[From Bruce Abbott (980622.0915 EST)]
i.kurtzer (980622.0200) --
this goes out to those math whizes out there. If A is correlated with B and B
is correlated with C, must A be correlated with C? Why?
No. A may be exerting a direct influence on C that opposes the direct
influence of B on C. A's direct and indirect effects on C cancel out. B in
this case is termed a "suppressor variable." You are already familiar with
the somewhat different case in which A (disturbance variable) is correlated
with C (output variable) and yet, in the ideal case, neither A nor C are
correlated with B (controlled variable), through which the influence of A on
C apparently must pass, the explanation having to do with negative feedback
and the circular loop of causality.
Why do you ask?
Regards,
Bruce