Fitting success estimator from lsqcurvefit.m

lsqcurvefit.m is attractive relative to nlinfit.m because of its support for ranges associated with the initial conditions / parameter estimates. nlinfit.m has a nice feature which is the output of a fitting success estimator which is the variance of the error term - the MSE.

The question is how can one programmatically collect or compute an MSE-like, goodness-of-fit estimator from the lsqcurvefit.m call? How can one know whether the fit operation failed, or converged on reasonable coefficient estimates in a programmatic sense?

Note that we are collecting all possible outputs produced by the lsqcurvefit.m function call: X, RESNORM, RESIDUAL, EXITFLAG, OUTPUT, LAMBDA, JACOBIAN

Note also that a contributor has suggested that certain outputs of the lsqcurvefit.m can be used as follows:

conf = nlparci(X, RESIDUAL, 'jacobian', JACOBIAN)