How to use it
- Pick proportion or mean.
- Enter the confidence level and the margin of error.
- For a proportion give the expected percentage (50 if unknown); for a mean give the standard deviation.
- Optionally enter the population size.
How it works
For a proportion n₀ = z² · p(1 − p) / e². For a mean n₀ = (z · σ / e)². With a population N the size is reduced to n = n₀ / (1 + (n₀ − 1) / N). The result is always rounded up to a whole number.
Worked example
95% confidence, margin ±5%, p = 50%: z = 1.959964, n₀ = 3.841459 × 0.25 / 0.0025 = 384.15, so 385 people. In a population of 1000 the correction gives 384.15 / 1.383 = 277.7, so 278. For a mean with σ = 15 and e = 2: (1.959964 × 15 / 2)² = 216.08, so 217.
Tips
- Halving the margin of error needs about four times the sample.
- Sample size does not repair a biased sampling method.