Sample Size Calculator
This calculator plans the number of completed independent responses needed to estimate a population proportion at a selected confidence level and margin of error. It uses the normal z planning approximation, optionally adjusts for a finite population sampled without replacement, and always rounds the final requirement upward.
By Dr. Hassan Khames EliwaLast updated
Result
- Unrounded large-population size
- 384.145882
- Unrounded planned size
- 384.145882
- Planning assumption
- Independent simple random sample
How it works
Start with a precision target
Enter the margin in percentage points, not as a decimal probability. Five means plus or minus five percentage points. Convert this to e = 0.05 for the formula. Enter the expected proportion as a percentage as well, so 50 becomes p = 0.5. With critical value z, the large-population planning size is n₀ = z² × p × (1−p) / e².
The available two-sided confidence levels are 90%, 95%, and 99%, with full-precision normal critical values approximately 1.644854, 1.959964, and 2.575829. The formula is a precision calculation for one proportion, not a power calculation for detecting a difference between groups. It does not plan a mean, regression model, or cluster trial.
Why the default proportion is 50%
The product p(1−p) is largest at p = 0.5. Choosing 50% therefore gives the largest large-population sample requirement for a given margin and confidence. If you have credible prior information, a different expected proportion can reduce the plan. Do not use a convenient extreme estimate merely to make recruitment look achievable. Endpoints of zero and 100% are excluded because they misleadingly imply no planning variance.
Finite populations and the last rounding step
For a known population N and simple random sampling without replacement, the corrected size is n = n₀ / (1 + (n₀−1)/N). Apply the correction to the unrounded number, then take the ceiling once. A blank population field means the correction is not applied. Population means every eligible unit in the sampling frame, not the number invited or the number you hope will reply.
The planning approximation is useful but does not guarantee a realized Wilson interval of exactly the target width. Actual counts, discreteness, and the final estimation method matter. NIST explains how confidence levels connect to interval construction; the confidence interval calculator evaluates the collected data afterward.
Worked example
Scenario 1: a large customer population
Plan to estimate a binary satisfaction proportion at 95% confidence within five percentage points. Leave population blank and use the conservative 50% estimate. The unrounded calculation is 384.145882. Taking the ceiling gives 385 required completed responses. This is not 385 invitations unless every invitation yields an eligible completed response.
Scenario 2: an inventory of 1,000 records
Keep the same confidence, margin, and expected proportion, but enter population 1000. Applying the finite population formula to 384.145882 before rounding gives 277.733453. The requirement becomes 278 records sampled without replacement. The reduction comes from observing a meaningful fraction of a fixed finite population. It would be inappropriate if records were sampled with replacement or the target were an ongoing stream of future records.
Expert tips
Common mistakes
- Do not enter 0.05 when you mean five percentage points. Here 0.05 requests a much narrower margin, because the field is already expressed in percentage points.
- Do not round to the nearest whole person. Any fractional requirement must be rounded upward; rounding down can miss the stated planning target even under the approximation.
- Do not apply the finite correction simply because your contact list is short. The list must adequately represent the actual finite population to which the conclusion is intended to apply.
Frequently asked questions
Is 385 always enough?
No. That number belongs to a particular combination of 95% confidence, five-point margin, 50% planning proportion, and independent random sampling. Tighter precision, higher confidence, clustered designs, or subgroup reporting can require substantially more observations.
Can I plan separate subgroups?
Run a separate calculation for each subgroup that needs its own precision target. A sample adequate for an overall proportion may leave very few observations in smaller segments. The overall total alone does not guarantee useful subgroup intervals.
What inputs are bounded?
Margins run from 0.01 to 99 percentage points, expected proportions from 0.01% to 99.99%, and finite populations from one to one billion whole units. These limits keep calculation bounded; they do not certify every combination as statistically practical.
Does higher confidence reduce the sample?
No. Holding margin and expected proportion fixed, higher confidence increases the critical value and therefore increases the sample requirement. A wider confidence procedure needs more information to keep the same precision target.
Is this an exact binomial sample calculation?
No. It is a normal-approximation planning calculation. Rare outcomes, small populations, strict coverage requirements, and regulated studies may require exact or specialized design methods rather than this convenient first-pass estimate.
Related calculators
Related guides
MathSample size and confidence intervals without the jargonSeparate sample planning from interval estimation, and make the assumptions visible before reporting a margin of error.15 November 2026 · 8 min read
MathStandard deviation without the jargonThe average tells you the centre. Standard deviation helps describe the spread, provided you know what the data represents.14 November 2026 · 7 min read
MathPercentages people get wrong: change, difference and pointsThe denominator determines the answer. Learn to distinguish percentage change, symmetric difference and percentage-point movements.14 November 2026 · 7 min readSources and standards
About the author
Dr. Hassan Khames Eliwa holds a PhD in Construction & Built Environment and has more than 20 years of international project controls experience. He is Director of Eliwa Project Controls Limited in Auckland and founder of PMMilestone.org and PMMilestone.com.
Buy me a coffee