Confidence Intervals for the Difference Between Two Proportions
You'll learn: How to build and interpret confidence intervals that estimate how much two population proportions differ from each other.
The Core Idea
You've already learned to build confidence intervals for a single proportion (like estimating the true conversion rate of one website). Now imagine you have two groups—say, customers shown ad version A versus ad version B—and you want to know: how much do their conversion rates truly differ?
Just like with single proportions, we use sample data to estimate the difference between two population proportions. The confidence interval gives us a plausible range for that true difference.
The Formula
The confidence interval for the difference (p₁ - p₂) follows this pattern:
(p̂₁ - p̂₂) ± z × SE*
Where:
- p̂₁ and p̂₂ are your two sample proportions
- z* is the critical value (typically 1.96 for 95% confidence)
- SE is the standard error for the difference, calculated as:
SE = √[(p̂₁(1-p̂₁)/n₁) + (p̂₂(1-p̂₂)/n₂)]
Notice the formula combines both sample proportions' standard errors using the square root of their sum—this accounts for variability in both groups.
Interpretation
If your 95% CI for (p₁ - p₂) is [0.02, 0.08], you'd say: "We're 95% confident the true difference in proportions falls between 2% and 8%." Since this interval doesn't contain zero, there's likely a real difference between the groups.
Key Takeaway: Confidence intervals for the difference between two proportions let you quantify how much two groups differ, with a margin of error, helping you make evidence-based comparisons like A/B test results or demographic disparities.