Explainer
Margin of error: what the plus-or-minus in a poll means
A poll's margin of error describes how close its result can reasonably be expected to fall to the true value for the whole population, because the poll talked to only a sample.[1] A margin of plus or minus 3 points at the usual 95% confidence level means the same survey repeated 100 times would land within 3 points of the true value about 95 times.[2] The margin for the gap between two candidates is roughly twice the reported figure, subgroups have larger margins, and errors from wording, nonresponse or turnout models are not counted at all.[3][4][5]
Almost every published poll comes with a line such as “margin of error plus or minus 3 percentage points”. The figure exists because a poll interviews a sample rather than everyone, so its result will usually differ a little from the true value for the whole population.[1] This page explains what the number measures and the common ways it is misread. For how the sample is drawn in the first place, see how opinion polls work.
What the margin of error measures
The margin of sampling error describes how close a survey result can reasonably be expected to fall to the true population value.[1] Pollsters generally calculate it at a 95% confidence level.[6] A margin of plus or minus 3 points then means that if the same survey were run 100 times, the result would be expected to land within 3 points of the true value about 95 times.[2] About five times in 100 it will not, which is one reason the American Association for Public Opinion Research (AAPOR) says even the best surveys should be read with caution.[6]
AAPOR’s worked example is a statewide poll that finds 58% approval of the governor with a margin of plus or minus 3 points, which puts the true figure somewhere between 55% and 61%.[7] The margin is always expressed in percentage points, not as a percentage.[8] It also varies with the result being estimated: it is largest at 50%, and that is the figure pollsters usually report.[9]
Think of the margin of error as a range around the poll’s number rather than a single point. A governor “on 58%” approval with a 3-point margin is really somewhere between about 55% and 61%, and the poll cannot say exactly where.[7]
The US Census Bureau describes a confidence interval as a range of values that describes the uncertainty around an estimate, and notes that the interval is itself an estimate.[10] The Census Bureau routinely uses 90% intervals, and a higher confidence level produces a wider interval for the same estimate, so check which level a source uses before comparing margins.[11] The Australian Bureau of Statistics notes that for a normal distribution the 95% confidence interval spans two standard errors either side of the estimate.[12]
Leads, changes and “statistical ties”
The margin printed with a poll applies to each candidate’s share, not to the gap between them. To judge a lead, you need the margin of error for the difference, which is generally about twice as large.[3] In an example from Pew Research Center, a poll with a 3-point margin per candidate shows a 5-point lead, but the margin for the difference is about 6 points, so the true lead could plausibly be anywhere from -1 to +11.[13]
Changes between two polls are noisier still. In the same example, a 3-point shift in a candidate’s lead from one poll to the next carries a margin of plus or minus 8 points.[14] Some polls will also miss by more than their margin purely by chance: of the 590 national polls listed on the 2012 presidential race, about 30 would be expected to fall outside their margins, and Pew advises waiting to see whether a surprising result is repeated.[15] A pattern across many polls gives more confidence than one or two.[16]
Sample size and subgroups
Bigger samples have smaller sampling error, but the gains shrink quickly. Doubling a sample from 1,000 to 2,000 reduces the margin by only about one percentage point.[17] A simple random sample of 1,067 people gives the familiar plus or minus 3 points.[18]
Results for subgroups are less precise because they rest on fewer interviews.[4] In Pew’s example, a group of about 160 people within that sample has a margin of plus or minus 8 points for a single estimate and 16 points for the gap between two candidates.[4] AAPOR’s example is similar: in a 1,000-person poll with a 3-point margin, figures for a 200-person subgroup carry a margin of 6.9 points.[19]
Weighting and the design effect
Pollsters adjust, or weight, their samples to match the population, which helps avoid bias but also makes the margin of error larger. Statisticians call this the design effect, and a pollster that leaves it out of the reported margin is claiming more precision than the survey warrants.[20] AAPOR’s disclosure standards ask probability surveys to say whether their reported sampling error has been adjusted for it.[21] How surveys are weighted explains the adjustment itself.
