Admissible Necessary Condition of Quadratic Estimate of Error Variance on Growth Curve Model
Zhao Jian-xin · Qingdao Haiyang Daxue xuebao · 2004
For the growth curve modelY=ABC+e Ee=0,cov(vece)=σ2(I pG)and the quadratic loss function, a necessary condition that the quadratic estimate of error variance is admissible is obtained when A is not a full row rank matrix.