Dear All,
I think this is a general stat plus R-related question.
Firstly, the 'problem': I fitted non-linear (exponential-type) models to
moisture ratio-time data using nls(); one of its parameters, k, is commonly
designated as the constant of drying rate (the other are less important); in
the literature, several authors calculate the diffusivity, D, from k,
knowing that k=(D*pi^2)/(4*L^2) (L is the known length of the material) and
present it as one value without any info on precision.
Finally, the question: using info from the nls() fit (namely the vcov
matrix), is it reasonable to use the delta method (as in alr3 or car
packages) to obtain D AND its SE(D)?
Thanks in advance,
Eduardo
[[alternative HTML version deleted]]