摘要
利用蒙特卡罗(MonteCarlo)模拟考察了随机区组设计、裂区设计和条区设计因数据缺失形成的数据非平衡性对试验分析结果的影响。结果表明,数据非平衡性可导致方差组分估计的精准度降低、固定效应误差的偏低估计以及固定效应测验一类统计错误率的增大。缺失数据越多,这一现象表现越明显。数据缺失对随机区组设计分析精准度的影响小,而对裂区和条区设计分析精准度的影响大。
This paper examined the influence of data imbalance on the analysis results for random block design,split-plot design and strip-plot design using the Monte Carlo simulation.The simulation results showed that the imbalance of data due to missing observations led to lower precision and lower exactness of variance component estimation,to underestimation of the error of fixed effects as well as to higher type Ⅰ error rate for fixed effect test.With more missing observations these influences would be heavier.The imbalance of data influenced the analysis results in split-plot design and strip-plot design much heavier than in random block design.
出处
《西北农林科技大学学报（自然科学版）》
CSCD
北大核心
2004年第1期103-107,112,共6页
Journal of Northwest A&F University(Natural Science Edition)
基金
西北农林科技大学校长基金资助项目
