缺失数据下广义线性回归拟似然估计的强相合性

Strong consistency of quasi-likelihood estimator in generalized linear models with missing data

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.