مقارنة بين اختبارات مشكلة عدم تجانس تباين الخطأ في أنموذج الانحدار الخطي المتعدد في حالة ابتعاد البيانات عن التوزيع الطبيعي

Abstract EN

the most of the statistical tests for testing the hypothesis of Heteroscedasticity need specific conditions and hypotheses form condition normal distribution data if this condition is broken .

the use statistical tests will not be right because it gives inaccurate results.

So that circular will be through the study of four types of tests((EP, G — F)and(, 'F) test)in the linear regression.in order to achieve the comparison among these statistics in tow case The first case is when the data follow the normal distribution and The second case the data follow the Cauchy distribution .the simulation style was employed using (Monte Carlo) method to test of the homoscedasticity .

through counting the possibility of type I error and the power of test for these statistics .

We have found that (•#*, M') statistics in case (normal and Cauchy) distribution of The observations used in testing hypothesis in the multiple linear regression model model is the best tin both cases.