Bahadur efficiencies of the Epps--Pulley test for normality

Bruno Ebner,Norbert Henze

DOI: https://doi.org/10.48550/arXiv.2106.13962

2021-06-26

Abstract:The test for normality suggested by Epps and Pulley (1983) is a serious competitor to tests based on the empirical distribution function. In contrast to the latter procedures, it has been generalized to obtain a genuine affine invariant and universally consistent test for normality in any dimension. We obtain approximate Bahadur efficiencies for the test of Epps and Pulley, thus complementing recent results of Milošević et al. (2021). For certain values of a tuning parameter that is inherent in the Epps--Pulley test, this test outperforms each of its competitors considered in Milošević et al. (2021), over the whole range of six close alternatives to normality.

Statistics Theory

What problem does this paper attempt to address?

The paper attempts to address the problem of evaluating the Bahadur efficiency of the Epps-Pulley normality test and comparing it with other normality tests based on the empirical distribution function. Specifically, the authors aim to:

1. **Derive the Bahadur efficiency of the Epps-Pulley normality test**: Bahadur efficiency is a method of measuring the performance of a statistical test as the sample size tends to infinity. By calculating the Bahadur efficiency, the performance of the Epps-Pulley test under different parameter settings can be assessed.

2. **Supplement existing research**: Milošević et al. (see reference [15]) have studied the Bahadur efficiency of normality tests based on the empirical distribution function but did not address the Epps-Pulley test. This paper aims to fill this gap.

3. **Compare the performance of different test methods**: By calculating the Bahadur efficiency under different tuning values, the authors hope to demonstrate the performance of the Epps-Pulley test under various near-normal alternative hypotheses and compare it with existing classical test methods (such as the Shapiro-Wilk, Shapiro-Francia, and Anderson-Darling tests).

4. **Provide theoretical support**: Through theoretical analysis and numerical simulations, the effectiveness and superiority of the Epps-Pulley test in different situations are verified, thereby providing a theoretical basis for its use in practical applications.

In summary, the main goal of this paper is to comprehensively evaluate the performance of the Epps-Pulley normality test through the analysis of Bahadur efficiency and compare it with other commonly used test methods to demonstrate that it may perform better in certain situations.