What the margin of error leaves out
The margin of error applies only to probability-based surveys, where every person has a known, non-zero chance of selection, and not to opt-in online polls.[22] Some non-probability polls publish a “credibility interval” in similar plus-or-minus language, but AAPOR says it rests on assumptions that may be hard to validate and urges caution.[23]
Even for a well-run probability poll, sampling error is only one source of error. Question wording, who can be reached, how candid people are and models of who will turn out to vote can all push results off, and unlike sampling error these are hard to quantify and rarely reported.[5] AAPOR puts it plainly: there is no measurable overall margin of error for a poll.[24]
Checklist for reading a poll’s margin
- Double it for a lead. Compare the gap between candidates with about twice the reported margin.[3]
- Check the base. Subgroup results have larger margins than the headline figure.[4]
- Look for a trend, not one poll. A single surprising poll may be one of the expected outliers.[15][16]
- Ask what kind of sample it was. A margin of sampling error only applies to probability samples.[22]
- Remember the errors it leaves out. Wording, nonresponse and turnout models are not in the number.[5]
Read next
Next in the crash course
How surveys are weighted: raking, matching and propensity scores
Why pollsters weight survey data, how raking, matching and propensity weighting work, and what Pew's research shows weighting can and cannot fix.
Questions readers ask
What does "plus or minus 3 points" mean in a poll?
It means that if the same survey were fielded 100 times, the result would be expected to fall within 3 percentage points of the true population value in 95 of them. AAPOR's example is a poll finding 58% approval with a 3-point margin, which puts the true value somewhere between 55% and 61%.[2][7]
Is a 4-point lead "outside the margin of error" if the margin is 3 points?
Not necessarily. The margin of error for the difference between two candidates is generally about twice the margin reported for each one. In Pew's example, a 5-point lead with a 3-point margin per candidate could plausibly be anything from -1 to +11 points.[3][13]
Why are results for young people or other subgroups less reliable?
Subgroups contain fewer respondents, so their margins of error are larger. AAPOR's example is a 1,000-person poll with a 3-point margin in which a 200-person subgroup has a margin of 6.9 points.[4][19]
Does a bigger sample always mean a much more accurate poll?
Larger samples reduce sampling error, but with diminishing returns. Doubling a sample from 1,000 to 2,000 cuts the margin of sampling error by only about one percentage point.[17]
Does an online opt-in poll have a margin of error?
AAPOR says the margin of sampling error applies only to probability-based surveys, not to opt-in online polls. Some non-probability polls report a "credibility interval" instead, which AAPOR says rests on assumptions that may be hard to validate.[22][23]
Sources
· 24 claims from 4 publishers
Each numbered claim is a statement we checked against the sources listed with it. Tap a number in the text to see its sources in place.
- [1]
Pew Research Center explains that because a survey talks only to a sample, its result will probably not exactly match the true population value, and the margin of sampling error describes how close the result can reasonably be expected to fall to that value. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 1 (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- [2]
A margin of error of plus or minus 3 percentage points at the 95% confidence level means that if the same survey were fielded 100 times, the result would be expected to fall within 3 points of the true population value in 95 of them. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 1 (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- [3]
To judge whether one candidate's lead is real, the relevant margin of error is the one for the difference between the two candidates, which is generally about twice the margin reported for each candidate. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 2 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 2 (retrieved 2026-11-14)
- [4]
Subgroups have larger margins of error than the full sample; Pew's example is a group of about 160 people within a 1,067-person sample, giving plus or minus 8 points for a single estimate and plus or minus 16 points for the gap between two candidates. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 4 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 4 (retrieved 2026-11-14)
- [5]
Sampling error is only one source of survey error; question wording, who is reached, how candid respondents are and likely-voter models can also introduce error, and unlike sampling error these are hard to quantify and rarely reported. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 5 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 5 (retrieved 2026-11-14)
- [6]
AAPOR notes that at the usual 95% confidence level, about five times in 100 the survey answer will fall outside the margin of sampling error, one reason to interpret even the best surveys cautiously. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- [7]
AAPOR's example is a statewide survey finding 58% approval of the governor with a margin of plus or minus 3 points, which means the true value would be expected to lie between 55% and 61%. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- [8]
AAPOR says the margin of sampling error is always expressed in percentage points, not as a percentage. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, second paragraph (retrieved 2026-11-14)
- [9]
The margin of sampling error varies with the percentage being estimated; it is largest at 50%, and that is the figure pollsters typically report. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, first paragraph (retrieved 2026-11-14)
- [10]
