Title

Sequential, Accelerated-Sequential And Three-Stage Estimation Of The Mean Of A First-Order Stationary Autoregressive Process: A Monte Carlo Study

Keywords

Coverage probability; Expected sample size; Fixed width confidence interval

Abstract

By way of Monte Carlo techniques, this paper compares the moderate sample size properties of three methods for fixed-width confidence interval estimation of the mean of a series whose errors form a first-order stationary autoregressive process. The three methods are sequential (Sriram, 1987). accelerated-sequential (Hall, 1983) and three-stage (Hall, 1981) estimation. Furthermore, four distributions on the errors are assumed - normal, double-exponential, uniform and "shifted" gamma. Our results show that the performance of the three estimation procedures is dependent on the autocorrelation and the ratio of interval width to the standard deviation of the errors. Furthermore, the performance of the sequential procedure relative to the accelerated-sequential and three-stage procedures also depends on the aforementioned ratio; with the accelerated-sequential and three-stage procedures behaving similarly for all ratios. Also, no one distribution clearly dominates any other with respect to performance of the three methods.

Publication Date

1-1-1996

Publication Title

Journal of Statistical Computation and Simulation

Volume

54

Issue

4

Number of Pages

333-353

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.1080/00949659608811738

Socpus ID

0039564564 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/0039564564

This document is currently not available here.

Share

COinS