Thesis Type: Postgraduate
Institution Of The Thesis: Gazi University, Fen Bilimleri Enstitüsü, İSTATİSTİK ANABİLİM DALI, Turkey
Approval Date: 2019
Thesis Language: Turkish
Student: HASAN DEMİRCİ
Supervisor: HAMZA GAMGAM
Open Archive Collection: AVESIS Open Access Collection
Abstract:Many tests have been proposed to compare the ordered location parameters under the simple order. Most of these tests were developed under the assumptions of homogeneous variances and normality. However, situations where these two assumptions are not met are often encountered in most real-life problems. In cases where these assumptions are not met, there are also nonparametric multiple comparison tests in the literature. In this study, the multiple comparison methods are called the M, NPM, T, NPT, NPY and NPY tests were introduced. While the M and T tests were parametric methods, the NPM, NPT, NPY and NPY tests were nonparametric methods. By a simulation study, these tests were then compared for type I error rate and power. In this simulation study, the sample number was taken as 5, the samples sizes were 5, 7, 13, 25, and the various simulation scenarios were used under different median vectors. In the simulation study, the data generated from the Student-t, Chi-square and Gamma distributions were used. According to the simulation results, the nonparametric NPT test gave the best results in terms of type I error rate and power. In addition, the NPY test also gave the good results. The nonparametric NPT and NPY tests gave better results than parametric tests for unequal sample sizes in all distributions. These results support the findings in the literature. On the other hand, the parametric M test gave the worst results