Wilson Score Interval
What you'll learn: How the Wilson score method provides a more reliable confidence interval for proportions, especially when your sample is small or your proportion is near 0 or 1.
Why We Need an Alternative
You've learned the standard confidence interval for proportions: p̂ ± z*√(p̂(1-p̂)/n). This works well most of the time, but it has a problem. When your sample size is small (say, n < 30) or when your proportion is extreme (very close to 0% or 100%), this formula can give you confidence intervals that extend below 0 or above 1—which makes no sense for a proportion!
What Makes Wilson Score Different
The Wilson score interval "adjusts" both the center point and the standard error of your interval. Instead of centering the interval directly on your observed proportion p̂, it pulls the center slightly toward 0.5 (the middle). This correction becomes more pronounced when you have less data or extreme proportions.
Think of it like adding "virtual observations" that gently nudge your estimate away from the extremes, making your interval more conservative and realistic.
When to Use It
Use the Wilson score interval when:
- Your sample size is small (typically n < 30)
- Your proportion is very small (< 10%) or very large (> 90%)
- You want more accurate coverage (the true proportion falls within your interval the stated percentage of times)
The standard formula works fine for large samples with moderate proportions, but Wilson score is safer in edge cases and never produces impossible intervals.
Key Takeaway: The Wilson score interval corrects the standard proportion confidence interval by adjusting both center and width, preventing impossible values and providing better accuracy for small samples or extreme proportions.