Margin of Error Calculator

What Does "± 3%" Actually Mean?

A news anchor reports that 52% of likely voters support a ballot measure, with a margin of error of ± 3%. The race looks close. But what does that ± 3% really tell you? It does not mean the true support is definitely between 49% and 55%. It means that if you ran the same poll many times under identical conditions, about 95% of the resulting intervals would contain the true value. The real support could be 50%. It could be 54%. It is unlikely to be 60%. That gap between the headline number and the underlying uncertainty is what a margin of error captures, and this calculator shows you exactly how it is built.

What This Calculator Does and Why It Matters

This tool computes the margin of error for a survey or experiment. It handles two common cases. The first is a proportion, like the percentage of respondents who favor a candidate. The second is a mean, like the average weight of parts coming off an assembly line. You pick a confidence level, enter your sample data, and the calculator returns the margin of error along with the standard error, the z-score used, and the full confidence interval.

The margin of error is half the width of a confidence interval. It tells you how far your sample estimate might sit from the true population value, purely because you did not survey everyone. If you want the full interval rather than just the error band, our Confidence Interval Calculator covers that directly. And if you are planning a study and want to know how many people to survey, the Sample Size Calculator works the problem from the other direction.

• Proportion mode: inputs p and n; outputs the margin of error as a percentage, the confidence interval, the standard error, and the z-score.
• Mean mode: inputs the sample mean, standard deviation, and n; outputs the margin of error, the confidence interval, the standard error, and the z-score.

How the Calculation Works

The margin of error is the critical value times the standard error. The critical value comes from the confidence level you choose. The standard error measures how much your sample estimate bounces around from sample to sample.

For a proportion, the formula is:

For a mean, the formula is:

The z-score is tied to the confidence level. At 95% confidence, z is 1.960. At 99%, it is 2.576. A higher confidence level widens the margin of error because you are demanding more certainty. The standard error shrinks as the sample size grows, which is why bigger samples produce tighter intervals. Notice the square root in the denominator. Doubling the sample size does not halve the margin of error. It reduces it by a factor of about 1.41. To cut the margin of error in half, you need roughly four times the sample size. That fact drives most survey budget decisions.

For proportions, the term p · (1 - p) is largest when p is 0.5. That is why pollsters use 0.5 as a conservative default when planning a sample size before they have data. It gives the widest, safest margin of error estimate. For a deeper look at the underlying statistics, see the Statistics Calculator.

How to Use the Calculator

1. Choose Proportion or Mean at the top, depending on what you are estimating.
2. Pick a confidence level from the buttons. 95% is the standard for most survey work.
3. For Proportion mode, enter the sample proportion (0 to 1) and the sample size.
4. For Mean mode, enter the sample mean, the standard deviation, and the sample size.
5. Read the margin of error, the confidence interval, the standard error, and the step-by-step breakdown on the right.
6. Use the Copy Result button to save the output for a report or presentation.

Example Calculations

Example 1: A political poll. A survey of 1,000 likely voters finds 52% support a measure. The pollster uses 95% confidence, so z = 1.960. The standard error is √(0.52 · 0.48 / 1000) = √(0.0002496) = 0.01580. The margin of error is 1.960 · 0.01580 = 0.03097, or about 3.10%. The confidence interval runs from 48.90% to 55.10%. Because that interval includes 50%, the race is statistically too close to call. This is the exact scenario the calculator reproduces when you enter p = 0.52 and n = 1000.

Example 2: A customer satisfaction survey. A company surveys 400 customers and measures their satisfaction score on a 100-point scale. The sample mean is 72 with a standard deviation of 14. At 95% confidence, the standard error is 14 / √400 = 14 / 20 = 0.70. The margin of error is 1.960 · 0.70 = 1.372. The confidence interval is 72 ± 1.37, or 70.63 to 73.37. The company can report that average satisfaction is 72, give or take about 1.4 points. Switch to Mean mode and enter 72, 14, and 400 to verify these numbers.

Real-World Scenarios

Election polling. Polling organizations report a margin of error with every survey. A typical national poll surveys about 1,000 people, which yields a margin of error near 3.1% at 95% confidence. When two candidates are within that band of each other, news organizations label the race a "statistical tie." The margin of error explains why a 2-point lead in one poll can become a 1-point deficit in another without either poll being wrong.

Market research. A brand tracks the share of customers who would recommend its product. With a sample of 600, the margin of error at 95% confidence is about 4 percentage points when the share is near 50%. If the tracked share moves from 42% to 45% quarter over quarter, that change sits inside the margin of error. A careful analyst will not call it growth until the sample size rises or the shift persists across multiple waves.

Quality control sampling. A factory pulls a sample of 50 parts each shift and measures a critical dimension. The margin of error on the sample mean tells engineers how much the sample average might differ from the true process average. If the tolerance band is narrower than the margin of error, the sample is too small to detect meaningful shifts, and the engineer needs to increase n.

Common Mistakes to Avoid

• Confusing margin of error with confidence level. The confidence level is how often the method produces an interval that captures the true value. The margin of error is the width of that interval. Raising the confidence level from 95% to 99% widens the margin of error, it does not shrink it.
• Ignoring sample size. A margin of error of 5% from a sample of 400 is very different from 5% from a sample of 40. The formula uses the square root of n, so small samples are far less stable than they look. Always report n alongside the margin of error.
• Using the margin of error for biased samples. The formula assumes a random sample. If your sample overrepresents one group, the margin of error does not fix that. A biased sample of 10,000 people can be further from the truth than a random sample of 500.
• Forgetting that the margin of error only covers sampling error. It does not account for non-response, question wording, or interviewer effects. Total survey error is usually larger than the reported margin of error, sometimes much larger.

Limitations

This calculator assumes a simple random sample from the population. It uses the normal approximation, which works well for proportions when n · p and n · (1 - p) are both at least 10. For very small samples or extreme proportions near 0 or 1, exact methods based on the binomial distribution are more accurate. For means, the calculator uses the z-distribution. When the sample size is below 30 and the population standard deviation is unknown, a t-distribution is technically more appropriate, and you can handle that case in the Confidence Interval Calculator. The margin of error reported here reflects only sampling error. It does not capture bias from non-response, self-selection, or poorly framed questions. Real-world survey error is often dominated by these other sources.

Authoritative Research and Resources

For a plain-language explanation of how pollsters compute and report margins of error, the Pew Research Center methods pages are an excellent starting point. They cover sampling, weighting, and the distinction between sampling error and total survey error.

The American Association for Public Opinion Research (AAPOR) publishes standards for survey reporting. The AAPOR standards and ethics resources explain what a margin of error should and should not be used to claim, including the limits of the concept for non-probability samples.

For a textbook treatment of the underlying theory, Penn State's STAT 500 course materials walk through confidence intervals, standard errors, and the role of sample size with worked examples. It is a solid free reference for anyone who wants the math behind the calculator.