|
|||||
|
|
A synthesis of stein-rule and mixed regression procedures inlinear regression models under non-normality |
|
| Authors: | Derrick S Tracy Anil K Srivastava |
| Institution: | 1. Department of mathematics and statistics , University of Windsor , Windsor, Ontario, Canada;2. Dept. of statistics , University of Lucknow , Lucknow, India |
| Abstract: | Stein-rule philosophy and mixed regression technique are combined to develop two families of improved estimators of regression coefficients in the linear regression model under incomplete prior information. The properties of these estimators are studied when disturbances are small and non-normal. Conditions for their dominance over mixed regression estimator are derived taking risk as the criterion for performance. |
| Keywords: | linear regression model mixed regression stein-rule incomplete prior information non-normality |