Research Methods Addendum

[From Rick Marken (941230.2320)]

Bruce Abbott (941230.1700 EST) --

I should note that I am aware that your example of using the IV-DV
method to study the relationship between the RC transmitter stick and
angles of the rudder and elevators on your plane does illustrate the use
of this methodology to study a control system. As you note, the stick
adjusts reference settings for the angle variables. Even without applying
disturbances to test for controlled variables you learn some very
interesting things about the behavior of this control system; you are
learning precisely what you said you were learning -- the relationship
between stick position and angle variables. You have demonstrated a
use of the IV-DV method that CAN tell us something useful about control
systems. Unfortunately, what you described is NOT the way the IV-DV
approach is used in psychology. Psychologists do not have direct
access to the reference inputs to living control systems; so the
IV in psychological experiments is never variation in control system
reference signals and the DV is generally not a measure of a (probable)
controlled variable. In psychological experiments, the IV is an
environmental variable; the only environmental variable that can be manipulat-
ed by the experimenter (without active resistance from the control system) --
the IV is what we call the disturbance variable in PCT. The DV is typically
some arbitrary measure of system "output"; it's not likely to be a controlled
variable becuase, if it were, it would show no relationship at all to
the IV and would be rejected as an irrelevant response.

So my critique of IV-DV research applies most emphatically to the
situation where we are forced to study control systems "from the
outside" -- from whence there is no direct access to reference signals.
In other words, it applies to all behavioral science research.