iSixSigma

link between confidence interval Ppk

Six Sigma – iSixSigma › Forums › Old Forums › General › link between confidence interval Ppk

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • August 13, 2008 at 10:50 am #50738

    Sorour
    ★ 10 Years ★
    Participant

    Hi
    Is it possible to calculate the Ppk for a corresponding confidence interval and sample size?
    For example for a confidence interval of 95/99.9 the Ppk for 30samples is 1.34 (using aql)
     
    Are there tables or a calculation I can use to develop Ppk for 95/99.9 for different sample sizes and different confidence intervals? E.g. 9 17 21 etc.
     
    Thank you
    Paul

     

    0
    August 13, 2008 at 11:02 am #174832

    Ron
    ★ 10 Years ★
    Member

    Quick answer is no.
    Ppk is a long term process capability typically with 100 samples taken or more.
    Confidence Intervals are the probability that a specifc value will fall within this range.
    Ppk is very specfic  Confidence intervals are general.
     

    0
    August 13, 2008 at 12:05 pm #174833

    The Kid
    ★ 10 Years ★
    Member

    Hi Paul,
      I am not 100% sure about this, but I think some time ago I did read some articles describing this. I think the outcome of the study was a look up table with sample size and confidence interval for CPK values.
    So I would suggest you do some googling on the same. Would be great if forum experts can provide some insight on this.
    Cheers,
    The kid

    0
    August 13, 2008 at 5:04 pm #174843

    Sorour
    ★ 10 Years ★
    Participant

    Hi Guys.
    I was talking to a work colleague. He said you can determine the Ppk by using your confidence intervals and k value. Simply by using the confidence formula 95/999 = mean –k X sd and LSL- mean divided by 3 x sd = Ppk you can work out that the Ppk is 1/3 the k value.
    Don’t know if anybody heard this before or if there some data behind it. I would be interested to hear your comments.
     
    Secondly, typically what would be the smallest sample size you would use to work out Cpk/Ppk? I am recommending a minimum of 15 but ideally 30.
    Thanks again
    Very much appreciated
    Paul

    0
    August 15, 2008 at 5:10 pm #174912

    Mikel
    ★ 10 Years ★
    Member

    Pp can be related to the tolerance interval which is a containment interval of individual values at a given level of confidence. E.g., a 95/95 tolerance interval contains 95% of the expected values at a 95% confidence level. Tolerance Intervals are sensitive to departures from normality so transformation is required of most data.

    0
  • Author
    Posts
Viewing 5 posts - 1 through 5 (of 5 total)

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