Calibration curve linearity validation

[dibbadijam]

A question on confirmation of calibration curve verification.
In the past i have used r2 value greater than 0.999 to confirm linearity. I have received the follow statement from a cilent.

"The acceptance of the model must be verified by evaluating the relative bias between the back-calculated and nominal concentrations of the calibration standards. The use of the correlation coefficient is not recommended for confirmation of the regression model."

I am not sure how to apply this. i assume the following:
say i used nominal concenration of 2, 4, 6, 8 and 10 µg/ml stds and plot these to obtain the responses. Then re-plot with actual concentrations based on weight taken, i.e. 2.2, 4.1, 6.3, 8.1 and 10.0 µg/ml. Then compare both responses, expressed as a %.

[aceto_81]

You are correct in your interpretation of the statement, it is exactly where they ask for.

r2 isn't a good tool for confirmation of linearity. Do some search on the internet (lack of fit test for example) and you will found out why.

Ace

[WK]

Hi,
Try reading the chapter on instrumental regression in this:

Statistics and Chemometrics for Analytical Chemistry J. N. Miller, J. C. Miller

WK