On the one-sample location hypothesis for mixed bivariate data

Abstract

We study a hypothesis testing problem involving the location model suggested by Olkin and Tate (1961). Specifically, we derive a likelihood ratio lest of the associated location hypothesis as an alternative to the conventional method of carrying out separate tests for each of the parameters. A small sample Monte Carlo comparison indicates the general superiority of the former in terms of statistical power. We also comment briefly on the properties of the test.

Keywords:

• consistency
• likelihood ratio test
• location model
• power function

Additional information

Notes on contributors

A. R. De Leon