论文部分内容阅读
对含有非随机缺失数据的潜变量增长模型,为了考察基于不同假设的缺失数据处理方法:极大似然(ML)方法与DiggleKenward选择模型的优劣,通过Monte Carlo模拟研究,比较两种方法对模型中增长参数估计精度及其标准误估计的差异,并考虑样本量、非随机缺失比例和随机缺失比例的影响。结果表明,符合前提假设的Diggle-Kenward选择模型的参数估计精度普遍高于ML方法;对于标准误估计值,ML方法存在一定程度的低估,得到的置信区间覆盖比率也明显低于Diggle-Kenward选择模型。
For the latent variable growth model with non-random missing data, in order to investigate the missing data processing method based on different assumptions: the superiority and inferiority of the maximum likelihood (ML) method and the DiggleKenward selection model, Monte Carlo simulation study, comparing two methods The difference between the accuracy of the growth parameter estimation and the standard error estimation in the model is considered, and the influence of sample size, non-random missing ratio and random missing ratio is considered. The results show that the accuracy of the parameter estimation of the Diggle-Kenward selection model that meets the prerequisite hypothesis is generally higher than that of the ML method. For the standard error estimate, the ML method has a certain degree of underestimation, and the obtained confidence interval coverage ratio is also significantly lower than that of the Diggle-Kenward selection model.