How do I select the correct SPC chart in Process Screening

[#1]

I have a data set that includes the percentage of patients seen within 13 weeks, pre and post a kaizen event.

I selected Process Screening which produced the following...

The pre selected SPC is I-MR.  Should it not be a P chart?  It doesn't seem to allow me to change this.

Is there a function in Process Selection that allows me to visually display the changes more effectively in a presentation?

Thank you

[#2]

Hi @Noeleen20350465 ,

The Process Screening platform is primarily focused on measurements of process capability within set specification limits, as the P-chart is focused on a proportion of defects (where the number is produced by the amount of samples that have exceeded spec limit). This is why in the process screening on I-MR appears (along with other options available when booting up the platform).

Here is another discussion in the community on this.

If you want to present the data in an effective way, you could consider using a JMP project, with each tab showing different measurements of your defects. Or you could use the 'New Y Chart' option in the control chart builder to stack your observations.

Thanks,

Ben

[#3]

Here are my thoughts:

Before randomly clicking buttons in the software, What questions are you trying to answer?

It appears you have one response variable, % of patients seen within 13 weeks?  % of what? Why 13 weeks? Does it matter the results of them being "seen"?  Do you care about the consistency of the % of patients seen?

It seems you are trying to determine the effectiveness of the "kaizen"?

One could argue the response variable could be considered continuous (perhaps a binomial approximation...), so use of a continuous control chart may be appropriate (see Wheeler).  Since it does not appear you have any rational subgroups, then the I-MR chart is the correct one.  This chart will not be useful in assigning components of variation, but may be useful in assessing consistency.  You could also argue this is count data and use attribute charts.

In any case, you would be much better off presenting the information in graphical form.  Either using graph builder or control charts.  If you attach your data set, I could add some graphical scripts to it.

[#4]

Thank you for your response.

I hope this answers some of your questions.

I am trying to assess whether the Kaizen event as a whole, or the individual inputs that were put in place, made a statistically significant difference to the outputs.

The actions of the Kaizen were to introduce staff training, introduce standard work for chronological scheduling, routine waiting list management and urgent waiting list management, and increase utilisation of available theatre capacity.

The outputs I wish to analyse pre vs post Kaizen, and pre vs post implementation of the actions above are:

• Number of patients waiting over 28 days for an appointment (target = 0)
• Number of patients on routine waiting list (no target or spec limits)
• Longest wait time (no target or spec limits)
• Number of routine patients waiting ≤ 13 weeks for an appointment (no target, spec limits)
• % of routine patients waiting ≤ 13 weeks for an appointment (KPI is ≥ 65%)
• Number of routine patients waiting ≤ 9 months for an appointment (no target, spec limits)
• % of routine patients waiting ≤ 9 months for an appointment (KPI is ≥ 95%)

The clinical outcomes were exceeding KPIs and therefore were not included in the scope of this project.

I am hoping to present the data graphically, and also determine if the change is statistically significant, and if the processes are more stable and more capable.

I'd be grateful for any direction that you can provide.  I've attached the data set as requested.

Thank you

[#5]

Noeleen,

Your questions whether any of the changes made by doing the Kaizen or elements of the Kaizen are "statistically significant" is very challenging.  There is not enough data to suggest the changes made had the effects you were interested in, although it appears there were some improvements. With only 7 data points post Kaizen, there really is not enough to conclude anything about stability.

Here is what I did with your data table (I added the scripts, simply click on the green arrows to run the scripts):

1. Coded the columns for more flexible analysis options (whenever you use words or letters in the data table, you restrict analysis).

2. Color coded the pre and post Kaizen.

3. Ran multivariate on the 7 responses indicated above.  There were some outliers in the data (see script).

4. Ran IMR charts for each of the 7 responses.  These charts included all of the data, both pre and post kaizen

5. Ran distributions of the 7 responses including the Kaizen.  The output of this will be dependent on the preferences for Distribution in JMP. Check only Histogram & Vertical.  You will see I placed the Kaizen first in the list.  When you highlight the pre section of the distribution, it will show corresponding results in the 7 distributions.

6. I ran IMR charts By Kaizen.  So you will the charts done for pre-kaizen and again post kaizen.

7. I ran a fit model with just kaizen in the model.  There is not enough balanced data for the elements within the kaizen.

8. Graph builder.  Too many options to try, and you can play with this.  Just drag and drop to the zones.  I added some to get you started.

In the future it might be useful to design experiments or components of variation studies as a more effective and efficient means of testing the effects of your improvement efforts.  I also would suggest using actual wait time vs. categorization.

[#6]

Wow!  Thank you so so much for that. I really appreciate it.

I'm very new to six sigma and JMP (only using it a few months) so it will take me a few days to go through everything you have sent, try to understand your process and how to reproduce it myself.  If you don't mind, I'll come back to you then if I have any questions.

Thank you also for your suggestions for future studies. I will look at this for the next project.

[#7]

Happy to help.  Let me know if you need any other assistance.

You might take advantage of the on-line JMP support to familiarize yourself with the software:

https://community.jmp.com/t5/Learn-JMP/ct-p/learn-jmp

[#8]

Selecting the appropriate Statistical Process Control (SPC) chart in JMP's Process Screening platform depends on your data type and specific analysis goals. For instance, continuous data might require different charts than attribute data. Similarly, when facing complex academic assignments, utilizing an essay writing service at here 

[#9]

I can select P charts when I'm in the control chart section, and can do a pre-post comparison there.

I don't seem to be able to change it in the process screening section though.  As another responder mentioned, maybe I shouldn't be using the process screening for this metric??  I like that it gives a stability index and sigma level pre and post kaizen and that's why I'd hoped to use it.