Sample Size for Estimating a Proportion
You'll learn how to calculate the minimum sample size needed to estimate a population proportion with a desired level of precision.
The Core Idea
When planning a survey or study to estimate a proportion (like approval rating or conversion rate), you need enough data to achieve your desired margin of error. Too few samples and your confidence interval will be uncomfortably wide; too many wastes resources. The sample size formula lets you work backward from your precision goal.
The Formula
The sample size needed is:
n = (Z² × p × (1-p)) / E²
Where:
- Z = critical value for your confidence level (1.96 for 95%)
- p = estimated proportion
- E = desired margin of error
Two Approaches for p
Conservative Approach
When you have no prior information, use p = 0.5. This gives the largest possible sample size because variance is maximized when p = 0.5. It's like planning for the worst-case scenario—you'll have enough samples no matter what the true proportion is.
Pilot Estimate Approach
If you have a reasonable guess from past studies or a small pilot sample, use that value for p. For example, if historical data suggests 30% conversion rate, use p = 0.3. This typically requires fewer samples but relies on your estimate being reasonably accurate.
Example Logic
Say you want ±3% margin of error at 95% confidence. Using the conservative approach (p = 0.5):
n = (1.96² × 0.5 × 0.5) / 0.03² ≈ 1,068 samples
Key Takeaway: To determine sample size for proportion estimates, use p = 0.5 for maximum safety when planning, or use a pilot estimate if you have prior knowledge—both plug into the same formula with your desired margin of error and confidence level.