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.

What method to validate?

Discussion in 'Capability - Process, Machine, Gage …' started by Chris Glover, May 24, 2017.

  1. Chris Glover

    Chris Glover Active Member

    Joined:
    Aug 7, 2015
    Messages:
    54
    Likes Received:
    32
    Trophy Points:
    17
    We have a piece of equipment that will "kick-out" parts that fail..there is a load displacement spec measure by an LVDT.

    We are trying to validate that the test works as designed.
    30 (No Good) pieces were ran..and the data collected (the "measurements from the LVDT)

    I was not part of the test..just given the data and ask to provide a statistical study showing the equipment did what it was supposed to do (i did..it rejected all 30 parts..but someone was some stats to back it up)
    What would be the best test or method using this data?
     
    Chris Glover, May 24, 2017
    #1
  2. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    606
    Likes Received:
    664
    Trophy Points:
    92
    Location:
    Maine
    When given data from a flawed study design all you can do is tell the requester what his experiment died from...
    This study design doesn't give you any stats. If you tested 30 'bad' parts and it rejected them all then you only get a number: 100%
    In order to do any statistical analysis you need to have 'good' parts as well. Then you can do a cross tab with mcnemer's test or the less powerful kappa statistic.

    When using automatic testers it is important to understand both false rejections and false accepatances...
     
    Bev D, May 26, 2017
    #2
  3. Chris Glover

    Chris Glover Active Member

    Joined:
    Aug 7, 2015
    Messages:
    54
    Likes Received:
    32
    Trophy Points:
    17
    As I dug into this a bit deeper I found that only parts that were above the limits were tested - no acceptable parts not parts below the limits were used...
    This is what happens when you let engineers test their own processes...
     
    Chris Glover, May 31, 2017
    #3
    Candi1024 likes this.