Again, and again the hotline or Q-DAS employees are asked whether the Cp/Cpk value can be calculated and displayed in the O-QIS products CMM-Reporting or procella directly after the value input.
For many reasons, the question alone is wrong. In this very untypical document, the author of this document (Thomas Gastgeb, hotline, and head of the hotline for 16 years) tries to explain these questions, why this is not possible, and what the real question behind this question is.
Procedural error with this question
Due to the fact that the O-QIS products do not have a setting to calculate the capability characteristic values directly after the value input, the "simulation" is carried out here in the value mask in qs-STAT.
Consider a feature, which is generally capable first:
Now a sample of values is recorded which makes the C-values go bad:
If this request where to be realised, it would only be an alarm in the O-QIS products.
So far, this would not be a problem.
Now the user records the next sample (after successful corrections of the process). A subgroup which again lies completely precisely in the middle of the tolerance:
In order to simulate again what would happen in the O-QIS, the evaluation is carried out:
As was to be expected, the index is of course still "bad". Because the measured values that led to the bad index are of course still in the calculation:
Means:
The user gets an alarm again: the capability indices would be bad.
This alarm now runs through the entire coming data collection as long as the "bad values" are part of the calculation.
The result is over-regulation and massive user uncertainty during data collection.
At this point, the following counter-statement is often made:
"The values, this last subgroup, which led to the bad C-values, you would delete them"!
This statement must be rejected directly. This is a form of data manipulation that must not be welcomed by the applying client, its clients, nor by the official side (auditors). This would be data manipulation. The recorded values / samples reflect an extract from the population, which must be included in the recording and calculation in order to ensure a realistic representation in the statistics. Anything else must be considered data manipulation, in the worst case it will be interpreted as fraud.