A Systematic Approach to Planning for a Designed Industrial

3 and c the knowledge sought through experimentation

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ; and (c) the knowledge sought through experimentation (col. 4). It is helpful to know the current state of use, and if it is unknown, the experimenters are advised to include some trial runs prior to the experiment or “checkpoint” runs during the experiment (perhaps these data have not been previously acquired). The current distribution serves as one of several possible reference distributions for judging the pracrical magnitude of the effects observed. Given a typical standard deviation for a response variable of u, a lowto-high control-variable effect of u/2 may be of no practical significance, but one of 4a may be important. Another advantage to knowing the current state of use is a check on credibility. Process or design limitations may constrain a response variable to be bounded on one or two sides. An experimental result outside that range may be erroneous or due to an abnormal mechanism (which may, however, be of interest). Measurement precision (and, in some cases, bias) and how to obtain it (i.e., choice of measurement system or repeated measurements) is a thorn in the flesh for many experimenters. The admonition of Eisenhart (1962) serves as a relevant (if overstated) warning, “until a measurement operation . . has attained a state of statistical control it cannot be regarded in any logical sense as measuring anything at all” (p. 162). It has been our experience that many TECHNOMETRICS, FEBRUARY 1993, VOL. 35, NO. 1 C. MONTGOMERY experimenters do not know the state of control nor the precision and bias of most measurement systems measuring a response or a control variable. The measurement systems were not useless, they were just of unknown utility. Important and poorly understood systems should be evaluated with a measurement capability study, a designed experiment of the measurement process. As a compromise, one might be forced to resort to historical data and experience or weaken the experimental objective to obtain ranking, selection, or a binary response instead of quantifica...
View Full Document

Ask a homework question - tutors are online