研究了响应变量随机缺失下的广义线性模型, 利用处理缺失数据的完全数据方法,构造了广义线性模型中未知参数的拟似然估计. 结合随机缺失机制和若干极限定理,证明了该拟似然估计的渐近存在性、强相合性和收敛速度.

Generalized linear model is studied when the responses are missing at random. By use of the complete-case data method, the quasi-likelihood estimator of unknown parameter is constructed. Combining the missing mechanism and some limit theorems, the asymptotic existence, the strong consistency and the convergence rate of the proposed quasi-likelihood estimator are proved.