A statistic obtained by arranging a sample of observations in nondecreasing order and selecting the k-th element; the k-th order statistic generalizes minima, maxima and sample quantiles.

Principle

Ordering a finite sample induces a sequence of dependent random variables whose distributions reflect sample size and parent distribution; order statistics summarize distributional position without parametric assumptions.

Demonstration

For a sample of size n, the sample median is the order statistic with k = (n+1)/2 (for odd n) or any value between the n/2 and n/2+1 order statistics (for even n); the minimum and maximum are first and nth order statistics respectively.

Misapplication

Treating an order statistic as if it equals the corresponding population quantile without accounting for sampling variability or finite-sample bias leads to overconfident inferences.

Consequence

Order statistics underpin nonparametric interval estimation, robust estimators like trimmed means, and extreme-value analysis; their sampling laws enable confidence statements and bias corrections.

Reversal

The unordered sample (multiset) retains the same values but lacks positional information; without ordering, one cannot define minima, maxima or medians as single-sample statistics.

Boundary

Defined only for samples drawn from a totally ordered set; does not apply directly to vector-valued observations without a chosen ordering or to data where ties require tie-breaking conventions.

Semantic Tension

Closely related to sample quantiles and empirical distribution functions; order statistics are the raw positional elements, while quantiles often interpolate between order statistics and carry additional convention-dependent definitions.

Synthesis

An order statistic is the k-th smallest observation in a sample: a simple, assumption-light summary that maps samples to distributional position and serves as the basis for robust and nonparametric inference.