
 Posts: 269
 Joined: Fri Jun 29, 2007 1:48 pm
 Location: Atlantic Beach, FL
I've got a method where I need to use quadratic regression as the fit type for a calibration. A 5 point calibration curve is generated using the following points:
Code: Select all
Concentration (ug/mL) Peak Area
50 396287
71 717880
107 1363599
125 1757618
143 2247428
If I setup a quadratic fit in Empower, I generate the following equation for a best fit line:
79.2x^2 + 4400x  11600
If I use Excel to generate an equation for the best fit line, the following is produced:
82.407x^2 + 3874.2x + 5919.9
I get the closeness of results on both the x^2 and x constant. I am confused at the difference in intercept as well as the total sign change. Clearly, results calculated using the two different approaches are not exactly the same; or even close for that matter.
For what it is worth, I can calculate the regression line on my TI89, by hand, using Maple, and using Mathematica and I get the same equation that Empower generated.
My question is: why is there a difference? What is Empower et al doing differently that Excel is not?
TIA...