On Uses of Van der Waerden Test: A Graphical Approach
share
Although several nonparametric tests are available for testing population identical distributions or equal means in multiple groups problem, the Van der Waerden test has asymptotically the same efficiency as the classical one-way analysis of variance test under normality assumption where it depends on inverse normal score. In this study the Waerden test statistic is used to create a form that assists in deriving adjusted p values chart in terms of Bonferroni and Benjamini-Hochberg methods. The sampling distribution of the proposed form is derived as a gamma distribution that utilized to acquire initial p values. The proposed method will provide well recognition and deep knowledge where the changes occur. Simulation study is executed to study the characteristic of the proposed method in terms of size and power of the test and compared with original Waerden test and one-way analysis of variance test. The results are of great practical values since the proposed method stretches visible shape to decision maker to gives complete picture and deep insight where the differences happen, keep size and power of the test almost similar to original Waerden test and gives a glimpse of various information that completes the pairwise comparisons. To show the benefits of the proposed method, two applications are presented.
READ FULL TEXT
