TESTING THE POPULATION INVERSE-COEFFICIENTS OF VARIATION AND ITS APPLICATION

Authors : Nihan POTAS, Hamza GAMGAM

Pages : 503-514

Doi:10.18038/estubtda.574511

View : 21 | Download : 15

Publication Date : 2019-12-30

Article Type : Research Paper

Abstract :In this paper, some tests are introduced and compared for testing the equality of inverse-coefficients of variation. Monte-Carlo simulation method is used for comparisons. In this simulation study, various simulation scenarios were designed with different population numbers insert ignore into journalissuearticles values( k =  3, 6);,  sample sizes, parameter values and  type I error rates insert ignore into journalissuearticles values( alpha =  0.01, 0.05);. The tests were compared in terms of type I error rate and power in these scenarios. When the sample sizes are small, the D and WT tests showed good results in terms of type I error, but the LR and ST tests did not give good results. As the sample sizes increased, the experimental type I error rates of the LR and ST tests converged to the nominal type I error and all tests showed good results in general. While the sample sizes were equal, it was found that the LR test was the most powerful test and the ST test sometimes yielded good results. For these sample sizes, the D test yielded the worst results. When the sample sizes are different, the LR and D tests are powerful than the other tests, and the ST test is the worst test in terms of power. As expected, as the sample sizes and nominal type I error rate increased, the powers of the tests also increased. In addition, an application for the tests was made on real data. It was seen that the results of this application and simulation study coincide.
Keywords : Inverse Coefficient of Variation, Monte Carlo Simulation, Likelihood Ratio Test, Score Test, Wald Test