Although it is often claimed that statistical techniques are ways of lettin
g the objective data speak for themselves, in both the contrast and correla
tional modes of statistical inference, all the real work is done by the a p
riori decisions imported into the analysis-which categories are to be used
to create contrasting populations, which categories are to be measured, whi
ch categories are to be held constant while others are compared, and which
is cause and which is effect? The authors explore here the problem of direc
tionality of causation and the relationship between cause and effect, on th
e one hand, and dependent and independent variables, on the other. In syste
ms of any complexity there are feedbacks-negative and positive feedbacks fo
rming loops, embedded in larger contexts and subject to influences that can
impinge on the loop at any point, such that the same pair of variables may
show positive correlations in some situations and negative correlations in
others.