Page 144 - Six Sigma Demystified

Chapter 6 a n a ly z e S tag e 125

owing to common causes. This provides valuable input in analyzing the sources
of variation.
Special causes of variation are evidenced by out- of- control points. The
time- ordered nature of the control chart provides an indication of when the
process was influenced by a special cause to aid in an understanding of what
happened (in process terms). Since the control chart will not detect all shifts,
nor necessarily detect shifts as soon as they occur, neighboring subgroups within
the control limits also could result from the assignable (special) cause under
investigation. Bear this in mind when you try to associate the timing of the
special cause with possible changes in process factors. Table T.19 in the section
“Statistical Process Control Charts” of Part 3 provides an indication of the num-
ber of subgroups required (on average) to detect a process shift of a given
magnitude.
The common- cause variation (represented by the control limits) is due to all
elements of the process. These sources of variation cannot be easily associated
with the point- to- point variation because they represent a combination of fac-
tors common to all the subgroups. There are likely to be several sources of
variation that contribute to that common- cause variation, and a subgroup’s
specific value is a reflection of the combination of those sources.
Teams brainstorm to understand potential sources of variation for a process.
Sources of variation are found in a process’s methods, its materials, the man-
power or personnel involved in the process, the machines or equipment used,
the measurement techniques, and the surrounding environment. Collectively,
these potential sources of variation are sometimes referred to as the 5M’s and
E. In service processes, the 4P’s may be more useful— policy, procedures, plant,
and people.

Categorizing the potential sources of variation using one of these techniques
aids the team in brainstorming. Each of the categories provides a focus to ensure
that potential causes are not ignored. As in any brainstorming exercise, indi-
vidual responses should not be critiqued or criticized. The purpose of the brain-
storming is to develop a comprehensive list of all possible sources, not just the
likely ones. A typical cause- and- effect diagram provides a fishbone layout for
displaying these root causes, as shown in Figure 6.2. The actual format of a
cause and effect can take any form so long as the hierarchal structure is main-
tained, with causes and subcauses grouped by an appropriate classification
scheme.
The nominal group technique described in Part 3 also can be used to reach
consensus on the list of causes to investigate. In this approach, each team mem-