deviation from linearity

[trentar]

hallo

how would you suggest to write function deviation from linearity? Just regress linear curve and write parameters with error or do you for engineering purposes use some more suffisticate term? If yes can you please define it!
thnx
r.

 

[btrueblood]

Depends. If I believe the function should be linear, then I use the first idea. But, when the underlying physics are non-linear, I might discuss why I think so, and what form the non-linear equation might have, if it's germane to do so (i.e. in discussing the limits of a linearized model).