The US Census Bureau describes a confidence interval as a range of values that describes the uncertainty around an estimate, and notes it is itself an estimate. confirmedas of 2026-11-14
- SAIPE: Confidence Intervals · US Census Bureau · Primary source · Confidence intervals (retrieved 2026-11-14)
- SAIPE: Confidence Intervals · US Census Bureau · Primary source · Confidence intervals (retrieved 2026-11-14)
- [11]
The Census Bureau routinely uses 90% confidence intervals, and a higher confidence level produces a wider interval for the same estimate. confirmedas of 2026-11-14
- SAIPE: Confidence Intervals · US Census Bureau · Primary source · Confidence intervals (retrieved 2026-11-14)
- SAIPE: Confidence Intervals · US Census Bureau · Primary source · Confidence intervals (retrieved 2026-11-14)
- [12]
The Australian Bureau of Statistics defines the standard error as a measure of how far an estimate based on a sample may vary from the true population value, and notes that for a normal distribution the 95% confidence interval spans two standard errors either side of the estimate. confirmedas of 2026-11-14
- Measures of error · Australian Bureau of Statistics · Primary source · Standard error (retrieved 2026-11-14)
- Measures of error · Australian Bureau of Statistics · Primary source · Confidence intervals (retrieved 2026-11-14)
- [13]
In Pew's example, a poll with a 3-point margin per candidate shows a 5-point lead, but the margin for the difference is about 6 points, so the true lead could plausibly be anywhere from -1 to +11 points. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 2 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 2 (retrieved 2026-11-14)
- [14]
The margin of error for a change between two polls can be larger than expected; in Pew's example a 3-point shift in a lead between two polls carries a margin of plus or minus 8 points. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 3 (retrieved 2026-11-14)
- [15]
Pew notes that of the 590 national polls listed on the 2012 Obama-Romney race, about 5% (around 30) would be expected to differ from the true value by more than their margin of error, and advises waiting to see whether a surprising result is replicated. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 3 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 3 (retrieved 2026-11-14)
- [16]
Pew says trends and patterns across a number of polls provide more confidence than one or two polls. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 3 (retrieved 2026-11-14)
- [17]
Larger samples have smaller sampling error, but with diminishing returns; doubling a sample from 1,000 to 2,000 reduces the margin of sampling error by only about one percentage point. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · Sample Size and MOSE (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · Sample Size and MOSE (retrieved 2026-11-14)
- [18]
A simple random sample of 1,067 people has a margin of error of plus or minus 3 percentage points for estimates of overall support for a candidate. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 4 (retrieved 2026-11-14)
- [19]
AAPOR's example is a 1,000-person survey with an overall margin of plus or minus 3 points, in which results for a 200-person subgroup carry a margin of plus or minus 6.9 points. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · Sample Size and MOSE (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · Sample Size and MOSE (retrieved 2026-11-14)
- [20]
Weighting helps avoid biased results but also makes the margin of error larger; statisticians call this increase in variability the design effect, and a margin of error that ignores it claims more precision than the survey warrants. confirmedas of 2026-11-14
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 5 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 5 (retrieved 2026-11-14)
- 5 key things to know about the margin of error in election polls · Pew Research Center · Primary source · 2016-09-08 · section 5 (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · Sample Size and MOSE (retrieved 2026-11-14)
- [21]
AAPOR's disclosure standards ask that probability surveys report their sampling error and say whether it has been adjusted for the design effect of weighting, clustering or other factors. confirmedas of 2026-11-14
- Disclosure Standards · American Association for Public Opinion Research · Primary source · Section III, A, sample sizes (retrieved 2026-11-14)
- [22]
AAPOR says the margin of sampling error applies only to probability-based surveys, in which participants have a known and non-zero chance of selection, and not to opt-in online surveys or other non-probability polls. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, second paragraph (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · page 1, second paragraph (retrieved 2026-11-14)
- [23]
Some non-probability polls report a "credibility interval" in similar plus-or-minus language, but AAPOR says it relies on assumptions that may be hard to validate and urges caution. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · The Credibility Interval (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · The Credibility Interval (retrieved 2026-11-14)
- [24]
AAPOR says there is no measurable overall margin of error for a poll, because other errors such as question design and interviewing cannot be measured. confirmedas of 2026-11-14
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · final note (retrieved 2026-11-14)
- Margin of Sampling Error/Credibility Interval · American Association for Public Opinion Research · Primary source · final note (retrieved 2026-11-14)
Revision history (1)
- Page created.
Created Nov 14, 2026. Last reviewed by an editor on Nov 14, 2026. Next scheduled review: Nov 14, 2027.
Cite this page
"Margin of error: what the plus-or-minus in a poll means." ContentLora, updated Nov 14, 2026. https://contentlora.com/explain/margin-of-error-explained
Spotted an error? Suggest a correction or emailcorrections@contentlora.com.