Getting Insight into Your Process Stability, Quality and Capability

Video using JMP 18 was posted in June 2025.

Do you need to identify routine (common-cause) and abnormal (special cause) process variation over time? Are you tasked with comparing how well  a process is performing compared to its given spec limits or after a process is changed? Are you responsible for understanding stability and capability metrics for a large number of processes across time?

In this case study we use process information to see how to:

• View and building control charts to find variations.
• Manage spec limits and attach to data.
• Screen processes to identify possible problems, out-of-control points, and visual summary of process issues.
• Compare variability in processes related to changes (phases) in process.
• Plot the relative frequency or severity of process problems, including before and after an intervention.

Suggested prerequisites:

• Familiarity with common process control terminology.
• Background in control charting and common analyses used for identifying process issues.            

Here are two related activities on this topic:

• 90-minute instructor-led complimentary Jump into Control Charting course - August 6.
• Complimentary slef-paced, online JMP course on Statistical Process Control. which you can take at any time.

Questions answered by @ChristianStopp and @alisa_h_lowery at the live webinar:

Q: What is the difference between Ppk and Cpk?

A: Cpk represents the short term (within-subgroup) variation, whereas the Ppk represents long term (overall) variability.  There is also there's a great module on Quality in the free STIPS content we offer, which can be found here: https://www.jmp.com/en/online-statistics-course.

Q: What do you do for Ppk if the data you have is not normal, skewed or a normal 2 mixture, close to normal but really not quite there?

A: You can specify non-normal distributions within the capability platform or using the column properties for the measure.

Q: Is there a way to combine all these individual process capabilities to one overall in a way, perhaps, to grade two different plants?

A: You can use a 'by' variable in process capability.  If you do combine, the goal plot can do it. Capability plot lets you look at range. The reports may look busy due to the number of variables. What you are requesting may be addressed by a multivariate SPC framework, which currently isn't supported here, but it's something that JMP Development  is looking into.  On the other hand, comparing individual capabilities across plants might be a reasonable strategy for comparing performance across them and there are other practical 'gauge offset' strategies for which you can use JMP.

Q: Can you use Ppk/Cpk on logistics data?

A: I could see them being used in that sense as a way to gauge how your process is performing relative to desired benchmarks.

Q: How many points do you need in order for Control Chart analysis to be useful?

A: Generally the more the better. You can  find short-run charts available in JMP if you're looking at sets of products for which you're exploring their stability.

Q: What is the Alarm Rate in Process Screening?

A: Alarm Rate contains information about the subgroups that result in alarms for a variety of tests, including each of the 8 Western Electric rules. The standard deviation estimate is the Within Sigma value. By default, only the Alarm Rate, Test 1, and Latest Alarm columns are shown in the summary table. Alarm Rate is the number of subgroups that resulted in alarms for any of the tests selected under the Choose Test option (Any Alarm) divided by the number of non-missing subgroups (Subgroups).

Q: If you are able to identify an assignable cause to for the special cause variation, is the remainder of the process stable and within control?

A: In such a case, we would suggest, if feasible, to collect more data and use phases to prove it.  It is not optimal to hide or exclude information.