Estimate how many data quality checks to inspect when measuring a proportion. Choose a confidence level, margin of error, expected issue rate, usable-sample rate, and optional finite population. The result includes the completed sample target, the number to draw, and the effect of expected exclusions.
The completed target is rounded up, then divided by the expected usable-sample rate to obtain the draw size.
The draw size is a planning target for estimating one proportion under random sampling assumptions. It does not repair biased selection or inconsistent inspection rules.
The calculator substitutes 50% when the expected rate is entered as exactly 0% or 100%, avoiding an unrealistic zero sample.
At 95% confidence, ±3% margin, 10% expected issue rate, 95% usability, and a population of 100,000, the target is about 383 usable observations, so draw 404 records.
Rates near 50% require the largest sample because a proportion has its greatest variance at 0.5.
When the population is very large, the correction is negligible. For smaller known populations, it reduces the required sample.
Yes. The calculator rounds up because a fractional observation cannot satisfy the target margin of error.
Yes. The formula assumes observations are selected independently and representatively; convenience samples can remain biased regardless of size.
Yes, through the expected usable-sample rate. The requested draw is increased to compensate for anticipated exclusions.
| Variable | Meaning |
|---|---|
| z | Confidence-level critical value |
| p | Expected issue proportion |
| E | Desired margin of error |
| N | Finite population, when known |