Mean square error tests for restrictions in singular linear models |
| |
Authors: | Burt S. Holland |
| |
Affiliation: | Department of Statistics, Temple University, Philadelphia, PA 19122, USA |
| |
Abstract: | Weak and strong mean square error tests of restrictions presented in Wallace (1972) are generalized to apply to singular linear models. The singularity necessitates a slight change in the strong m.s.e. criterion and the requirement that the restrictions be estimable, but otherwise the tests are applied in a fashion analogous to the non-singular case. Use of those tests implies that the solution for the linear model parameter vector is contingent on a test result. The risk behavior of these contingent solutions is discussed. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|