Jan,

I'm not an expert on this but the answer seems to me to be that you should go with the model with the simplest parametrization because as you say "data I have to compare these different mechanisms is a special case for the common part of all models (not related to the mechanism I'm interested in): both rate parameters have to be equal and the third parameter has to be zero."

Other than that, in the future I guess you may want to apply a full parametrization to new data that do not fall in the special case you describe.

Hth,

cheers,

SCM

Dear SCM,

Thanks for the answer. What you say agrees with my intuition,  especially that application of the full model to the general situation will be necessary. 

My concern about the special case data is that when I apply the full model, some of the noise in the data will be pushed into uncertainties of the additional parameters, which may be good for looking at the parameters of interest.  On the other hand it inflates the parameter space leading to practical difficulties in model comparison and possibly reduces the power (?)

Best regards

Jan

Hi,

From what you describe (and without details of the models), I think you could use the full model rather than the reduced one and incorporate the knowledge you have about the system into your prior, if you then wish to test these models on new data. i.e. if in a preliminary experiment a parameter is measured to be zero, set the prior over this parameter to 0. You could change this to a normal prior centred at 0 if you want to account for the uncertainty in the parameter.

Cheers,

Adam