Dear all,
I have a question concerning the size of model space in DCM. Let's assume I'd have 80 000 a priori equally likely models. Would it be suitable to test them all (barring the time the calculations would take) to be able to subsequently do some family comparisons?
Further, would it eventually be struggling or improper to interpret the result of two families with unequal amount of members, e.g. in family #1 you compare 50% of all your models with the other half, but in family #2 all models will be divided in 70% of one category and 30% of the other. In the comparison of familiy #2 the 30% category wins. Following this, would it be possible, that the majority of winning models of family #2 do not correspond to the category of winning models in family #1? How would one deal with that?
Thank you very much in advance for your response.
Kind regards
Cris
|