Premium
ASYMPTOTIC ABSOLUTE EFFICIENCY OF PROJECTION TESTS WITH AN APPLICATION TO N‐RANKING TESTS
Author(s) -
Rothe Günter
Publication year - 1984
Publication title -
statistica neerlandica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.52
H-Index - 39
eISSN - 1467-9574
pISSN - 0039-0402
DOI - 10.1111/j.1467-9574.1984.tb01108.x
Subject(s) - minimax , mathematics , nonparametric statistics , ranking (information retrieval) , goodness of fit , statistical hypothesis testing , class (philosophy) , type (biology) , projection (relational algebra) , statistics , mathematical optimization , algorithm , computer science , artificial intelligence , ecology , biology
A class of χ 2 ‐type goodness‐of‐fit tests for the hypotheses of equal cell probabilities is introduced and the quality of the tests under arbitrary model restrictions is compared, where a suitable efficiency concept is used. As a special case, several nonparametric tests on treatment effects in the randomized block design can be considered. Some well known cases, such as the Anderson‐, Friedman‐ and the Page‐test, turn out to possess a certain minimax efficiency property under reasonable model assumptions. Finally a direct comparison between the Anderson‐ and Friedman‐tests is given.