This preview shows page 1. Sign up to view the full content.
Unformatted text preview: tion.
The relationship of a response variable to the objective may be direct. An objective may be defined
in terms of a response variable-for
quantify the effect that thermal cycle B has on tensile
strength measured on customer qualifying tester X.”
In the case of CNC-machining, a response variable
is blade profile (see Fig. 4). This is related to the
objective through two measurement-performance indicators-mean
absolute difference of blade profile
and the target, and standard deviation of the difference. Sometimes a response variable may be a SUTrogate for the true response of interest. This is often
the case in destructive testing, in which a standard
stress-to-fracture test, for example, represents performance under conditions of use. Another example
is yield rate or failure rate, which are inferior responses that often represent where a specification
falls relative to a distribution of continuous-scale values (the collection of which provides superior information).
As discussed previously, the relationship of a response variable to the objective may be through performance measures that involve a comparison of the
response to a target or desirable outcome.
4. CONTROL VARIABLES As with response variables, most investigators can
easily generate a list of candidate control variables.
Control variables can be attribute or continuous.
They can be narrowly defined, such as “percent of
copper, by weight,” or broadly defined, such as
“comparably equipped pc: Apple or IBM.” In either
case, control variables should be explicitly defined.
When discussing potential control variables with
experimenters, it may be helpful to anticipate that
held-constant factors and nuisance factors must also
be identified. Figure 5 is a Venn diagram that can
be used to help select and prioritize candidate factors. It illustrates different categories of factors that
affect response variables, based on three key characteristics-magnitude
of influence on response
variables, degree of controllability,...
View Full Document
- Fall '14
- The Land