Quality in Design - Medical Device Design: Innovation from Concept to Market - page 125

Table 7.5: Experiment Randomized and Completed

Random
Run Number

Experiment Variable 1

(X1)

Variable 2
(X2)

Variable 3
(X3)

T P Tm Result

(Q)

7 1

+ 1 + 1 + 1

Max Max Max 4

6 2

+ 1 + 1 − 1

Max Max Min 1

4 3

+ 1 − 1 + 1

Max Min Max 2

5 4

+ 1 − 1 − 1

Max Min Min 4

3 5

− 1 + 1 + 1

Min Max Max 2

8 6

− 1 + 1 − 1

Min Max Min 3

2 7

− 1 − 1 + 1

Min Min Max 4

1 8

− 1 − 1 − 1

Min Min Min 4

We analyze the results taking each +1 and −1 of each variable in turn in relation to the result
Q. So for variable X1 the +1 results were experiments 1-4. For variable X2 the +1 results
were experiment 1, 2, 5, and 6. We analyze them in this fashion:

Average +Q = Sum (+1 values of Q)/Number of results
Average −Q = Sum (−1 values of Q)/Number of results
Variance of Q = (Average +Q) − (Average −Q)

Or, more elegantly

∑

Q () 1

E

1

k 1

2

∑

Q () 1 (7.1)

E

2

k 1

2

EEE

1 2

For variable X1 this would be as shown in Equation 7.2 and Table 7.7 .

Maximu m M inimum

E (. 137142 +++ . 11114/4132 . .) . E = (92184 0 . . 11137)/4 . 128 .

1 2

(7.2)

Overall
E 132128 004 . . .

Similarly we can determine the values for the system for variables X2 and X3 ( Table 7.7 ). Or
we can plot these on an effect graph ( Figure 7.4 ).

Both Table 7.7 and Figure 7.4 illustrate that the variable X3 has the highest effect - its slope
is negative which means the effect is decreasing. X1 and X2 seem to have a similar effect.