Applications of probabilistic seismic hazard analysis demand the adopt
ion of a model (i.e. of the forms of a certain number of correlations
and probabilistic distributions) and the estimate of the parameters of
the model. As a measure of uncertainty in the calculation of the expe
cted value of a given quantity (for instance the peak ground accelerat
ion corresponding to a given return period at a given site) a coeffici
ent of variation is frequently adopted, which is intended to include u
ncertainties due to both the choice of the model and the estimate of p
arameters. The following three statements are illustrated in this pape
r: (1) in theory, the use of a coefficient of variation, when uncertai
nties in modeling are involved, is not correct, (2) in practice, the a
foresaid use can lead to unreliable results and (3) the analysis of un
certainties can be carried out in a more satisfactory way if uncertain
ties in modeling and uncertainties in the estimate of parameters are c
onsidered separately and viith different approaches. (C) 1998 Elsevier
Science Ltd. All rights reserved.