Max Min Uniform Surprise

This question looks at the extremes of a sample of independent draws from a continuous uniform distribution on a bounded interval. The candidate is asked to characterize the distribution and expectation of both the maximum and the minimum of these draws, focusing on how the collective behavior of many simple random variables produces nontrivial distributions for their extremes. This is a canonical exercise in order statistics and shows up frequently in interviews for quant research and statistics-heavy roles, because it ties together basic distribution theory with reasoning about sample behavior.

Solving it leans on writing events involving maxima and minima in terms of the original variables and exploiting independence. The core step is expressing the relevant probabilities using the underlying cumulative distribution function, then differentiating to obtain densities. From there, the expectation can be computed either by integrating against the density or by using integration-by-parts style identities for nonnegative random variables. Interviewers are looking for comfort with distribution functions, facility with translating verbal "max/min" statements into set events, and clean, error-free calculus on simple integrals.

What it tests

When dealing with the maximum or minimum of independent and identically distributed (iid) random variables, the key is to realize that their joint behavior is governed by the product of their individual probabilities. For the maximum, the event that all variables are less than or equal to a value x is the intersection of independent events, so the probability multiplies: $P(Z_n \leq x) = [F_X(x)]^n$. For the minimum, the event that all variables are greater than x is also an intersection, so $P(Y_n > x) = [1 - F_X(x)]^n$, and thus $P(Y_n \leq x) = 1 - [1 - F_X(x)]^n$. This pattern arises because the maximum and minimum are determined by the most extreme outcomes among the variables, and independence allows us to multiply probabilities for joint events. The structure holds for any iid random variables, not just the uniform case, and is a direct consequence of how order statistics relate to the underlying distribution's CDF.