Sample size calculator
Precision costs non-linearly: halving the margin of error means quadrupling the sample. The tool shows how many people you need and applies the correction for small populations.
The resulting number assumes simple random sampling. Real surveys use strata and quotas, which change the effective error by a design factor usually between 1.2 and 2.
Frequently asked questions
Why is 50% the cautious choice?
Because the variance p(1−p) peaks at 0.5. With no prior estimate, 50% gives the largest, therefore safest, sample.
When does population size need correcting?
When the sample exceeds roughly 5% of the population. In a company of 300 staff, the correction cuts the required number substantially.
Why does response rate matter?
Because you have to invite more people than the responses you want. At a 30% response rate, 1,000 responses need over 3,300 invitations.