Hi all,
The LV curve fitting functions output a covariance matrix, which I would like to use to estimate the error on the fitted parameter values in different linear and nonlinear fits. At first I thought the Vii matrix elements correspond to the variance = (error)^2, but some of the values were ridiculously low. I tested a linear fit to a straight line on some dummy data, found the slope and its variance from the covariance matrix, multiplied all the Y values by two, and got double the slope but the same variances. This would indicate that the covariance matrix is normalized by the parameter values.
Now I have tried a few different generalized linear and LM nonlinear fits, and cannot comprehend the results. The results for the diagonal elements of the covariance do not seem to correspond at all to the error except in the case of the coefficients of linear terms. Does anyone know what is actually being calculated in the covariance matrix and/or how it relates to the error in the parameter values?
thanks,
Dan
