Page 496 - Design for Six Sigma a Roadmap for Product Development
P. 496

Fundamentals of Experimental Design  455


                   Half normal plot of the effects  Half normal plot of the effects
                   (response is Y, Alpha = 0.05)   (response is Y, Alpha = 0.05)
                                      Effect type                      Effect type
              98                     Not significant  98              Not significant
                                     Significant                      Significant
              95                              95
                                     Factor Name                      Factor Name
                                      A  A                             A  A
              90                              90
                                      B  B    85                       B  B
            Percent  80               C  C   Percent  80               C  C
              85
              70
              60                              70
                                              60
              50                              50
              40                              40
              30                              30
              20                              20
              10                              10
              0                                0
                0  1  2  3  4  5  6  7           0  1  2  3  4  5  6  7
                       Absolute effect                 Absolute effect
             Lenth’s PSE = 2.175             Lenth’s PSE = 2.7
                      (a) ABC = 0, m = 24.7       (b) AB = 0, m = 19.5
                   Half normal plot of the effects  Half normal plot of the effects
                   (response is Y, Alpha = 0.05)   (response is Y, Alpha = 0.05)
                                       Effect type                    Effect type
              98                      Not significant  98             Not significant
                                      Significant                     Significant
              95                      Factor Name  95                 Factor Name
                                       A  A                            A  A
              90                       B C  B C  90                    B C  B C
            Percent  85                      Percent  80
                                              85
              80
              70                              70
              60                              60
              50                              50
              40                              40
              30                              30
              20                              20
              10                              10
              0                                0
                0  1  2  3  4  5  6  7  8  9     0   5    10   15  20
                        Absolute effect                 Absolute effect
             Lenth’s PSE = 3.15              Lenth’s PSE = 7.8
                     (c) AC = 0, m = 18.3            (d) BC = 0, m = 40.3
           Figure 12.19  Half-normal plots of Example 12.9 with one missing data value.
             point at zero, but other points in the half-normal plot form a line that is not
             pointing to the zero. So m   40.3 will be judged as an inappropriate esti-
             mate for the missing data m.
               In Draper and Stoneman’s method, we can use any of the estimates of m,
             from assumptions 1, 2, or 3, that is, m   24.7, m   19.5, or m   18.3, to put
             back to Table 12.19 and we will complete the experimental data analysis.
               Case 2:  Two data points are missing    In this case, we assume that there two
             missing data points. For example, we can assume that the responses for the
             third and seventh experimental runs, m   21.9, and x   27.5, are missing.
             In this case, if we use Draper and Stoneman’s method, we have to assume
             that two of the factorial effects are equal to zero. So we can obtain two equa-
             tions to solve for two unknown values m and x.
               In Example 12.9, we can make at least the following possible
             assumptions:
             1. If ABC   0, then
                      21.1   16.3   m   17.3   16.1   14.5   x   23.3   0
   491   492   493   494   495   496   497   498   499   500   501