I use non linear curve fitting in both versions 6.1 and 8. When I fit the same data to the same fitting function, I get about the same fitting parameters, as i should. But the calculated errors are in v8 substantially high, generally one order of magnitude higher than the fit parameter, whereas the calculated errors in 6.1 are generally one order of magnitude lower than the fit parameter.
So how come the results differ so much, and what can I do to lower the calculated error in v8.
Please send your project with data and 6.1 result and 8.0 result and what function you used for fitting. If it was a user-defined function please include that. Send these to tech support so they could look at what is causing the discrepancy.