Minitab Capability Analysis – Sigma Shift

Six Sigma – iSixSigma Forums General Forums Tools & Templates Minitab Capability Analysis – Sigma Shift

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
  • #54185


    Is it possible that an experienced Six Sigma professional could answer my question?

    My understanding regarding Sigma Shift is as following:

    If you have collected Long term data then:
    1) You can work out long term Sigma level and label it ZLT.
    2) You can convert your long term data into PPM or DPMO.

    Furthermore you can work out the short term performance of the process, to see how good process could be:

    1) You can find short term performance of your process from within long term data (Minitab Capability Analysis).
    2) You can then convert short term performance into sigma level and work out sigma shift. (Zst ย– Zlt).

    If you have collected short term data then:
    1) You can work out your sigma level, make sure you label is Zst, so you know it represents short term.
    2) Do NOT multiply short term into PPM or DPMO as these are for long term only.
    3) Subtract 1.5 Sigma in order to ESTIMATE Long term Zlt.
    4) Ultimately look to collect Long Term Data.

    That being said does Minitab assume that the data used is LONG TERM DATA, the Capability Graph would then provide Observed, Ex Within and Overall and Potential results.


    Joel Smith

    @frym You have mostly described things correctly, although I don’t know that you’re doing anything terrible by converting short term sigma into PPM or DPMO as long as you know how to correctly interpret them (after all, they’re just a conversion from sigma anyway which you’re already comfortable using).

    To answer your question about Minitab, it doesn’t know how the data were gathered and whether they truly represent long-term data or not. So it assumes long-term data, and if your data were short-term then you would treat the reported long-term statistics as short-term (although either should work…if your data were short-term then there shouldn’t be much difference between the “long-term” and “short-term” stats).

    As for the 1.5 sigma shift, there are many articles and forum posts on the subject. In short, there’s nothing magic about 1.5 and some processes may show no shift while others may have a 5 sigma shift. 1.5 was a guess based on experience but is highly variable. Rather than re-hash the whole topic, I’ll leave it at that and you can either find some articles on it or perhaps someone else not already weary of the topic will type about it here.



    Others may disagree but IMH the basis for the shift and the consensus on using 1.5 is statistical esoterica. Unless you’re doing “rocket science” short term data is sufficient


    Mike Carnell

    @straydog What have you developed your IMHO (it needs an O on the end) on?



    @frym – 1.5 is fantasy, and as Joel said, different processes may have higher or lower levels. The issue is determining what is “short” vs. “long” term. Theoretically, you can never have truly “long term” data, as there may always be some other cyclical factor (Halley’s commet anyone?) that could impact your output.
    You should always try to account for all reasonable sources of variation, and don’t worry about “shifting.” all that using a shift does is let you fool yourself that you’re better than you think you are (see the z table – find 4.5 and 6.0 and you’ll see that what most call 6 sigma is really only 4.5, this is due to the mythical 1.5 sigma shift).

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.