Let F1 and F2 be two competing models proposed for the interpretation of the seismic catalog regarding a given site. Which one is more reliable than the other?
If the test bench is the consistency of the models with the catalog, the scarcity of data prevents meaningful comparisons. In what follows it is shown that a different test bench (however interesting from the engineering point of view) opens new statistical perspectives. A few methods aiming at a reduction of the epistemic uncertainty in the choice between F1 and F2 are described.
This paper is available in Italian only.