Forced to Zero

[syx]

When could we use 'forced to zero' for regression of calibration curve?

Thanks and regards,
Siswanto

[Mattias]

The accuracy investigation in the validation will show you what calibration mode is most suitable and in what range. (Accuracy = quantitation of spiked blank sample solutions).

[Jumpshooter]

Syx, use a "forced to zero" when you wish to prepare a calibration curve that has three standards (e.g. 0.1 mg/mL, 1 mg/mL, 10 mg/mL) and a reagent blank (e.g., containing the extraction solvents and some mobile phase solution--but NO analyte). The "reagent blank" will serve as your "zero" data point.

[JI2002]

"forcing (0, 0)" and "including (0, 0) are not same thing.

[rnelson]

I'm not sure I completely understand the question. Are you performing an analysis?

If you perform regression on a calibration curve, you should also get a number for intercept. Depending on how large that number is should tell you how close your line is to going through the origin (zero point). Whether or not that is acceptable would depend on the requirements of your application.

I'm currently struggling with some linearity questions myself related to calibration.

[Noser222]

You should not force to zero unless you know that your y-intercept is insignificant. For example, you run a standard curve and the y-intercept is less than 2% of the response at the target concentration of your analyte. In that case, forcing the intercept to zero would contribute no more than 2% error to your measurement. 2% is an example; whether or not you force the intercept to zero will depend on the level of accuracy and range that you need.

[HW Mueller]

syx, all you have to do not to be belligerent to a poor old straight line is to inject mobile phase, etc., a few times. . . . . . .