A likelihood ratio test for equality of deviations from randomness |
| |
Authors: | D. L. Johnson and A. J. Petkau |
| |
Affiliation: | (1) Agriculture and Agri-Food Canada Research Centre, P.O. Box 3000, T1J 4B1 Lethbridge, AB, Canada;(2) Department of Statistics, The University of British Columbia, V6T 1Z2 Vancouver, BC, Canada |
| |
Abstract: | The use of the sample variance-to-mean ratio as a measure of deviation from randomness in spatial pattern is reviewed. The likelihood ratio method of constructing a statistical test for the equality of several population variance-to-mean ratios is described, and details are provided for the special case where counts are modelled as arising from a negative binomial distribution. This test is illustrated by application to example data sets in ecology. Likelihood ratio tests represent a general methodology whereby relationships among several indices of aggregation can be systematically investigated, provided one is able to specify a suitable parametric form for the underlying distributions. |
| |
Keywords: | variance-to-mean ratios deviations from randomness spatial pattern and counts Poisson negative binomial |
本文献已被 SpringerLink 等数据库收录! |
|