Find the minimum sample size needed for a survey, or work out the margin of error for a sample you already have.
Computes the minimum number of samples needed to meet your desired confidence level and margin of error.
Computes the margin of error (confidence interval) for a survey or observation you've already run.
A larger sample gets you closer to the true population value, shrinking your margin of error — but with rapidly diminishing returns. Cutting the margin of error in half requires roughly quadrupling the sample size, since the relationship follows an inverse-square pattern.
If you don't know the true proportion you're estimating, use 50% — it's the most conservative assumption, since it maximizes the required sample size (p(1-p) is largest when p=0.5). If you have a rough estimate already (from prior research or a pilot study), using it gives a more efficient, smaller required sample.
The finite population correction factor shrinks toward 1 as the population grows large relative to the sample — once your population is, say, 20× your sample size or more, the correction becomes negligible, which is why very large or "unlimited" populations give essentially the same answer.
95% is the most common default across research and polling. Higher confidence levels (99%+) require larger samples for the same margin of error, since you're demanding more certainty.
Pick another tool to jump straight to it.