ISO 17025 Method Verification - One sided “F” test or two sided “F” test

tacom

Involved In Discussions
The Standard deviation obtained in house method verification study must be compared to precision found in the colloborative study. The aim of this study is to demostrated that the result produced by laboratory is acceptable (method verification). That mean my standart deviation must be smaller or equal than found in the colloborative study
So, I decided to use “F test . What should i do? One sided “F” test or two sided “F” test
Thaks in advance
 

itsbiodiversity

Involved In Discussions
Are you performing testing such as chemical analysis or are you performing calibrations? Are you trying to conform to 17025 (2005) Section 5.9? Are you familiar with En score and z scores? The most important is the En score.

The En score is calculated according to this formula: (Xi - Xref) / (SQROOT(ucsquared + ucrefsquared)) where Xi is your value, Xref is the "assigned" value, ucsquared is your uncertainty squared, and ucrefsquared is the "assigned" uncertainty.

Z score is calculated according to this formula: (Xi - Xref) / stdev for proficiency assessment.

The z score is commonly used, but it does not take measurement uncertainty in to account which can lead to "false" failures. For instance - some instruments like calipers have a resolution of 0.0005". If you have a value one digit of resolution different than the "assigned" value your z score may be magnitudes greater than is allowed. I would analyze any z score results that are out of tolerance.

Here is a great link that helped me along the way:
https://www.eurachem.org/images/stories/Guides/pdf/Eurachem_PT_Guide_2011.pdf
 
Top Bottom