By Anonymous on Tuesday, October 1, 2002 - 02:46 pm:
Please excuse my ignorance,but what does it mean to weight a standard curve? I am looking at a standard curve that is theoretically linear, but if I look at the software, it says that there is a 1/X weighting applied. I am not familiar with this kind of data handling. Can anyone help?
By Tom M. on Tuesday, October 1, 2002 - 05:41 pm:
http://www.lcresources.com/discus/messages/2401/2367.html?MondayApril2220020506am
Sorry, it's a long thread.
By Anonymous on Wednesday, October 2, 2002 - 05:50 am:
Thanks for the help. Did Marcelo ever start a differnt thread about this issue?
By vestelshirley on Friday, October 11, 2002 - 07:09 am:
I had always thought that 1/x, etcetera, weighting was a way to force a linear relationship where none existed - somewhat like the physicists going through elaborate fixes so they could take a differential equation and reduce it to a first order differential equation. I could be naively wrong since I have been before ... many times.
By chris on Friday, October 11, 2002 - 07:46 am:
The link that was mentioned above was very helpful information. It turns out that larger deviations at higher concentrations tend to influence the linea more than smaller deviations at smaller concentrations. Therefore, the accuracy at the lower end is impaired.
Since I originally posted this question, I found a paper that explains it and goes over its uses:
Almeida, A.M., Castel-Branco, M.M. and Falcao, A.C., Journal of Chromatography B, 774(2002),215-222. Linear regression for calibration lines revisited: weighting schemes for bioanalytical methods.