If spatial data do not have an underlying normal distribution, the traditional tests for spatial correlation may lead to erroneous conclusions. This is illustrated by means of an example. A class of tests based on the ranks of the observations is proposed and the large sample normality of each test is established. A theorem providing the basis for locally most powerful tests is also presented.
Get full access to this article
View all access options for this article.
References
1.
BeranR J, 1972“Rank spectral processes and tests for serial dependence”Annals of Mathematical Statistics431749–1766
