Asymptotic Theory of Sequential Fixed-Width ConfidenceIntervals for Location Parameters,
Abstract
Consider a sequence of confidence intervals (I sub n) with widths (2(D sub n)) and noncoverage probabilities (2(alpha sub n)). Previous sequential solutions to the problem of finding a short confidence interval with a high coverage probability have taken the following approach: Introduce a stopping rule nu(alpha,d) to be used with a sequence of intervals and designed to select an interval of width 2d which is such that, as d nears 0, the coverage probability 1 - (2(alpha sub n)) nears 1 - 2 alpha for preassigned alpha > 0. In the present investigation, the authors are concerned with the fixed-width (d fixed) case and are primarily interested in rules which perform well as alpha nears 0. A general approach toward solution of this problem in the case of a location parameter is presented and applied in connection with confidence intervals based on the sample mean and with confidence intervals based on the sample median. Asymptotic relative efficiencies are provided. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1974
- Accession Number
- AD0787552
Entities
People
- D. D. Wackerly
- Robert Serfling
Organizations
- Florida State University