Sample size calculator
How many completed interviews do you need for the precision you want?
385 completed interviews
To reach that many completes you would need to invite about 1,540 people at the completion share above.
How to use it
Choose how sure you want to be, how much error you can accept and, if you know it, the size of the group you are describing. The calculator returns the number of completed interviews and the number of people you would need to invite at the completion share you expect.
If your sample is weighted or comes from a panel, raise the design effect so the answer holds up. See our standards for how we report panel results, or the margin of error calculator to check a sample you already have.
Planning a real study?
Use the scoping tool to see what makes a study a bigger job, or see how we run pulse studies, trackers and full studies.
Frequently asked questions
What does the calculator assume?
A simple random sample of a yes or no question, with the usual normal approximation. The design effect lets you widen the answer for weighting and for panels that are not random samples.
Why use a higher design effect?
Weighting and clustering make a result less precise than a simple random sample of the same size. A weighted sample usually needs a higher design effect, so plan for a wider margin than the textbook one.
Does a bigger sample always mean a better result?
A bigger sample shrinks random error. Who is in the sample and how it is weighted matter just as much, which is why we set quotas, weight to national figures and publish the method with every result.
What if I only want to compare small groups?
Each group needs enough interviews of its own. Run the calculator for the smallest group you want to report separately, not for the whole sample.