Thesis Type: Postgraduate
Institution Of The Thesis: Gazi University, Fen Bilimleri Enstitüsü, İSTATİSTİK ANABİLİM DALI, Turkey
Approval Date: 2020
Thesis Language: Turkish
Student: NİLGÜN NURSU ÖZTÜRK
Supervisor: HAMZA GAMGAM
Open Archive Collection: AVESIS Open Access Collection
Abstract:Equality of variances is one of the basic assumptions of the classical variance analysis method. There are many known tests in the literature to determine whether this assumption is met. In this study, to test the homogeneity hypothesis of variances, the bootstrap test based on the ratio of average absolute deviations, Bartlett test, Hartley test, Levene test, Fligner-Killeen test were introduced and compared with the simulation method. In these comparisons, various simulation scenarios were created using design parameters such as sample number, variances and sample sizes. The tests in this simulation study were compared in terms of experimental type I error rate and the power of the test. The gamma, weibull, uniform and chi-square distributions were used for the nonnormal distributions in the simulation study. According to the simulation findings, the Bartlett and Hartley tests for the normal distribution gave the best results in terms of experimental type I error rate and power. The bootstrap test based on the ratio of the absolute deviations, Levene test and Fligner-Killeen test gave better results in terms of the experimental type I error rate for the nonnormal distributions. On the other hand, the power of the Bartlett, Hartley and Levene tests are generally better for these distributions.