Glossary

Margin of error

A statistic expressing how much a poll's result might differ from the true value because only a sample was surveyed.

Also called: sampling error, confidence interval, MOE

The margin of error is a measure of the uncertainty in a poll caused by surveying a sample rather than everyone. A poll reporting a candidate at 48 percent with a margin of error of plus or minus 3 points means that, if the same method were repeated many times, the true value would fall between 45 and 51 percent in about 95 percent of the samples.

It depends mostly on sample size: the larger the sample, the smaller the margin, but returns diminish. Typical polls of about 1,000 respondents have a margin near 3 points. The margin for a subgroup, such as women under 30, is larger because the sample is smaller, and the margin on a lead is roughly double that for a single candidate's share.

The statistic captures only random sampling error. It does not cover other sources of mistakes, such as who answers (nonresponse bias), how likely voters are chosen, question wording or weighting choices, and these can matter more. Pollsters also adjust results with weighting, which raises the effective margin. Commentators often misinterpret a "statistical tie," and polls in close races often fail to determine a leader. The generic ballot and an exit poll have their own uncertainties.

Last reviewed November 8, 2026

In the index now

Related terms

Likely voter A pollster's estimate of who will actually vote, used to narrow a sample of adults or registered voters. Generic ballot A poll asking voters whether they would vote for a Democrat or a Republican for Congress without naming candidates. Exit poll A survey of voters as they leave polling places, used to explain who voted and why rather than to project results alone. Bellwether A place or group whose election results have historically matched the national outcome and are watched as an early indicator.