iSixSigma

Determining correct sample size

Six Sigma – iSixSigma › Forums › Old Forums › General › Determining correct sample size

Viewing 16 posts - 1 through 16 (of 16 total)
  • Author
    Posts
  • April 19, 2002 at 9:44 pm #29284

    Miguel Camargo
    ★ 20 Years ★
    Participant

    Hi all,We want to know what percentage of the rejects are caused because of assembly issues (the failures because of assembly have the same symptoms than the ones failing because of some other reason). However, we need to perform a failure analysis that consumes a lot of time. So that, we would like to analyze the rejects for several lots as a way to get a picture of the process. Take in mind that there is a lot variation in the process, so that we need to analyze as much lots as we can & we would like the sample to be representative for all the working shifts.If there any statistical methodology for calculating this?Any comment is welcome,

    0
    April 23, 2002 at 12:13 pm #74673

    Sridhar
    ★ 10 Years ★
    Member

    Hi Camargo,
           Sample size can be determined based on the following three factros
    1. size of population
    2. margin of error
    3. probability of failure or sucess
        There is a simple formula to calculate the sample size. Send me your mail id i will send you one excel sheet which calculates the sample size if you give the paratmeters whatever required.
    thanks
    A.Sridhar
     

    0
    April 23, 2002 at 12:44 pm #74675

    Corina
    ★ 20 Years ★
    Participant

    Could you be so kind as to send me the Excel file as well? It would be very helpful…
    My e-mail address is: [email protected]
    Thanks much!
    Corina

    0
    April 23, 2002 at 12:48 pm #74676

    abasu
    ★ 10 Years ★
    Participant

    Assuming you have a normal distribution you can estimate an appropriate sample size using the formula below
    n = Z^2*s^2/d^2 where
    Z = sigma level of the confidence (1.96 for 95% confidence level)
    s = standard deviation of your process and
    d = the difference you want to detect
    If your data is binomial
    n = Z^2*p*(1-p)/d^2 where
    Z = sigma level of confidence (1.96 for 95% confidence)
    p = proportion or failure rate and
    d = the difference you want to detect

    0
    April 23, 2002 at 1:18 pm #74677

    john beaudoin
    ★ 20 Years ★
    Participant

    My e-mail is [email protected]
    Could you please forward me a copy as well?

    0
    April 23, 2002 at 1:22 pm #74678

    john beaudoin
    ★ 20 Years ★
    Participant

    You are exactly right, and have also shown why sample size hangs up a lot of individuals.  There is no real simple way to determine sample size.  However, I have heard that there are software solutions to such.  Are you doing these calculations by hand (calculator) or on a spreadsheet?  Do you know of any software such as Excel addins or PDA (Palm Pilot) software, etc. that would aid in this.  We picked up a lot of documentation on ANSI standards with a slide rule type calculator and some other tables and things, but wanted something a little more solid.

    0
    April 23, 2002 at 3:13 pm #74687

    abasu
    ★ 10 Years ★
    Participant

    Prostats for palm has sample size calculator.  In addition I use an excel worksheet to calculate sample size.

    0
    April 23, 2002 at 10:19 pm #74739

    Miguel Camargo
    ★ 20 Years ★
    Participant

    Hi,Here is my e-mail address:[email protected]

    0
    April 24, 2002 at 4:13 pm #74769

    JJS
    ★ 20 Years ★
    Participant

    Please send me a copy
    my email [email protected]
     
    thanks
     
    JJS

    0
    April 24, 2002 at 4:22 pm #74770

    Ted J. Loeffelholz
    ★ 20 Years ★
    Member

    Be careful with your sample selection.  I noticed that you mentioned lots and shifts.  Do you know for sure that your failures are not lot or shift dependent?  If you don’t then you will probably want to make sure you know if your manufacturing process is a fully nested process and pick your samples accordingly.  This would give you more information as to where in your process the variation is occurring.

    0
    April 24, 2002 at 7:43 pm #74786

    CAconsultant
    ★ 20 Years ★
    Participant

    A simple formula to use would be n x p >5, where n = sample size and p=current proportion defective. 
    First calculate your current % defective. (total rejects/total produced).  Plug that number in for p and solve for n.
    Once to have that you can develop subgrouping strategies and use X-bar R control charts to sample across days, shifts, operators, etc.
    Hope this helps.
    Doug

    0
    April 24, 2002 at 7:53 pm #74788

    DC
    ★ 20 Years ★
    Participant

    This is a popular excel file!  Could I get a copy?
    [email protected]
     
    Thanks!

    0
    August 14, 2002 at 2:14 am #78074

    Saherngu
    ★ 10 Years ★
    Participant

    Hi, Sridhar. I’ll appreciate if you can send me a copy. Email [email protected]. Thanks & Regards
    James

    0
    August 14, 2002 at 9:37 am #78076

    A. Jerez
    ★ 10 Years ★
    Participant

    My e-mail add is [email protected] yahoo.com.
    I am interested with your excel file. Kindly send me a copy as well..
    Thank You and More Power!
    >>TOXICO<<

    0
    August 14, 2002 at 9:39 am #78077

    A. Jerez
    ★ 10 Years ★
    Participant

    Kindly send me a copy of your excel file as well…
    E-MAIL ADD:   [email protected]..
    Thank You and More power!

    0
    August 14, 2002 at 11:02 am #78079

    vidya Kulkarni
    ★ 10 Years ★
    Member

    Hi Sridhar,
    Iam also interested to know how to determine the correct sample size.
    Pl. mail me excel sheet too.
    my mail ID is [email protected]
     
    vidya

    0
  • Author
    Posts
Viewing 16 posts - 1 through 16 (of 16 total)

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