A New Goodness of Fit Test for the Uniform Distribution with Unspecified Parameters.
Abstract
Separate techniques of interpolation, reflection, and parameter estimation are combined to develop a new goodness of fit test for the uniform distribution. The Kolmogorov-Smirnov, Anderson-Darling, and Cramer-Von Mises statistics are used in the generation of critical value tables of sample sizes from 3 to 50. The methods for estimating parameters are the Maximum Likelihood and the Best Linear Unbiased Estimators. Separate tables for each are presented. These tables are built with and without employing the reflection technique. The reflection technique is one in which the data points are reflected about the sample mean to double the size of the sample set. With these tables of critical values, a power study is done to test the power of the three statistics with the reflection procedure versus the same statistics without the reflection procedure.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1982
- Accession Number
- ADA124835
Entities
People
- Larry B. Woodbury
Organizations
- Air Force Institute of Technology