What causes Ppk to be greater than Cpk when a basic capability analysis is run?

Hello Everyone,

I know this have been a hot point of discussion in many threads here on the Cove however I am bringing the whole Cpk/Ppk thing again with a different question.

The question of the hour is this: What causes Ppk to be greater than Cpk when a basic capability analysis is run on a set of data? Generally speaking the Ppk will usually be lower than the Cpk, due to the overall sigma calculation vesus the R bar/d2 calculation however on occasion we will have data that shows a higher Ppk.

Does this have to do with the subgroup variation?

Any insight would be appreciated.

Regards,

Jon

I suppose this could be related to the fact that Ppk's variance is estimated using the sample standard deviation and Cpk's variance is estimated using the range. I haven't taken the time to verify that though. My guess is that when you see Ppk larger, it isn't by much.

Ken

Sigma from Ppk is calculated using all of the individual measurements.

Sigma for Cpk is calculated using the variation of the subgroup totals. RbarD2

Since there is more variation in the individual measurements the subgroup totals, Ppk will usually be lower that Cpk.

That is why Ppk is a preffered methododology to use during initial capability analysis.

Generally I see <10% difference between the two.

Thanks for the clarification!

But what should be the interpretation when the difference is bigger than 10%...

Can we interpret that the sample for the subgroup is bad or too similar... ???

Can you post your data?

I don't have any data, just questioning the logic between Ppk vs Cpk when the results are different...

To found out if the sampling are appropriate, or what that have impact...

Imho,

if the process average doesn't move during production the Ppk may be greater then Cpk (tested by numerical simulation).
Please consider that d2 is an estimator within a confidence level

But if the Ppk is getting bigger and bigger apart from Cpk, gap growing between the two, is this means that the average still stable, but the subgroup are more and more disperse...???

This is too better understand the difference between them, and also be able to read it when this happen.

By understanding what influence the result of Ppk and Cpk.

PpK is measuring your potential. CpK is measuring your actual. So for example your firing darts at a target, they are all grouped together but not on the bullseye. Your friend then tells you that the wind is 5mph to the left(THE CAUSE) You now readjust your aim and fire again taking into consideration that the wind is 5mph to the left. All the darts now hit the bullseye and your PpK & CpK values are identical.

A

Is it possible or logic that Cpk is getting bigger than Ppk that indicate you the commun cause are more present (something wear out) or the employe is having fun to adjust all the time the process parameter to reach the mean but having an increasing of range into the subgroup

???