Sağlam Kestiricilere Dayalı Varyans Analizi

Abstract

Certain assumptions must be provided in the Analysis of Variance (ANOVA) calculated over sample means. If the assumptions are not met, the Analysis of Variance cannot fully reveal the real group differences, and in this case, a test with low power may emerge. Even a slight deviation from the assumptions will introduce these problems. One of the methods suggested as a solution to this situation is robust methods. In the thesis study, Type 1 errors of the proposed tests based on robust estimators were compared with the classical least squares estimators (LSE).