Sample size reference table.
How many responses you need for a given confidence level and margin of error, computed for a proportion at the worst case (p = 0.5).
This table gives the sample size needed to estimate a proportion at common confidence levels and margins of error, using the conservative worst case (p = 0.5). It assumes a large population; for small populations the requirement is lower.
| Confidence | ±1% | ±2% | ±3% | ±5% | ±10% |
|---|---|---|---|---|---|
| 90% | 6,764 | 1,691 | 752 | 271 | 68 |
| 95% | 9,604 | 2,401 | 1,068 | 385 | 97 |
| 99% | 16,588 | 4,147 | 1,844 | 664 | 166 |
↓ Download as CSV · open data, free to reuse (CC BY 4.0)
n = z² × p(1 − p) / E², with p = 0.5 (worst case) and z the critical value for the confidence level.
How to cite it.
Conyso. (2026). Sample size reference table [Data set]. Conyso. https://conyso.com/reference-sample-size-table.html
DOI: pending — Zenodo archival plannedThe free sample size calculator computes the exact number for your confidence, margin, and population.
Reading the table.
Pick your confidence level (how sure you want to be) down the side and your margin of error (how precise) across the top. The cell is the number of responses needed. Notice the cost of precision: halving the margin of error quadruples the sample size.
This uses p = 0.5, which maximises the required sample, so it is a safe upper bound. If you know the true proportion is far from 50%, you can get away with fewer.
Sample size, answered.
How many people do I need to survey?
For a ±5% margin at 95% confidence you need about 384 responses regardless of population size (for large populations). Tighter margins need many more.
Why does p = 0.5 give the largest sample?
Because p(1 − p) is maximised at p = 0.5, so assuming it gives the most conservative, safest sample size when you do not know the true proportion.
Does population size affect the sample size?
Very little for large populations. The formula barely moves once the population is in the tens of thousands, which is why ±5% at 95% needs about 384 whether the population is 50,000 or 50 million. For small finite populations you apply a correction and need fewer.
What sample size do I need for a 99% confidence level?
More than for 95%, because the critical z rises from 1.96 to 2.576. At ±5% margin you need about 666 responses at 99% confidence, versus 384 at 95%.