Quantitative research tool
Sample size and margin of error calculator
Plan a robust quantitative study or assess the precision of an existing sample. Get an immediate estimate, understand the assumptions behind it and see what the number means for your research design.
Default settings use a 95% confidence level and the most conservative 50% expected proportion.
Quick reference
Completed interviews needed for a 95% confidence level, a 50% expected proportion, a design effect of 1 and a large population. Other settings change these numbers: use the calculator above.
| Margin of error | Completed interviews |
|---|---|
| ±10% | 97 |
| ±7.5% | 171 |
| ±5% | 385 |
| ±4% | 601 |
| ±3% | 1,068 |
| ±2% | 2,401 |
What the margin of error does not measure
The result is a theoretical sampling estimate based on a simple random probability sample. It does not account for coverage error, non-response bias, questionnaire wording, respondent quality or other non-sampling errors.
- Access panels and quota samples may not meet probability-sampling assumptions; their theoretical margin of error should be treated as an indicative precision measure.
- Weighting and complex sample designs can reduce the effective sample size, which is why the design-effect option is provided.
- A smaller margin of error does not compensate for biased recruitment or poor questionnaire design.
- Comparisons between groups, mean estimates, experiments and significance tests may require a different power or sample-size calculation.
How the calculation works
The calculator uses the standard formula for a proportion, then applies the corrections you select.
Initial sample
n₀ = z² × p × (1 − p) / e²z is 1.96 at 95% confidence, p the expected proportion and e the margin of error, both as decimals.
Finite population correction
n = n₀ / (1 + (n₀ − 1) / N)Applied only when you enter a population size N.
Design effect
n × DEFFThe result is then rounded up to the next whole interview.
Margin of error
MOE = z × √(p × (1 − p) × DEFF / n)Multiplied by √((N − n) / (N − 1)) when the population is known.
Worked example: at 95% confidence, ±5% and p = 50%, n₀ = 1.96² × 0.5 × 0.5 / 0.05² = 384.1, rounded up to 385 completed interviews.
Frequently asked questions
How many respondents do I need for a survey?
For a large population, 385 completed interviews give a margin of error of ±5% at a 95% confidence level, assuming the most conservative 50% proportion. You need 1,068 for ±3% and 97 for ±10%.
Does population size change the sample size?
Only when the population is small. Above a few tens of thousands, the requirement barely moves. For a B2B customer base of 2,000 companies, ±5% at 95% needs 323 completed interviews instead of 385; for 500 companies, it needs 218.
What is a good margin of error?
As an indicative guide: up to ±3% for high-precision tracking, ±3% to ±5% for most business decisions, ±5% to ±8% for directional reads, and above ±8% for exploratory work.
I need results for each region of Belgium. How large should the sample be?
If Brussels, Flanders and Wallonia each need ±5% at 95%, plan about 385 completed interviews per region, 1,155 in total. Splitting a single sample of 385 into three equal regions gives each region a margin of about ±8.7%.
What is a design effect?
Weighting, clustering or strict quotas reduce the effective sample size. A design effect of 1.5 means you need 1.5 times more interviews: 577 instead of 385 for ±5% at 95%.
Does the margin of error apply to online panels?
The formula assumes a random probability sample. Access panels and quota samples rarely meet that assumption, so treat the result as an indicative precision measure, and give as much attention to recruitment and questionnaire quality.
Turn the number into a fieldwork-ready design
Sample size is one decision among many: audience, sourcing, quotas, languages, questionnaire and analysis. We can help with all of them.