iSixSigma

How to Determine Sample Size

Six Sigma – iSixSigma › Forums › Old Forums › General › How to Determine Sample Size

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • June 9, 2006 at 4:09 pm #43679

    Manu
    ★★★★★★★★★★
    Participant

    I have 5 boxes on Item A ( each box contain 1500pieces) and another 4 boxes of Item B(850 pieces in each box). I am trying to design an experiment with combinations from both Items A & B. Total 20 combinations. An assembly of one piece from A and one piece from B can be considered as a sample. I would like to know how many samples(assemblies) from each combination I should take so that I can statistically prove that there is no variation between the samples. I have a way of measuring the strength of each assembly. So there is going to be some value(in Newtons) which the machine displays when I load the assembly to the machine.
    Thanks
    Manu

    0
    June 9, 2006 at 5:12 pm #138880

    Schuette
    ★★★★★★★★★★
    Participant

    Manu – You want to know the sample size necessary to show that there is an acceptable amount of variation – if you have no variation you just have an incapable measurement system.
    You need to specify what amount of variation you need to be able to detect, and the power you want to assign the test.  Plug the numbers into Minitab or other program, and you will get your sample size recommendation.

    0
    June 9, 2006 at 7:45 pm #138887

    Manu
    ★★★★★★★★★★
    Participant

    Thanks Jim.
    One more question. Which test shall I use. 2 sample T test or 2 Proportion Test.
    Manu

    0
    June 9, 2006 at 7:53 pm #138888

    Patrick
    ★★★★★★★★★★
    Participant

    Hi Manu,
    Since your measurement is a continuous variable, I’m wondering if the following would work:
    1. Determine the standard deviation (SD) of the sample measurements you’ve collected thus far.
    2. Determine the degree of desired precision (P) for your assessment purposes (e.g., to the nearest 1 Newton? 5 Newtons? 10?)
    3. Plug this information into the following formula:

    Required sample size = (2*SD/P)^2 (that’s 2 times the standard deviation, divided by the desired precision, quantity squared).
    We used this method in a recent Six Sigma project when we were collecting baseline data on patient readiness times.
    Hope this helps!
    Patrick

    0
    June 9, 2006 at 8:23 pm #138890

    Mighty Mike
    ★★★★★★★★★★
    Participant

    Manu,
    Use a 2 Sample T test to prove your hypothesis…
    Mighty Mike

    0
    June 10, 2006 at 2:24 am #138894

    Ritz
    ★★★★★★★★★★
    Member

    Mighty Mike –
    Did you just state “prove your hypothesis”?  Shame!

    0
  • Author
    Posts
Viewing 6 posts - 1 through 6 (of 6 total)

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