Question: Brush Ferrules. Adams, Harrington, Heemstra, and Snyder did a quality improvement project concerned with the manufacture of some paint brushes. Bristle fibers are attached to a brush handle with a so-called ferrule. If the ferrule is too thin, bristles fall out. If the ferrule is too thick, brush handles are damaged and can fall apart. At the beginning of the study, there was some evidence that bristle fibers were falling out. "Crank position," "slider position," and "dwell time" are three production process variables that may affect ferrule thickness.
(a) What feature should analysts measure on each brush in this kind of problem?
(b) Suggest how an engineer might evaluate whether the quality problem is due to poor conformance or to poor design.
(c) From the limited information given above, what seems to have motivated the investigation?
(d) The students considered plotting the variable identified in (a) versus the time at which the corresponding brush was produced. One of the analysts suggested first sorting the brushes according to the different crank position, slider position, and dwell-time combinations and then plotting the variable chosen in (a) versus time of production on separate graphs. The others argued that no insight into the problem would be gained by having separate graphs for each combination. What point of view do you support? Defend your answer.