The following artificial data are taken from Daniel (1976, p. 65) to illustrate the effect of one “bad value” in a factorial design. Find the test that seems to be influenced by a special cause. First make a run chart and identify the most likely special cause. Then put the data on a cube and identify the point. Why is it easier to find the special cause on the cube? What value would be more consistent with the rest of the data?

Transcribed Image Text:

Factors Run Test Order A B C Response 1 4 158 2 3 45 5 + 132 2 8 + + + 212 136 5 3 + 264 17 6 7 + + 188 1 1 + + 318 8 6 + + + 242