摘要
统计软件SAS中PROCMIXED的发展使得混合线性模型非平衡试验数据的分析在技术上变得简单可行。但在小样本非平衡数据条件下,选择怎样的自由度计算方法用于固定效应统计检验,才能取得较准确的分析结果,则是一个尚未明确的问题。本文利用PROCMIXED提供的几个自由度计算方法对3个常用农业试验设计的非平衡数据进行了分析。MonteCarlo模拟研究结果表明,Kenward Roger法能够取得第一类统计错误率的最佳控制。
The development of PROC MIXED of SAS has made analysis of the linear mixed model with unbalanced data technology accessible.However,a sticky problem for the procedure has been the specification of appropriate degrees of freedom for test of moxed effects in unbalanced data with small sample sizes.This paper considered the analysis of unbalanced data from 3 experimental designs frequently used in agricultural research. Several methods for aproximating the degrees of freeom are employed,all of which are available in PROC MIXED. The Monte Carlo simulation results showed that the Kenward-Roger method provided the best control of the Type I error rate.
出处
《数理统计与管理》
CSSCI
北大核心
2005年第1期45-51,57,共8页
Journal of Applied Statistics and Management
