Eger, Karl-Heinz : Likelihood ratio tests for grouped observations

Author(s):

Eger, Karl-Heinz

Title:

Likelihood ratio tests for grouped observations

Preprint series:

Technische Universität Chemnitz, Fakultät für Mathematik (Germany). Preprint 10, 2003

Mathematics Subject Classification:

62F03			[ Hypothesis testing ]
62L10			[ Sequential analysis ]
62B99			[ None of the above, but in this section ]

Abstract:

The paper deals with likelihood ratio tests for discriminating between two simple hypotheses based on grouped observations. The computation of the characteristics of these tests is considered. Corresponding efficiency criteria for grouped tests are derived based on statistical an information theoretical aspects. Numerical examples are presented for the normal and exponential distribution. They show that only a small number of groups is required to obtain tests having approximately the same statistical properties as in the non-grouped case if the group bounds are chosen properly. For sequential likelihood ratio tests based on grouped observations a direct method is presented for the computation of the OC- and ASN-function.

Keywords:

Hypotheses testing; likelihood ratio tests; grouped observations; grouped data; optimal grouping; sequential analysis; sequential likelihood ratio tests; OC-function; ASN-function

Language:

English

Publication time:

11 / 2003