Jump to content

  • Log in with Facebook Log in with Twitter Log in with Windows Live Log In with Google      Sign In   
  • Create Account

Submit your paper to J Biol Methods today!
- - - - -

Established Statistical Method for I.D.ing Multiple Outliers

  • Please log in to reply
1 reply to this topic

#1 hyphae



  • Members
  • Pip
  • 3 posts

Posted 06 July 2010 - 07:29 AM

When we do radioactive work, our data can be pretty scattered. We have an internal policy to eliminate up to 1/3 of the data, but we just eye-ball the bad points. We want a systematic, and more importantly, a verified (or at least an established) method to eliminate the outliers. Most methods (Q-test, etc. ) only eliminate one point, but I did discover the Peirce's Criterion that is supposedly a valid way to identify and eliminate multiple points.

However, I can't see too many people using this method (there is a paper from NASA that talks about the virtues of this method, but the boss wants to see the method used in biotech, or at least in radiation work). If anyone uses that method, or knows of another method that is commonly used for multiple outliers, please let me know!

#2 DRT



  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 176 posts

Posted 06 July 2010 - 03:41 PM

Would an alternate approach, given that you are loosing a third of your data, be to look for a transformation to normalise the data? That way you wouldn’t need to eliminate outliers.

Home - About - Terms of Service - Privacy - Contact Us

©1999-2013 Protocol Online, All rights reserved.