Sample size for ±1% margin of error at 99% confidence
For a margin of error of ±1% at 99% confidence you need a sample of 16,588 (using the worst case p = 0.5 and rounding up). If you expect a proportion nearer 0 or 1, fewer are enough.
How to work it out
- 1
The margin of error for a proportion is E = z*·√(p(1−p)/n). Solve it for n.
- 2
The critical value for 99% confidence is z* = 2.576. The margin of error is E = 1% = 0.01.
- 3
Without a prior estimate use p = 0.5, the worst case, which gives the largest n.
- 4
Always round up to the next whole person.
LaTeX source of the formulas on this page
\[ n = \frac{z^{*2}\,p(1-p)}{E^2} = 16,588 \]
\[ n = \frac{(z^{*})^{2}\,p(1-p)}{E^{2}} \]
\[ z^{*} = 2.576,\qquad E = 0.01 \]
\[ n = \frac{2.576^2 \times 0.5 \times 0.5}{0.01^2} = 16587.24 \]
\[ n = 16,588 \]Sample size for ±1% at 99% confidence, by expected proportion
| Expected proportion p | Sample size n |
|---|---|
| 0.5 | 16,588 |
| 0.4 | 15,924 |
| 0.3 | 13,934 |
| 0.2 | 10,616 |
| 0.1 | 5,972 |
| 0.05 | 3,152 |
Sample size for ±1% at other confidence levels
| Confidence | Sample size n (p = 0.5) |
|---|---|
| 80% | 4,106 |
| 90% | 6,764 |
| 95% | 9,604 |
| 98% | 13,530 |
| 99% | 16,588 |
Questions
How many people do I need for a margin of error of 1% at 99% confidence?
16,588 (worst case p = 0.5).
Why use p = 0.5?
Because p(1 − p) is largest at 0.5, so the sample size is large enough whatever the true proportion turns out to be.
What if the population is small?
If the sample is more than about 5% of the population, apply the finite population correction n′ = n / (1 + (n − 1)/N), which lowers the number needed.