Enter your sample size and confidence level to calculate your survey's margin of error.
Margin of error tells you how much your survey results might differ from the true opinion of your entire target population, simply because you surveyed a sample of people rather than everyone. It's expressed as a plus-or-minus percentage around your result.
For example, if 60% of respondents say they prefer Product A with a margin of error of ±4%, the real figure across your whole population most likely falls somewhere between 56% and 64%. The smaller the margin of error, the more precisely your sample reflects the full population.
These three terms are often confused, but they describe different pieces of the same statement:
Put together: "We are 95% confident (confidence level) that the true value falls between 56% and 64% (confidence interval), a margin of error of ±4%."
When you also know your total population size, a finite population correction is applied so the margin of error shrinks slightly for smaller populations — which is exactly what the calculator above does automatically.
| Confidence level | Z-score |
|---|---|
| 90% | 1.645 |
| 95% | 1.96 |
| 99% | 2.576 |
You survey 400 people from a large population at a 95% confidence level:
Larger samples produce a smaller margin of error, but with diminishing returns — doubling your sample from 400 to 800 respondents only tightens a ~4.9% margin of error to roughly ~3.5%, not half of it. This is why chasing extremely large samples usually isn't worth the extra time and cost.
A higher confidence level (say 99% instead of 95%) means you want to be more certain your interval captures the true value — which requires a wider range, and therefore a larger margin of error, for the same sample size.
Using p = 0.5 (a 50/50 split) is the standard, most conservative assumption because it produces the widest possible margin of error. If you already have good reason to expect a lopsided result (e.g. 80/20), your real-world margin of error will typically be smaller than the calculator's default estimate.
±5% at a 95% confidence level is the widely used industry standard for general audience research. Higher-stakes decisions (elections, product launches) often target ±3% or tighter.
Technically each question has its own margin of error based on how many people answered it and how their responses were distributed — so a question with more skipped answers will have a wider margin of error than the survey's headline number.
The finite population correction has almost no effect once your population is much larger than your sample (roughly 20x or more) — this is why national surveys with millions of people in the population use sample sizes similar to those for a population of just 100,000.