The normal approximation to the （正常的近似）.pdf

The normal approximation to the hypergeometric distribution Mark A. Pinsky, Northwestern University 1 Introduction In Feller [F], volume 1, 3d ed, p. 194, exercise 10, there is formulated a version of the local limit theorem which is applicable to the hypergeometric distribution, which governs sampling without replacement. In the simpler case of sampling with replacement, the classical DeMoivre-Laplace theorem is applicable. Feller’s conditions seem too stringent for applications and are difficult to prove. It is the purpose of this note to re-formulate and prove a suitable limit theorem with broad applicability to sampling from a finite population which is suitably large in comparison to the sample size. 2 Formulation, statement and proof We begin with rational numbers 0 p 1 and q = 1 − p . The population size is N and the sample size is n, so that n N and Np, Nq are both integers. The hypergeometric distribution is Nq Np P (k; n, N ) = k n−k 0 ≤ k ≤ n. (1) N n This is expressed in terms of the usual binomial distribution by writing Np (Np)k (Np)(Np − 1) · · · (Np − k + 1) = = k k! k! = p k N k 1 − 1 · · · 1 − k − 1 k! Np Np Nq = (Nq)n−k = (Nq)(Nq − 1) · · · (Nq − (n − k) + 1) n − k (n − k)! (n − k)! = qn−k N n−k 1 − 1 · · · 1 − n − k − 1 (n − k)! Nq Nq n N Nn N