Margin of Error

The margin of error is a statistic expressing the amount of random sampling error in a survey's results. It is the half width of a confidence interval: the sample estimate plus or minus the margin of error gives a range built by a method that captures the true population value in a specified share of repeated samples, the confidence level (usually 95%), and it is typically reported as a plus or minus value. A smaller margin of error indicates greater precision and typically requires a larger sample.

Survey results are estimates of population parameters, not exact measurements, and the margin of error communicates the precision of those estimates to both researchers and the public. When a poll reports that 52% of voters support a candidate with a margin of error of plus or minus 3 percentage points, readers can see that the 95% confidence interval runs from 49% to 55%, so the poll cannot rule out support anywhere in that range, including below 50%. Without the margin of error, survey results can be over interpreted, and small differences between groups or time points may be mistaken for meaningful changes. Understanding the margin of error is essential for designing surveys with adequate precision, interpreting survey results responsibly, and communicating findings to nonspecialist audiences.

The margin of error for a proportion is calculated as the critical value (z) multiplied by the standard error of the proportion. For a 95% confidence level, the critical value is approximately 1.96. The standard error of a proportion is the square root of p(1 minus p) divided by n, where p is the sample proportion and n is the sample size. The margin of error therefore depends on three factors: the confidence level, the sample proportion (maximum uncertainty occurs at p = 0.50), and the sample size.

For example, a survey of 1,000 respondents with a proportion of 0.50 and a 95% confidence level produces a margin of error of approximately 1.96 times the square root of (0.50 times 0.50 divided by 1000), which equals about 0.031 or plus or minus 3.1 percentage points.

The margin of error decreases as the sample size increases, but the relationship is not linear: quadrupling the sample size cuts the margin of error in half. This diminishing return means that very large samples are needed to achieve very small margins of error. The margin of error also assumes simple random sampling; more complex sampling designs (such as cluster or stratified sampling) require design effect adjustments that may increase or decrease the effective margin of error.

For continuous variables, the margin of error is likewise the half width of the confidence interval for the mean, and it depends on the sample standard deviation rather than the proportion.

Report the margin of error alongside all survey estimates to communicate precision. Use it during survey design to determine the sample size needed to achieve a desired level of precision for the primary outcome. It is most commonly associated with survey research and opinion polls but applies to any estimate derived from a probability sample. When comparing two estimates from independent samples (for example, two separate surveys at different time points, or two nonoverlapping subgroups), note that the margin of error for the difference is larger than the margin of error for either individual estimate.

A common error is interpreting the margin of error as accounting for all sources of error in a survey, when it actually reflects only sampling error. Nonsampling errors such as question wording effects, nonresponse bias, coverage bias, and social desirability bias are not captured by the margin of error and can be much larger in practice. Another mistake is applying the overall margin of error to subgroup estimates; subgroups have smaller sample sizes and therefore larger margins of error. Some survey reports omit the margin of error or bury it in fine print, encouraging overinterpretation of point estimates. Finally, comparing two estimates whose confidence intervals overlap does not necessarily mean they are not significantly different; a formal statistical test is needed.

A national health survey of 4,000 adults estimates that 28% of respondents report food insecurity, with a margin of error of plus or minus 1.4 percentage points at the 95% confidence level. This means the 95% confidence interval for the true prevalence is 26.6% to 29.4%. A subgroup analysis of 800 respondents aged 18 to 30 estimates 35% food insecurity, but the margin of error for this subgroup is plus or minus 3.3 percentage points (95% CI: 31.7% to 38.3%) due to the smaller sample size.