含有缺失机制的多元纵向数据分析

目的本文旨在对含有不可忽略缺失机制的多元纵向数据建立一个适当的统计模型。方法对纵向数据建立含有潜在变量的线性混合模型;由于潜在变量在本文中代表治疗效果,而且随着时间变化在不断改进,所以本文用一阶的马氏链来反应潜在变量之间的这种联系;引入logistic回归模型来描述数据的缺失机制。最后利用EM算法对参数进行估计,并且给出了数据的模拟结果。结果从模拟的结果可以看出,参数的96%置信区间包含了待估计参数的真值。结论本文所提出的方法对于估计类似数据的参数具有一定的有效性,相对于传统方法,适用性更广泛,且大大简化了计算的工作量。

Analysis of Multivariate Longitudinal Outcomes With Nonignorable Dropouts

Objective To set up an appropriate statistical model with regard to the analysis of multivariate longitudinal outcomes with nonignorable dropouts.Methods Linear mixed models are used to model the relationship between the observed outcomes and the latent variable.To account for the improvement of the latent variable though time points,the Markov Chain with first order was applied to express the relationship.Specially,in view of the data with nonignorable dropouts,logistic regression was used to model this missing mechanism.The EM algorithm was used to estimate the model parameters.Results The proposed method performs well in finite samples.Conclusion It comes to conclusion that the proposed approach was useful for dealing with this types of multivariate longitudinal outcomes.