Far East Journal of Theoretical Statistics
Volume 28, Issue 1, Pages 57 - 106
(May 2009)
|
|
SOME NEW NORMALITY TESTS FOR THE ERROR OF A LINEAR REGRESSION MODEL
Miguel A. Arcones (U.S.A.) and Yishi Wang (U.S.A.)
|
Abstract: We present two new normality tests for the error of a linear regression model. The tests are obtained by applying the normality tests in Arcones and Wang [3] to the residuals obtained using the least squares estimators. We show that the considered tests are omnibus. We also obtain the limit distribution of the considered tests under the null hypothesis. Simulations show that the power of the presented tests is competitive with common normality tests. |
Keywords and phrases: test of normality, linear regression model, residuals, U-processes. |
|
Number of Downloads: 302 | Number of Views: 733 |
|