ABSTRACT
Me thods for testing the val idi ty of models are also standard exer-
cises i n the literature, for example, residual analysis (Chatfield, 1989).
The same problems of the appl icabi l i ty of the strict condit ions which
were referred to above i n connection wi th parameter estimation
apply also to these methods. A visual investigation is, therefore,
also necessary i n this context. In fact, it may often be sufficient for
practical purposes. Fur thermore, it must be appreciated that, even
if a part icular model passes a residual analysis test, it does not prove
that there is not some other model which w o u l d give an even better
fit i f tested similarly.