Doe試驗中發現的一個難題

看中質協紅皮書doe試驗中,有個案例存在疑惑,一個三因子兩水平四個中心點沒有重複的試驗設計,數據如下
StdOrder RunOrder CenterPt Blocks Press Dis Angle strength
5 1 1 1 300 60 24 55.2
6 2 1 1 400 60 24 49.6
7 3 1 1 300 70 24 94.8
11 4 0 1 350 65 22 69.5
4 5 1 1 400 70 20 63.5
2 6 1 1 400 60 20 55.3
8 7 1 1 400 70 24 61.3
10 8 0 1 350 65 22 60.2
9 9 0 1 350 65 22 73.3
12 10 0 1 350 65 22 59.1
3 11 1 1 300 70 20 89.1
1 12 1 1 300 60 20 61.8
書中給出的分析結果如下
Factorial Fit: strength versus Press, Dis, Angle

Estimated Effects and Coefficients for strength (coded units)

Term Effect Coef SE Coef T P
Constant 66.058 1.619 40.80 0.000
Press -17.800 -8.900 1.983 -4.49 0.006
Dis 21.700 10.850 1.983 5.47 0.003
Angle -2.200 -1.100 1.983 -0.55 0.603
Press*Dis -11.750 -5.875 1.983 -2.96 0.031
Press*Angle -1.750 -0.875 1.983 -0.44 0.677
Dis*Angle 3.950 1.975 1.983 1.00 0.365


S = 5.60846 R-Sq = 92.35% R-Sq(adj) = 83.17%


Analysis of Variance for strength (coded units)

Source DF Seq SS Adj SS Adj MS F P
Main Effects 3 1585.14 1585.14 528.380 16.80 0.005
2-Way Interactions 3 313.45 313.45 104.485 3.32 0.115
Residual Error 5 157.27 157.27 31.455
Curvature 1 1.71 1.71 1.707 0.04 0.844
Lack of Fit 1 9.68 9.68 9.680 0.20 0.686
Pure Error 3 145.89 145.89 48.629
Total 11 2055.87


Estimated Coefficients for strength using data in uncoded units

Term Coef
Constant -320.167
Press 1.54200
Dis 6.05000
Angle -10.3250
Press*Dis -0.0235000
Press*Angle -0.0087500
Dis*Angle 0.197500
**問題點:
1.大家可以看到,雖然顯著因素有Press,Dis,Press*Dis,但是方差分析顯示二階交互作用不顯著,WHY??
2.關於第一步的分析結果,目前依很多教材得不到這個結果,原因是關於中心點的處理方式不同,LG/三星教材都是中心點在做完線性檢定後如果線性就直接delete**