Beware of commonly used approximations. Part II. Estimating systematic biases in the best-fit parameters