Tests for Exponentiality When Origin and Scale Parameters are Unknown.
Abstract
Several methods are examined for testing for the exponential distribution, when both the scale and location parameters are unknown. Percentage points are given for EDF (Empirical Distribution Function) tests, in which both parameters are estimated from the data, and the tests are compared for power with EDF tests where only the scale parameter is unknown. In addition, power studies are reported to compare these tests with other goodness-of-fit techniques for the exponential distribution, particularly those based on regression methods, and it is found that the EDF statistics W sub 2 and A sub 2 have overall good power properties.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 02, 1983
- Accession Number
- ADA131377
Entities
People
- John Spinelli
- Michael A. Stephens
Organizations
- Stanford University