Suppose you are analyzing a dataset using a Wald-type confidence interval for a proportion $p$. As the number of trials $n$ increases, the Wald interval is known to have poor coverage properties when the true $p$ is near $0$ or $1$. Which modification is mathematically equivalent to the Agresti-Coull interval, intended to improve coverage?