Asymptotic Power of EDF Statistics for Exponentiality against Weibull and Gamma Alternatives.
Abstract
Asymptotic distributions of EDF statistics - goodness-of-fit statistics based on the empirical distribution function - can be obtained by expanding a related function in a series of orthogonal functions. This method was given by Watson and developed especially in this connection by Durbin and Knott and by the author. an earlier approach by Anderson and Darling is different and the newer methods provide a sharper view of the approach to infinity. Similar approaches were employed by Durbin, Knott and Taylor to test for normality and also gave asymptotic power results for certain alternatives and in some cases for tests of exponentiality. In this paper, asymptotic powers of some EDF statistics for testing exponentiality against Weibull and Gamma alternatives are developed. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 04, 1981
- Accession Number
- ADA097107
Entities
People
- Michael A. Stephens
Organizations
- Stanford University