本文在椭球分布族下研究一阶自相关线性混合效应模型的约束极大似然估计问题.分别考虑位置参数在等式和不等式线性约束这两种情况下的极大似然估计值.同时对约束条件下的兴趣参数给出三种渐近等价的检验方法.蒙特卡洛模拟说明本文方法的有效性和稳健性.本文结合模型对Framingham心脏研究中的胆固醇水平进行了分析.
In this paper, we focus on applying restricted maximum likelihood to the linear mixed model with auto-correlated error structure under the elliptical distribution class. Maximum likelihood estimation under both linear equality and inequality constraints on location parameters are considered. Three asymptotically equivalent tests are also constructed for the interest parameters under restrictions. Monte Carlo simulations are conducted to evaluate the effectiveness and robustness of the new methodology. An illustrative example for the cholesterol levels from the Framingham heart study is reported.