iSixSigma

Mix Weekly and Monthly Data in a Control Chart?

Six Sigma – iSixSigma Forums General Forums Tools & Templates Mix Weekly and Monthly Data in a Control Chart?

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #55032

    Klaus M.
    Guest

    Hello,
    maybe a simple question, however, I like to get other consideration.
    Do you see any issue by showing monthly (before) and weekly (after) datapoints in the same control Chart? Data will be the same, just the cadence during the improvemment.

    Thanks

    0
    #198286

    Shelby Jarvis
    Participant

    Klaus,

    I am assuming you have run the charts separately but hope to find a way to communicate the results of the teams improvements.

    You have two potential issues.
    1) The value of SPC is using sampling to represent the population. During projects, it is not uncommon to measure more frequently, but in essence you have changed your sampling plan as compared to normal production. You must make this clear in your communication.
    2) What is the purpose of monthly? Does your process have a long term deterioration which is not detectable in shorter time periods? Make certain you can address any risk in the change in the measurement plan as your stakeholders need to understand this as well.

    If you feel you have mitigated your risk and have legitimate control charts, I suggest showing them in a manner which your company culture can understand and accept.

    Good Luck

    0
    #198287

    JB
    Participant

    @KlausM,

    I like the advice you got from @ShelbyJarvis. I’d advise against that significant of a change in the time interval of data collection for an SPC chart. The goal is to identify and react to adverse changes that occur over time.

    If your goal is simple to visually represent the improvement of the “before” vs “after” data, then I’d suggest using the “stages” option in Minitab (if you have it) so the audience will clearly see the difference between the two sample sets of data, if you want to stick with using an SPC chart. I’d suggest using histograms or capability analysis side by side to show the difference in the data, rather than a SPC chart.

    0
    #198289

    Amit Kumar Ojha
    Participant

    I Agree with JB. Better use the same period before and after for showing the data in SPC chart and use simple Box Plot to compare the peformance befor and after.

    0
    #198301

    Don Hutchcraft
    Participant

    My current position does this already. We get daily, weekly and monthly data given to us to show how well we are performing within the current process maps (which are 4 years old) I find this to be a horrible way to “manage” and show performance as a before and after scenario. The reason is that my current position has technicians with multiple responsibilities and not every failure (although the same nomenclature for the fault) has the same root cause and must be handled in a different way every time. To realistically show performance we would need about 20 different control charts to show the true measurement.

    0
    #198312

    The only reason you would require up to 20 control charts would be that you have up to 20 different variables you’re being measured on. Seems as though this is far too many to effectively manage, but not knowing anything about your process, maybe this is correct. However, reading between the lines, it seems you’re implying you’d need up to 20 different control charts to show the variation across different time bases. Unless your process is so unstable that it goes out of control daily, you should be able to use a longer time basis to show improvements. And, you should really show the improvement across a long time period to confirm the improvement is for real, and is maintained. In my company, a project must show sustained improvement for at least 3 months before we declare success. Using a weekly score to show improvement in what had been a monthly output may be a little premature.

    0
Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.