 |
 |
 |
|
 |
|
 |
|
 |
 |
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
© Copyright 2009 AACC International.
Published January 2009.
Managing outliers in collaborative studies
D. Palmquist. USDA ARS, Peoria, IL
Palmquist, D. 2009. Managing outliers in collaborative studies. Online. AACC International Cereal Science Knowledge Database
A laboratory showing consistently higher or lower values than other laboratories over a range of materials in a collaborative study can be considered to have a bias. Laboratory bias can have misleading consequences when evaluating the efficacy of a method. The between-laboratory variation may be inflated and lead to the conclusion that a perfectly acceptable analytical method will be rejected. A simple technique for identifying outlier laboratories will be presented by ranking the laboratories’ mean results for each material tested in the collaborative. A laboratory having an overall ranking outside the table value range (based on number of materials tested as well as number of participating labs) will be considered to be an outlier whose results can then be removed from the statistical evaluation of a method.
©AACC International
- 3340 Pilot Knob Road - St. Paul, MN 55121 U.S.A.
Phone: +1.651.454.7250 or Fax: +1.651.454.0766
Privacy Policy | Disclaimer
| Copyright |
Contact AACC International
|
Webmaster