iSixSigma

Z Bench Score and Sample size

Six Sigma – iSixSigma Forums Old Forums General Z Bench Score and Sample size

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #41047

    Patrick
    Participant

    We recently had a six sigma project done, the main goal was to improve the process capacity, so we don’t have to do final inspection anymore.
    After improvement, the new process capacity study shows that the new process has a 4.33 Z Bench Score,  the team decided to sample test every fifth product instead of testing every single product. But if some one asks why you decided to test every fifth instead of every tenth, we don’t have answer for that. I don’t know in this situation how to calculate SAMPLE SIZE.
    By the way, the test is Pass or Fail. Please help.
    Thanks
    Patrick

    0
    #128746

    Pablo Quintana
    Participant

    Patrick,We are in some sort of the same project. What I did is use the “Reduced” verification level. I use MIL-STD-1916 to calculate the sample sizes.For example if you use VLIII as your “Normal” level, then you should use VLII as the “reduced” and VLIV as the “tighter” level.That is a practical approach using a well known source.

    0
    #128749

    J Pilar Alcantar
    Participant

    Patrick,
    I recomend you to visit tha attached Web side, it has a lot of information for sampling (std-1916).
    http://www.variation.com/techlib/standard.html

    0
    #128762

    thevillageidiot
    Member

    Ok…I am confused…what does process capacity have to do with QC activity…I am guessing you meant capability?  And if you are using a pass/fail, why are you using Z bench, which is used for continuous data….forgive me if I misunderstood…..

    0
    #128778

    Craig
    Participant

    Patrick,
    Just to make sure I understand your situation….A Z-Bench score of 4.33 means that your projected defect rate is 7.455 PPM.  It took a little digging, but I found the definition of Z-Bench. 
    You can use JMP to generate an OC curve if you are doing acceptance sampling.  I would take a hard look at how you came up with the projected defect rate and also determine what the customer expectations are. (What PPM is acceptable to them? ) 
    If you are unsure about your true capability but you believe you have made a siginificant positive impact, be conservative and go to 50% sampling. Adjust the sampling based on your real pass/fail results. Hopefully you will rarely find a defect, and your capability study was accurate!  It is better to delay your internal benefit a little longer in order to protect the customer.
    Good luck.

    0
    #128868

    Rajesh Narasimhan
    Participant

    Patrick
    Take a look at this site which will give you how to calculate the sample size.http://www.ifad.org/gender/tools/hfs/anthropometry/ant_3.htm The following site will give you how many samples need to be inspected based on population size http://www.ifad.org/gender/tools/hfs/anthropometry/ant_3.htm. Also regarding the selection of fifth product for inspection, you are bascially using a systematic sampling.  In the systematic sampling once you know the sample size, you will then use the random number table to select the first sample and from there on you will start selecting the sample with a pre-defined number.  For example, if your sample size is 50 and population size is 1000.  You use the random number table to select the first sample let say 25.  The you can use a pre-defined number say 20. You will add 20 with 25 = 45, then you will pick the 45th one as your sample, likewise you keep on adding that till you reach the said sample size 50.  The formula for defining the pre-defined value (say K) is P (population) divided by the sample size (s).  If you really want to know much about sampling go thru this site:-http://www.mis.coventry.ac.uk/~nhunt/meths/listof.htm which will be useful for you understanding.

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

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