What is Margin of error?
Margin of error is a number that summarises the sampling uncertainty in a survey result — it shows how much the reported percentage (or average) might differ from the true value in the whole population. It is usually reported alongside a confidence level (commonly 95%).
The margin of error quantifies the expected sampling variability in a survey estimate. For example, if a survey says 60% of respondents prefer X with a 5% margin of error (at 95% confidence), the true population proportion is likely to lie between 55% and 65%. The margin of error depends mainly on sample size (larger samples → smaller margin), the variability of the measure (proportions near 50% give the largest margin), and the chosen confidence level (higher confidence → larger margin). It applies to sampling error from selecting a subset of people; it does not cover other mistakes such as biased sampling, low response rates, measurement or translation errors. For subgroup comparisons (for example different language groups), each subgroup has its own margin of error determined by its sample size.
Usage example
A school survey of 400 parents reports that 60% would attend an evening meeting. With a 95% confidence level the margin of error is about ±4.9%, so the true attendance rate for the whole parent population is likely between ~55.1% and ~64.9%. If responses from one language group are only 50 people, that subgroup's margin of error is about ±13.9%, so small differences between language groups may not be meaningful.
Practical application
Margin of error helps you decide whether observed differences are real or likely due to chance, plan how many responses you need, and allocate resources (for example, push translation and outreach for smaller language groups to reduce their margin). For Hearo users it also warns against over-interpreting small differences between translated or language-specific subgroups and reminds you to combine good sampling practice with translation-quality checks — sampling uncertainty is only one source of error.
FAQ
Does margin of error tell me everything about survey accuracy?
No. Margin of error only addresses sampling variability. It does not account for biases from non-representative samples, low response rates, question wording, translation mistakes or data-entry errors. All those factors can change results in ways the margin of error doesn't capture.
How can I get a smaller margin of error?
The most direct way is to increase the number of completed responses — margin of error shrinks roughly with the square root of sample size (four times the responses halves the margin). You can also oversample small language groups, accept a lower confidence level, or reduce variability by measuring more precise outcomes, but increasing real responses is usually the most reliable approach.
What does '95% confidence' mean in plain language?
It means that if you repeated the same survey many times with the same methodology, about 95% of those surveys' confidence intervals would contain the true population value. It does not mean a 95% chance that the single reported interval contains the truth — it refers to the long-run frequency.
Does margin of error work the same for language subgroups?
Yes in principle, but each subgroup has its own margin of error determined by its sample size. Small language groups often have large margins of error, so treat comparisons between languages cautiously and consider targeted outreach to increase responses in underrepresented languages.