How I calculate Standard Deviation

Six Sigma – iSixSigma Forums Old Forums General How I calculate Standard Deviation

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
  • #50090


    How do I calculate the standard deviation when the USL and the LSL are known ? I would like To see an example if someone has one and is willing to share. This is a new experience for me. 
    Thanks for any help,


    B Kumar

          In this case, do you have sigma level. If you know the Sigma level, you can calculate sigma value as below.
    Sigma Value = (USL – LSL)/ Sigma Level.
    If your data is continuous data
    Sigma Value  = (UCL – Mean)/3;
    or Sigma Value = (Mean – LCL)/3
    UCL and LCL are Upper and lower Control limits.
    Suppose, if your data is Attribute data, then
    Sigma Value = NORMSINV [ (1 – DPO) +1.5 ]
    DPO = Number of defects / Total opportunities for eror.
    So, generally USL – LSL = Sigma Level * Sigma Value.
    Ravi B                  



    sandrine berthet

    Not enough
    You have to know the Cp to use the equation and to figure the sd


    aiman farid

    You can’t calculate the standard deviation when only the USL and the LSL are known. As the USL and the LSL are specification spread.

    To calculate standard deviation you need real data or the UCL and the LCL as actual  process spread.

    Or, when you know the relationship between both (process and specification spread)

    The USL and the LSL + capability index Cp (no shift) or Cpk (shifted)

    The USL and the LSL + Sigma level (number of standard deviations lying between the mean and the nearest specification limit)
    Aiman Farid


    Bill Fowlkes

    Ravi, this is so wrong.  If it was true that sigma = (mean – LSL)/3, or better sigma = (USL – LSL)/6, then all Cp would equal 1 and everyone would have the same defect rates.  Of course that is not true.  
    If fact, the USL and LSL should be related to customer requirements, and the sigma to manufacturing varaibility.  So, the correct answer is that you CAN NOT calculate the standard deviation from the spec limits.  Those are two very different things. 

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

The forum ‘General’ is closed to new topics and replies.