THE POWER OF SOME TESTS FOR UNIFORMITY OF A CIRCULAR DISTRIBUTION.

Abstract

Let (x sub 1, x sub 2,...,x sub n) be independent observations on an arbitrary random variable which takes values on a circle of unit circumference. Suppose f(x) is a probability density in L sub 2(0,1). The main result of this paper is the asymptotic distribution of the class of test statistics T sub n = (1/n) multiplied by the integral taken between the limits 0,1 of the quantity (summation from j = 1 to j = n of f(x + x sub j)-n) squared dx. T sub n is used to test whether the observations are uniformly distributed on the circle. It includes as special cases several other statistics previously proposed for this purpose by Rayleigh, Watson, Ajne and others. Two distinct cases arise in the asymptotics -- for one class of alternatives T sub n yields a consistent test for uniformity, but for the others T sub n gives a test which is not consistent. A fair approximation to the power of T sub n can be obtained from its first two moments. The approximate Bahadur slope of T sub n is calculated from its asymptotic null distribution it does not appear to reflect the power of T sub n reliably. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1968
Accession Number
AD0668150

Entities

People

  • R. J. W. Beran

Organizations

  • Johns Hopkins University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Data Science
  • Information Science
  • Integrals
  • Mathematics
  • Observation
  • Probability
  • Probability Distributions
  • Random Variables
  • Statistics
  • Stochastic Processes

Fields of Study

  • Mathematics

Readers

  • Analytical Mechanics
  • Statistical inference.