ABSTRACT
The study of relations between variables is fundamental in every branch of
science and, in this respect, Statistics is no exception. A n important distinc-
tion, however, must be made from the outset: while in disciplines like, for
instance, Physics and Engineering these relations generally have an intrinsic
cause-effect meaning, it may not necessarily be so in Statistics. Let us con-
sider an example. When an engineer writes the equation F = kx for an elastic
spring of constant he/she is expressing a well-defined cause-effect linear
relation between the force F applied to the spring and its elongation x (with
respect to its rest position). If, however, we obtain a statistically satisfactory
linear relation between, say, the number of new-born babies in Italy and
the number of New Yorkers who quit smoking over the last 10 years, it
would be rather hard to believe that this linear relationship has some intrin-
sic meaning. If, nonetheless, somebody is willing to assume that a meaning
does exist, he/she must look for it outside the realm of Statistics.