Minitab 的 power and sample size 2n直交因子設計应用例
实验因子有无显著性的决定,除了因子本身具有效果外,还要实验误差(MSE)够小时就会有显著性,此涉及检出力与样本数(Power and sample size)的问题,例如怀疑A~E五因子对Y有影响,于是进行2**(5-2)筛选设计,进行8 runs 的配置与实验结果ANOVA如下表
Source
DF
Seq SS
Adj SS
Adj MS
F
P
Main Effects
5
450.476
450.476
90.0953
389.6
0.003
Residual Error
2
0.462
0.462
0.2312
Total
7
450.939
从Minitab 解析ANOVA表观察,此实验误差的估计值为√MSE = √0.2312=0.48
运用Minitab的 Power and sample size 本实验检出力
Stat > Power and sample size > 2-level factorial design 对话框填入
Number of factors:5
Number of corner points:8
Specific values for any three of following:
Replicates:1 2 3 4
Effects:
Power values: 0.8
Number of center points per block:0
Standard deviation:0.48
点Design钮
Number of terms omitted from model:2 (可删除的项目数) 按 [OK]、[OK] Minitab
输出结果如下
Power and Sample Size
2-Level Factorial Design
Alpha = 0.05 Assumed standard deviation = 0.48
Factors: 5 Base Design: 5, 8
Blocks: none
Number of terms omitted from model: 2
Center Points
Total Reps
Runs
Power
Effect
0
1
8
0.8
1.91886
0
2
16
0.8
0.74639
0
3
24
0.8
0.58056
0
4
32
0.8
0.49393解读:若高低水平差异1.92以上,不必重复,就有80%基会检出有显著性
TA的首页

