1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

Sample Size Justification

Discussion in 'Sampling, Standards and Inspection' started by TheWaeller, Mar 23, 2016.

  1. TheWaeller

    TheWaeller Member

    Joined:
    Jan 13, 2016
    Messages:
    6
    Likes Received:
    0
    Trophy Points:
    1
    Hi all,

    I'm looking for a guidance to justify sample size.

    Actually , the assumption is that the data sets (metric data) are following a student-t-distribution. Then based on a T/s
    (Tolerance range above standard deviation)-ratio the sample size is calculated.

    This approach is being used for design verification. Usually max.sample size is 60 in accordance with ISO 11608-1.

    What about if the assumption will not be true? Do Ineed to go to a distribution free approch?

    Thanks for any hint
    Waeller
     
    TheWaeller, Mar 23, 2016
    #1
  2. Vthouta

    Vthouta Member

    Joined:
    Dec 20, 2015
    Messages:
    10
    Likes Received:
    2
    Trophy Points:
    2
    You can justify using the level of significance based on your current problem definition. This would lead to type-1 and type 2 errors. I just got to know effect size could be used to justify, but I have not applied yet to any of my problem definitions. I am not sure whether I answered your question.

    Link:https://en.wikipedia.org/wiki/Effect_size
     
    Vthouta, Mar 29, 2016
    #2
  3. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    606
    Likes Received:
    664
    Trophy Points:
    92
    Location:
    Maine
    Assumptions are tough. the t-distribution is fairly robust tho. you could perform a goodness of fit test. But this is all about justifying a sample size based on statistical theory to who?
    if it's to your organization, I might suggest a better way depending on the risk (severity of effect) of the product you are validating. If it's to a regulatory agency, you are probably OK as is. (depending on your reviewer)

    I would also recommend that you read the following article on tolerance intervals by Donald Wheeler.
     
    Bev D, Mar 29, 2016
    #3
    TheWaeller likes this.
  4. TheWaeller

    TheWaeller Member

    Joined:
    Jan 13, 2016
    Messages:
    6
    Likes Received:
    0
    Trophy Points:
    1
    Hi Bev,

    thank you for your help.
    The article of Don Wheeler is of of great help.
    Unfortunately, it is not yet available in his reading room.

    Thanks
    Waeller
     
    TheWaeller, Apr 11, 2016
    #4