Skip to content
OneKitly

A Poll of a Thousand Is Just as Accurate in a Country of Sixty Million

Published 11/17/2026 · 3 min read · Everyday calculators

Daniel Okonkwo

Daniel Okonkwo — Front-end developer and tech writer at OneKitly

Web performance · File formats

Checked against 2 sources

View profile →
In short

The width of a polling margin depends on the size of the sample and essentially not at all on the size of the population behind it. A thousand interviews finding 500 in favour give 50% with a 95% interval of 46.9 to 53.1 — a margin of 3.10 points — whether the country has sixty thousand voters or sixty million. Reduce the sample to a hundred and the margin grows to 9.80 points, three times wider, because precision improves with the square root of the sample. To halve a margin you must quadruple the interviews, and that arithmetic is why national polls cluster around a thousand respondents.

Advertisement

500 out of 1000 gives 50% with a margin of 3.10 points. Cut the sample to 100 and the margin becomes 9.80 — three times wider, from a tenth of the interviews.

Why the population size drops out

Each interview is an independent draw, and what it tells you does not depend on how many other people were not asked. There is a correction for sampling a large fraction of a small group — asking 400 people out of 500 really does give a tighter answer than asking 400 out of a million — but it only bites when the sample exceeds a few per cent of the population. At a thousand out of sixty million the correction changes the margin in the sixth decimal place, which is why it is left out of every published figure.

The margin is the smallest of the errors

Sampling error is the only one that can be computed, which is why it is the only one quoted. Everything else in a poll — who refuses to answer, how a question is worded, whether the people reached resemble the people who vote, how undecideds are allocated — contributes error that no formula covers and that has repeatedly exceeded the published margin. A poll reported as accurate to three points is accurate to three points in one respect and unquantified in every other, and the honest reading treats the interval as a floor rather than a bound.

Margin
Margin at 50% and 95% confidence
InterviewsMarginInterval
1009.80 pts40.2 to 59.8
10003.10 pts46.9 to 53.1
40001.55 pts48.4 to 51.6

Worked with our own calculator

Confidence interval for a proportion

Given

Successes (x)
42
Sample size (n)
100
Confidence level
80%

Result

Sample proportion
42%
Wald lower
35.67%
Wald upper
48.33%
Wilson lower
35.85%
Wilson upper
48.4%

These figures are produced by the calculator below, not typed in by hand — they are recomputed whenever the tool changes.

Run it on your own figures →
Advertisement

Frequently asked questions

Why is the margin widest at fifty per cent?
Because a fifty-fifty split is the most variable outcome. A proportion near zero or near one has little room to move — if 99 out of a hundred agree, the next hundred will look much the same — while a proportion near a half swings the most from sample to sample. Published margins are usually quoted at 50% precisely because that is the widest case, so the real margin on a 20% figure is narrower than the headline claims. The tool shows the interval for the proportion actually observed rather than the worst case.
Is a lead of two points inside the margin meaningless?
Not meaningless, just not conclusive. A two-point lead inside a three-point margin still makes the leader more likely to be ahead than behind — often around a 70% chance rather than a coin flip. Reporting it as a statistical tie throws away real information, and reporting it as a lead overstates it. The comparison of two proportions from the same poll also has its own margin, wider than either alone, which is a further step most coverage skips.

Articles you may find interesting

All guides →

Related tools

Sources

Spotted a mistake in this article?