Margin of Error Calculator
Type how many people answered and the result to get the margin of error…
Type how many people answered and the result to get the margin of error, the plus or minus a poll reports, and see how it shrinks with more people.
More optionsLarge population
ExampleMargin of error, 95%
±3.1%
A result of 50% from 1,000 people is likely within 3.1% of the true figure: 46.9% to 53.1%.
Type your own numbers, or drag the chart.
Drag along the curve: bigger samples, smaller margins, with fast-falling returns.
- Standard error
- 1.58%
- z for the confidence level
- 1.96
- With twice the sample
- ±2.19%
How this works: 8 sources,
How this works
Method
The margin of error is the “plus or minus” on a survey result: the half-width of its confidence interval. For a percentage it is z × √(p(1 − p)/n). A poll of 1,000 people with 50% saying yes, at 95% confidence, has a margin of 1.96 × √(0.25/1,000) ≈ 0.031, so ±3.1 points: the true figure is likely between 46.9% and 53.1%. Full method
Sources
- Wikipedia, Margin of error
- NIST/SEMATECH e-Handbook of Statistical Methods
- Abramowitz and Stegun, Handbook of Mathematical Functions, chapters 6, 7 and 26 (1964)
- SciPy 1.18 scipy.stats (reference values for the tests)
- statsmodels 0.15 (reference values for the A/B test and regression intervals)
- Welch, The generalization of Student’s problem when several different population variances are involved (Biometrika) (1947)
- Fleiss, Levin and Paik, Statistical Methods for Rates and Proportions, 3rd edition (sample size for two proportions) (2003)
- Hyndman and Fan, Sample quantiles in statistical packages (The American Statistician) (1996)
How it’s tested
2 worked examples for this page are checked by automated tests before every release: given the inputs, the tool must show the expected answer.
Privacy
On your device. Nothing you enter or open leaves this browser. About privacy labels
Changes
- First published: live curves with draggable cut-offs, and tests on pasted data (t-tests, ANOVA, chi-square, Fisher’s exact, correlation, regression, A/B), checked against SciPy.
How to calculate the margin of error
The margin of error is the “plus or minus” on a survey result: the half-width of its confidence interval. For a percentage it is z × √(p(1 − p)/n). A poll of 1,000 people with 50% saying yes, at 95% confidence, has a margin of 1.96 × √(0.25/1,000) ≈ 0.031, so ±3.1 points: the true figure is likely between 46.9% and 53.1%.
It shrinks with the square root of the sample size. Going from 1,000 to 4,000 people halves it to about ±1.55 points, while going from 100 to 400 takes it from ±9.8 to ±4.9. The chart above shows the steep early gains and the long flat tail where extra responses buy little.
The margin of error covers only random sampling error. It says nothing about questions that lead, people who won’t answer, or a sample that isn’t representative. And the margin on the gap between two candidates is roughly twice the margin on each one’s share.
E = z* × √(p(1 − p) ÷ n) · for a mean E = z* × σ ÷ √n
- For a simple random sample of about 1,000 people, the 95% margin of error on a percentage near 50% is about ±3 percentage points. Source: Wikipedia, Margin of error.
Margin of error by sample size
The 95% and 99% margin of error for a percentage near 50%, by number of responses.
| Responses | 95% margin | 99% margin | Range for a 50% result (95%) |
|---|---|---|---|
| 50 | ±13.9% | ±18.2% | 36.1% to 63.9% |
| 100 | ±9.8% | ±12.9% | 40.2% to 59.8% |
| 200 | ±6.9% | ±9.1% | 43.1% to 56.9% |
| 300 | ±5.7% | ±7.4% | 44.3% to 55.7% |
| 400 | ±4.9% | ±6.4% | 45.1% to 54.9% |
| 500 | ±4.4% | ±5.8% | 45.6% to 54.4% |
| 750 | ±3.6% | ±4.7% | 46.4% to 53.6% |
| 1,000 | ±3.1% | ±4.1% | 46.9% to 53.1% |
| 1,500 | ±2.5% | ±3.3% | 47.5% to 52.5% |
| 2,000 | ±2.2% | ±2.9% | 47.8% to 52.2% |
Show all 13 rowsShow fewer
| 3,000 | ±1.8% | ±2.4% | 48.2% to 51.8% |
| 5,000 | ±1.4% | ±1.8% | 48.6% to 51.4% |
| 10,000 | ±1% | ±1.3% | 49% to 51% |
Common questions
Is a 2-point lead inside the margin of error a tie?
Not exactly, but it is not a clear lead either. The margin on a lead is roughly twice the poll’s stated margin, so a 2-point lead in a ±3 poll is well within the noise.
Why does the result percentage matter?
p(1 − p) is largest at 50%, so results near 50% have the widest margin. A 10% result from the same 1,000 people has a margin of about ±1.9 points instead of ±3.1.
Does population size matter?
Hardly, once the population is many times the sample. It matters when you survey a big share of a small group; type the population under More options for the corrected margin.