Currently, only the first order approximation is implemented. The second order approximation is still very easy to compute and may be slightly more accurate, although it still suffers from the general incorrect assumption of a normal distribution. The second order approximation essentially assumes independence of the estimators of a and b.
Currently, only the first order approximation is implemented. The second order approximation is still very easy to compute and may be slightly more accurate, although it still suffers from the general incorrect assumption of a normal distribution. The second order approximation essentially assumes independence of the estimators of a and b.