Conceptual

Statistics Sample Size and Margin of Error in Proportions

The relationship between sample size and margin of error in statistical inference for proportions is governed by a power-law scaling where precision improves inversely with the square root of $n$. Formally, under conditions satisfying normal approximation ($np \ge 10$ and $n(1-p) \ge 10$), the width of the confidence interval scales proportionally to $\frac{z_{\alpha/2} \sqrt{\hat{p}(1-\hat{p})}}{\sqrt{n}}$. This principle establishes that reducing uncertainty by a factor of two requires quadrupling the sample size, reflecting fundamental limits in binomial estimation within frequentist statistics.