The model solutions to 16.1 list incorrect dependencies and statistical distributions as sources of parameter error.
Yet these are listed under Model Uncertainty in chapter 4 (page 31).
Can we please get a clarification about these two?
If the actual model is all about fitting a curve, the statistical distribution selection per-se is the model/method and therefore this is a source of model error. This is even mentioned in 1.1 of chapter 16 therefore I trust it is best to consider incorrect choice of statistical distribution as a source of model rather than parameter error.
On the other hand if we are using some form of Bayesian analysis, the selection of the priori's stat distribution can be considered as a parameter error?
Last edited by a moderator: Jan 30, 2015