A walkthrough, end to end.
- 1
Enter desired margin of error (decimal, e.g. 0.05 for ±5%), confidence level, expected proportion, and population size (optional).
- 2
The calculator returns the minimum sample size needed.
- 3
If population size is given, applies finite population correction.
Sample size formula
Infinite population: n = (z* / E)² × p(1−p). Finite population: n_adj = n / (1 + (n−1)/N). Use p = 0.5 for most conservative estimate (maximum n).
What you can do with this.
National survey sampling
5% margin of error, 95% confidence, p=0.5: n = (1.96/0.05)² × 0.25 = 384.16 → 385 respondents. The classic political poll baseline.
Tighter precision
Drop margin of error to 3% → n = (1.96/0.03)² × 0.25 ≈ 1068. Halving the MOE quadruples the sample size — sample size scales as 1/E².
Higher confidence
99% confidence: z* = 2.576. n = (2.576/0.05)² × 0.25 = 663. About 1.7× the 95% sample size — confidence costs sample.
Finite population correction
When sampling more than ~5% of population, the FPC reduces required n. For a population of 1,000, the calculator's adjusted n will be smaller than the infinite-population baseline.
Known proportion
If pre-existing data suggests p ≈ 0.3, use that instead of 0.5. Smaller p (or larger 1−p) reduces variance and required n. Conservative default is p = 0.5.
A/B testing sample size
Different formula — you need to detect a specified difference with given power. Tools like Optimizely, VWO, and statistical-power calculators handle this.
Drop-out / non-response buffer
Recruit MORE than the calculated n. If 20% drop out, target 1.25× n. Real-world surveys regularly inflate by 20–50% to account for non-response.
Sample size 2026 — what's current
Online surveys via Pollfish, SurveyMonkey, Google Surveys make hitting target n affordable. Calculator gives you the target; recruitment platforms handle delivery.
Frequently asked.
Use 0.5 (the most conservative — gives maximum sample size). If you have prior data suggesting p ≈ 0.3, use that for smaller required n.
When sample is more than ~5% of population. For samples of national populations, infinite-population formula is fine. For small organizations/groups, FPC reduces required n materially.
This calculator is proportion-based (most surveys). For continuous variables (mean of weights, scores), use a different formula based on σ and desired margin.
No. Calculations run entirely in your browser.