Process Screening Chart Options - Show spec limits

[#1]

So it looks like this is only available for I-MR and not Xbar-R/S, do I have that correct?  If so does anyone know why?  I can see that spec limits are more valuable when looking at I-MR but its still helpful for xbar charts as well.

[#2]

Spec limits have no place on X bar charts. They provide no insight into conformance to specification because the plotted statistic on an X bar chart is the average of observations in a sub group. So it's potentially possible that you could have a plotted point within the spec limits and ALL the individual observations actually outside the spec limits...leading to a gross error in concluding that everything is just fine with the process wrt to conformance to specification.

[#3]

Hi @emorris1000 ,

To add @P_Bartell explanation, X-bar is displaying the average value where as the Spec Limit refers to individual value.

To avoid confusion or possibility of interpreting the result wrongly..

I also ponder the same question like you have when I first started using Process Screening function.

My comments:

• It is only helpful to display spec limit when the X-bar that you plot also display the individual points (i.e dotted individual points).
• It is still useful to add Spec limit in X-bar, provided the individual points being displayed (for this, I use the Control Chart builder).
• I've seen people suggesting to me an X-bar control chart that is "statistically" in-process control (i.e average is OK & within control limits), but the individual points outside of Spec limits. This suggestion is not helpful to the decision-makers. To avoid this issue, it is good to display spec limit in X-bar chart (as long as all individual points displayed)

Personally, I would exclude the display of control limits/I-MR in the process screening function because it cause confusion to our stakeholders. 

[#4]

To be honest I'm a bit surprised that people see no value in having it with an Xbar chart.  Granted, its not nearly as clear as in an I-MR chart as you have to have an intrinsic understanding of the dispersion to interpret it, but if you have it on there with the calculated control limits you can eyeball it fairly effectively. 

[#5]

What are you trying to 'eyeball' by placing spec limits on an X bar chart? 

[#6]

My thoughts, In addition to the points made by @P_Bartell putting spec limits on control charts (any) completely misses the purpose of control charts and confuses the application of the control chart methodology.  Specifications are almost always independently derived and have no relation to the control limits.  Control limits are a function of the process and depend on how the data was acquired (e.g., sampling frequency and rational subgrouping strategies).  Specification limits may answer questions about capability and may alert if there is an issue.  Control limits perform a completely different function.  Control limits on Range or Moving Range charts answer the question about stability.  Is the within subgroup variation (a function of the x's changing within subgroup) consistent, stable?  If not you should seek to understand why.  The control limits on the X-bar chart are a function of the within subgroup variation.  The X-bar chart is a comparison chart and answers the question about where is the leverage.  Is there more variation between subgroup (a function of the x's changing between subgroup) than predicted by the within subgroup variation?  If, the leverage (x's with the greatest affect on the plotted response) would be the between subgroup sources.  If not, your focus should be on the within subgroup sources.  Spec limits add confusion to the purpose of the control chart.

Now, if you want to plot data against spec limits, by all means go forth.  It is just not the control chart method.

For the Moving Range, Individual chart, the questions are a bit different as there are is no rational subgrouping strategy.  The MR chart still answers the question about stability, but it is the stability of consecutive measures.  The Individual chart then compares longer term changing x's (that might shift or drift, for example) to shorter term changing x's.

A quote from Shewhart (the inventor of the methodology):

“The engineer who is successful in dividing his data initially into rational subgroups based on rational theories is therefore inherently better off in the long run. . .”

Perhaps you might want to read:

Shewhart, Walter A. (1931) “Economic Control of Quality of Manufactured Product”, D. Van Nostrand Co., NY

Wheeler, Donald, and Chambers, David (1992) “Understanding Statistical Process Control” SPC Press (ISBN 0-945320-13-2)

[#7]

putting spec limits on control charts (any) completely misses the purpose of control charts and confuses the application of the control chart methodology.  

Totally agree.

[#8]

If you are bound and determined to present process performance vs. specification limits, JMP provides a large number of data visualization tools far more well suited to this analysis goal outside control charts...the Process Capability platform is a great place to start. Many other pathways as well...Control Chart Builder and on and on. Show the control charts right alongside meaningful and appropriate process capability visualizations and you've got a killer combination, correctly done. 

[#9]

All right, well thanks for the info.   Did a bit more digging online and the nature of the responses make a bit more sense to me now.