LARGE-SAMPLE INFERENCE FOR THE GENERAL HALF-NORMAL DISTRIBUTION

ABSTRACT

The paper considers the development of inferential techniques for the general half-normal distribution based on the maximum likelihood method. Point estimates and large-sample confidence sets are derived for the distribution's parameters, and the use of the confidence sets as a basis for hypothesis testing is discussed. Data on the percentage body fat of elite male athletes are used to illustrate the application of the new methods.

Keywords:

• Confidence sets
• General half-normal distribution
• Large-sample inference
• Maximum likelihood estimation
• Non-regular problem

ACKNOWLEDGMENTS

I would like to thank Toby Lewis, Chris Jones and, especially, Leon Harter, for their support and advice during the research which led to the production of this paper. I am also greatly indebted to an anonymous referee whose thoughtful comments on an earlier draft led to major improvements in the paper's content.