摘要
目的 探讨多水平模型在交叉设计资料分析中的应用价值。方法 通过两水平随机系数模型以及顺序效应的拟合，估计处理效应的大小及其在个体间的变异。结果 分别获得处理效应、阶段效应、顺序效应以及个体变异的估计。结论 在交叉设计资料分析中，当存在顺序效应时，多水平模型可分解并检验顺序效应的大小。
Objective To explore the special applicability of the multilevel models in analysis of crossover design.Methods Fitting two-level random coefficient model and sequence effect to estimate the treatment effect and its variation between individuals.Results Obtain the estimates of treatment effect,time period effect,sequence effect and individual variation separately.Conclusion While the sequence effect exists in analysis of crossover design,the sequence effect can be decomposed and tested,meanwhile,the treatment effect and the time period effect can be estimated by fitting multilevel models.
出处
《中国卫生统计》
CSCD
北大核心
1999年第5期273-274,共2页
Chinese Journal of Health Statistics
