138 – Weather and Climate Applications
Generalized P-Value Test to Compare Several Population Means for Unequal Variances
Evren Özkip,
Bursa Police College
Ahmet Sezer
Anadolu University
Berna Yazici
Anadolu University
The size of classical F tests are fairly robust against the assumption of equal variances when the sample sizes are equal. However when the sample sizes are different, failure of the assumption of equal variances can have serious problems on the power of the F test. Classical F test fails to reject the null hypothesis even when the data actually providens strong evidence to do so. The generalized p-value method provides a promising approach to solve such problems with no adverse effect on the size of the test. In this study we demostrate performance of the F test and generalized p value approach for different sample sizes and variances. Weather data will be used to compare temperature for different locations under some unequal variances situations.
