[关键词]
[摘要]
目的:采用觉规统计分析方法,分析最小化随机分组资料后的Ⅰ型错误和检验效能.方法:采用Monte Carlo模拟的方法,计算最小化随机分组的资料使用常规协方差分析的Ⅰ型错误和检验效能,并与简单随机化、中心分层区组随机化分析的结果进行比较.结果:简单随机化、中心分层区组随机化和最小化随机分组的资料采用协方差分析后,Ⅰ型错误均保持在0.05左右.3种分组方法的检验效能接近,无明显差异.结论:最小化随机分组的资料可采用常规的协方差分析方法,并能将Ⅰ型错误保持在0.05左右且不降低检验效能.
[Key word]
[Abstract]
This work aimed to discuss the type Ⅰ error and power of the conventional statistic analysis methods in minimization. Monte Carlo simulation was employed to compute the type Ⅰ error and power of ANCOVA, then the computation results were compared with those of simple randomization and center stratified blocked randomization. The simulations of the three randomization methods demonstrated that the type Ⅰ errors of ANCOVA were close to 0.05, and the test power was similar to each other. The results showed that ANCOVA is usable in the data analysis of minimization and may keep the type Ⅰerror close to 0.05 without decreasing the test power.
[中图分类号]
[基金项目]
国家科学技术部等14部委和广东省人民政府共同主办的“2009’传统医药国际科技大会暨博览会”第二分会——“传统医药在疾病防治中的作用与临床评价”论文,该论文为大会学术委员会评选的优秀论文;科技部国家“十一五”科技支撑计划课题(2006BA104A21-2):中医药防治重大疑难疾病综合治疗方案优化与临床评价共性技术和质量控制研究,负责人:何丽云;国家重点基础研究发展规划“973”计划(2006CB504601):中医辨证论治疗效评价方法基础理论研究,负责人:刘保延;中国中医科学院自主选题课题(Z02105):中医临床研究中动态随机化方法的选择和应用,负责人:闫世艳。