Page 1 of 1

Help regression linearity validation

Posted: Tue May 21, 2013 7:15 pm
by montana
Hi,

I need to study the matrix effect on an standard solution and plasmatic samples

For the first curve standards : y = 1348,22307 x -450,359967
For plasmatic curve: y = 1348,94668 x + 595,26428

My question is , is it ok that the intersept in standards curve négative -450 and for plasmatic is positive +595,26

Re: Help regression linearity validation

Posted: Wed May 22, 2013 3:42 am
by mattmullaney
My suggestion--answer is maybe. Look into performing the a statistical test on each calibration to determine if the y-intercepts are significantly different than zero. If both are, you then can be confident that you've no troubles...but the safest thing to do is to run say, six total curves under each manner of preparation and compare all of them statistically (F-test).

Hope that this helps, and good luck!

Re: Help regression linearity validation

Posted: Wed May 22, 2013 3:53 pm
by Prim
It would suggest a problem to me as the graph would look like a standard addition method :|

A negative intercept = the line of intercept crosses the y axes below y=0

As a first step, I would check if your intercept value is statistically different from zero. If it is then it would suggest a problem....

Ultimately you are saying a response of less than zero = a concentration

Good luck

Prim