On Berry-Esseen Rates for Statistical Functions, with Application to L-Estimates.
Abstract
A parameter expressed as a functional T(F) of a distribution function (d.f.) F may be estimated by the statistical function T(F sub n) based on the sample d.f. F sub n Typically, T(F sub n) is asymptotically normal. We investigate the rate of this convergence by utilizing the von Mises representation to express T(F sub n) - T (F) as an approximate U-statistic plus R sub n, and applying the Berry-Esseen rate 0(sq rt n) established for U-statistics by Callaert and Janssen. This essentially reduces the problem to a handling of R sub n. We carry out this method for linear functions of order statistics (L-estimates) and obtain results competitive with Bjerve and Helmers. Also, we briefly indicate the application of the method to M-estimates.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1979
- Accession Number
- ADA072131
Entities
People
- Dennis D. Boos
- Robert Serfling
Organizations
- Florida State University