For a general class of order selection criteria, we establish analytic and
non-asymptotic evaluations of both the underfitting and overfitting sets of
selected models. These evaluations are further specified in various situat
ions including regressions and autoregressions with finite or infinite vari
ances. We also show how upper bounds for the misfitting probabilities and h
ence conditions ensuring the weak consistency can be derived from the given
evaluations. Moreover, it is demonstrated how these evaluations, combined
with a law of the iterated logarithm for some relevant statistic, can provi
de conditions ensuring the strong consistency of the model selection criter
ion used. (C) 1999 Academic Press.