|
This article builds upon the fundamentals of ZScores discussed
in the March 2004 eNewsletter. For more information on
calculating ZScores follow this link to the March 2004
ZScore Article, "I Know I Passed
But How Well Did I Do?"
Let's face it, what's important is whether you passed or failed. Many
laboratories believe that all that is important is whether their
PT results were "acceptable" or "unacceptable". For accreditation,
that is certainly true, however, being "acceptable" really does not
tell you or your customers much about the quality of your results.
The State / USEPA Proficiency Testing limits are so wide that
they do not help you understand the actual quality of your results.
Acceptable PT results are a license to do business - not a definition of quality
Your customers (data users) want you to assure them that the
numbers you are generating are the same that a regulatory agency
laboratory would generate if they split a sample with you. What
your customers want is data comparability. Under the United States' current
system, laboratories which have a less than 5% chance of being
right can retain their accreditation in all 50 states, including the
NELAC states.
If data comparability is quality, how can you determine your
level of data comparability? Easy, look at your ZScore. APG is the
only US PT provider that reports this critical measure of quality.
We have been reporting ZScores since 1982. The ZScore is a measure
of the distance your result was from the Mean of the data. Since the
Mean is the most common result, how close you are to the Mean is
a measure of how your data would compare to other laboratories. Your
result will be judged "unacceptable" if your ZScore is greater than
3 for WP or greater than 2 for WS.
For WP, this means that there is less than a 1% chance that the results are
right and all the other laboratories are wrong. Similarly, you will
receive a "check for error" if your ZScore is between 2 and 3. At a ZScore
of 2 there is less than a 5% chance you are right but all states accept
this number.
So what's a great ZScore?
APG believes that a ZScore of less than 0.3 is a true measure of
analytical excellence. Only 12% of all laboratories can achieve this
level of data comparability. However, ZScores below 1 are good for
most laboratories. Your goal should be to receive less than 1 for
all tests run in your laboratory.
Now you know what a ZScore is and why APG reports them with every
study.
|
