变换数据生长曲线模型的参数估计

The Estimator of the Parameter in the Growth Curve with Transformed Data

本文主要讨论变换数据生长曲线模型有关参数估计的问题，给出了保持GME，容许性范围不变的变换（C1，C2），在γk（ciXi）＝γk（Xi），i＝1，2不成立时，提出了B的一种新的有偏估计，给出了其相对效率的上界，证明了这种有偏估计在MSE准则下可优于变换数据的GME，甚至也可优于无变换的GME。

In this paper, we mainly concern the estimator of the parameter in the growth curve withtransformed data. We give out the transformation (C1, C2) which retains the GME and region of admissibility . When γk(C1Xi)≠γk(Xi), i= 1,2. We propose a biased estimator of B give out the upperbound of it's efficiency. According to the mean square error (MSE) matrices . We have proved that thebiased estimator may even perform better than the Gauss - Markov estimator based on untransformed data.