Readers,
In chemical or pharmaceutical analysis and instrument calibration where
chemical or drug reference standards and instrument calibrators (standards) are
used, the question is stated as many reference standards or calibrators are
needed. In these situations, the analysis is of materials that fall close to
and about a nominal or target value, that associated with the product.
The question is then: how many calibrators or reference standards (points or
x-y pairs) are needed to prepare a calibration or standard line? Expressed
somewhat differently: What is gained by increasing the number of points? Or
lost by decreasing the number of points? Common practice is to use five points
for least-squares regression lines (instrument calibration or analytical
standard lines), the typical pharmaceutical industry practice with the exception
of clinical studies.
My answer is: The number of points that minimizes the variance surrounding
the line at an amount of calibrator or reference standard of most interest.
This answer is general and applies to any use of a least-squares regression
line, for example, quantitating analytes from a standard line or calibrating an
instrument. In chemical and pharmaceutical analysis, it is the target or
nominal value of the product to be analyzed.
Regards,
Stan Alekman
Let charm be thy spear, wit thine armour.
************************************** See what's new at http://www.aol.com
|