Cpk Basics without Charting?



Can it be explained how charting can actually benefit small production runs?
Last edited by a moderator:

Bev D

Heretical Statistician
Staff member
Super Moderator
sure we can explain it. can you give us more specifics about why you think it won't? what are trying to learn? how does charting tie to Cpk (from your title)

Steve Prevette

Deming Disciple
Staff member
Super Moderator
There are indeed some folks who say just calculate the standard deviation of the data and compare it to the specs.

However, if the data are not stable, the calculation will be meaningless, as will any histogram pretending to state it is the distribution of the data.

Generally speaking, the most effective and efficient way to determine if the data are stable and a CPK calculation is warranted is to use a control chart.

Bev D

Heretical Statistician
Staff member
Super Moderator
what do you mean by "charting has everything to do with it"? what is 'it'? why do you ask a question about the value of charting for short runs and then respond that "charting has everything to do with it"?
I'm sorry but I'm finding your question a bit cryptic...

However: my experience shows that the most powerful statistical or analytical tool we have is to plot our data in time series. simple statistics like average, standard deviations, p values and Cpk indices simply aren't very informative. patterns (trends, shifts, shape, cycles, etc) can't be seen with these numbers.

On the other hand, if you are only calculating a Cpk value because of some requirement that nobody will do anthing with beyond docuemtning that you did it and you aren't interested in understanding the process, plotting the data is a waste of time.

Can you exist without charting - of course man has existed for thousands of years without charting. But existing isn't really living, now is it? :notme:


Stop X-bar/R Madness!!
For short runs, correct charting may be your best information. Cpk is just a silly little single number - really has no depth of information at all. Shorts run processes may not even be capable, in the traditional sense. For machining, for example, if the run is short enough, the process never reaches steady state during most of the run - so, therefore not stable. Parts may still be good, but the process may not be predictable. So, charting will help you watch an unstable process, and let you know if any intervention may be needed.