Free research tool
Sample Size Calculator
Work out how many completed responses your survey needs for a given margin of error and confidence level, and how many people you need to reach once incidence and drop-off are counted. The classic answer: 385 responses for ±5% at 95% confidence.
Free, no sign-up. Last reviewed against source documents on .

How close to the true value you need results to be. ±5% is common, ±3% for headline numbers.
Leave blank for a large population, like all Indian adults or all app users.
50% is the safe default: it gives the largest sample.
Sample size table
Completed responses needed for a large population, assuming a 50/50 result (the most conservative case).
| Margin of error | 90% confidence | 95% confidence | 99% confidence |
|---|---|---|---|
| ±1% | 6,764 | 9,604 | 16,588 |
| ±2% | 1,691 | 2,401 | 4,147 |
| ±3% | 752 | 1,068 | 1,844 |
| ±4% | 423 | 601 | 1,037 |
| ±5% | 271 | 385 | 664 |
| ±7.5% | 121 | 171 | 295 |
| ±10% | 68 | 97 | 166 |
| Population | ±5% | ±3% | ±2% |
|---|---|---|---|
| 100 | 80 | 92 | 97 |
| 500 | 218 | 341 | 414 |
| 1,000 | 278 | 517 | 707 |
| 5,000 | 357 | 880 | 1,623 |
| 10,000 | 370 | 965 | 1,937 |
| 100,000 | 383 | 1,056 | 2,345 |
| 1,000,000 | 384 | 1,066 | 2,396 |
Above roughly 100,000 people, population size barely changes the answer.
How the calculation works
z is 1.645 for 90%, 1.960 for 95% and 2.576 for 99% confidence. p is the result you expect (0.5 if unsure). e is the margin of error as a decimal. N is the population, DEFF the design effect.
Worked example: 95% confidence, ±5%, p = 0.5: 1.96² × 0.25 ÷ 0.0025 = 384.2 (384.1 with the exact z of 1.95996), rounded up to 385. If only 25% of people qualify and 80% of those finish, reach 385 ÷ 0.25 ÷ 0.8 = 1,925 people.
Plan for subgroups. The margin applies to the whole sample. If you need to compare, say, North vs South, size each region for the precision you need, then check the gap with the significance calculator. Already have data? Use the margin of error calculator.
Questions, answered
How many responses do I need for a survey?
For a large population, 385 completed responses give a margin of error of ±5% at 95% confidence, and 1,068 give ±3%. Those numbers assume the most conservative case (a 50/50 result) and a simple random sample. If you will report results for subgroups, such as men and women or four regions, each subgroup needs its own sample of that size.
What is the formula for survey sample size?
n₀ = z² × p × (1 − p) × DEFF ÷ e², where z is the z-score for the confidence level (1.96 for 95%), p is the expected proportion (use 0.5 if unsure), e is the margin of error as a decimal and DEFF is the design effect (1 for a simple random sample). For a known population N, then apply the finite population correction: n = n₀ ÷ (1 + (n₀ − 1) ÷ N).
Does population size matter?
Much less than people expect. At 95% confidence and ±5%, a population of 10,000 needs 370 responses and a population of 10 million needs 385. Population size only changes the answer when you are surveying a large share of a small group, such as the employees of one company.
What is incidence rate and why does it change how many people I need to reach?
Incidence is the share of people you reach who qualify for the survey, for example 25% if only a quarter of adults bought a car in the last year. If you need 385 completes from a 25% incidence group, and 80% of qualified people finish, you need to reach about 1,925 people. This is also what drives panel cost: lower incidence means more people screened per complete.
What is a design effect?
The design effect (DEFF) is how much weighting, clustering or quota sampling inflates the variance compared with a simple random sample. A DEFF of 1.5 means you need 1.5 times as many responses for the same precision. Weighted online samples commonly sit between 1.2 and 2.0.
Should I use 95% or 99% confidence?
95% is the standard in market and social research. 99% is stricter and costs more: ±5% needs 664 responses at 99% versus 385 at 95%. 90% is sometimes used for quick directional reads.
Sources and method
Uses the normal approximation for a proportion (Cochran’s formula), multiplied by the design effect, then the finite population correction, rounded up. Every figure on this page is computed by the same code the calculator uses, which is tested against reference values from SciPy.
More free research tools

Margin of Error Calculator
How precise is your survey result? Get the ± and the range

Significance Calculator
Is the gap between two groups real? Percentages or averages

NCCS Calculator
Classify any Indian household into NCCS A1 to E3 in two questions

ISEC Calculator
MRSI’s new socio-economic classes, ISEC 1 to 12, for any household
Know your number? Field it in India.
Build the survey in PollPe, share it on WhatsApp, QR or link, and watch completes against your target live.