Discussion:
verification of a calibration
(too old to reply)
tancho
2008-03-20 20:33:41 UTC
Permalink
Due to the actual workload in our lab, it is impossible to perform
every day a 4 point calibration for 10 different cations. Therefore we
want to check the calibrationcurve with 1 standard solution. But what
is the max allowed deviation between obtained result and the
theoretical valuean how can I determin/validate this deviation.


Thanks.
David Stone
2008-03-20 20:49:09 UTC
Permalink
In article
Post by tancho
Due to the actual workload in our lab, it is impossible to perform
every day a 4 point calibration for 10 different cations. Therefore we
want to check the calibrationcurve with 1 standard solution. But what
is the max allowed deviation between obtained result and the
theoretical valuean how can I determin/validate this deviation.
I haven't seen any recommended values. What I would be tempted
to do, is plot the concentration* derived via the stored calibration
curve for a check solution on a control chart, with the warning
and action limits set according to the 95% confidence limits
for the interpolated value as measured on day 1.

That way, you'd have some indication of whether there was drift,
or variations were due solely to random errors in the actual
day-to-day measurements. I'd also run a blank the same way.

F- and t-tests would give you statistical tests for whether the
measurements on your check solution were significantly different
from the starting values.

* Actually, probably better to do this on the measured value,
rather than the derived concentration i.e. absorbance, peak
area, etc.
n***@gmail.com
2008-03-24 00:38:25 UTC
Permalink
I had the same problem in my lab, and I resolve it changing and
validating a new method of quantification. In my case, pesticides
response in GC is a real problem and calibration curves and statical
analysis arent quite fiable.

Day to day calibration is the best way of course but I resolved that
with the inclusion of an internal standard.

Good luck
Marvin
2008-03-24 15:31:28 UTC
Permalink
Post by tancho
Due to the actual workload in our lab, it is impossible to perform
every day a 4 point calibration for 10 different cations. Therefore we
want to check the calibrationcurve with 1 standard solution. But what
is the max allowed deviation between obtained result and the
theoretical valuean how can I determin/validate this deviation.
Thanks.
There is no global answer. For some applications (e.g.,
clinical chemistry) there are regulations that need to be
followed. If you are not covered by such regulations, you
need to set the parameters based on the acceptable
analytical error.

Loading...