moderate sample sizes.

Limitations

  • Unreliable for extreme proportions or small n.
  • Ignores the discreteness of the binomial, so the true type I error differs from nominal.

Assumptions

  • Observations are independent Bernoulli trials with constant probability.
  • Expected counts are large enough for the normal approximation to hold.
  • Simple random sampling with no clustering.

Applicable adjustments

Supports dropout inflation and, where the sampling framework applies, the finite population correction.

Conclusion

Adequate for mid-range proportions with moderate sample sizes. Check the four expected counts reported with the result; if any falls below about five, switch to the exact binomial method.

References

  1. Fleiss JL, Levin B, Paik MC. Statistical Methods for Rates and Proportions. 3rd ed. Hoboken: Wiley; 2003.
  2. Clopper CJ, Pearson ES. The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika. 1934;26(4):404–413. Link
  3. Chow S-C, Shao J, Wang H, Lokhnygina Y. Sample Size Calculations in Clinical Research. 3rd ed. Boca Raton: Chapman & Hall/CRC; 2017.