[Martin Taylor 951211 12:00]
Rick Marken (951208.0830)
Gee, this is getting boring... I've got to find something to disagree with
Rick about 
Chuck and Bill: Observations are obviously important. But why do you guys
think instructions are important? Obviously, things are a lot easier if you
can talk to your subjects and tell them what you'd like them to do. And it's
also nice if you can assume that your subjects' have the goal of trying to
understand and follow your instructions (when I was a grad student, some
of my best girlfriends were subjects with just this goal;-)). But why the
concern about the "precision" of instructions -- since you can have only the
vaguest idea how anything you say might be interpreted by your subjects.
I agree 100% (though I envy Rick his subject pool as a grad student). When
I was doing my thesis work, I carried on a continuing argument with the
faculty on this point. I wanted to perceive all my subjects as having the
same understanding of what they were to do, whereas the tradition (and it
is no more than that) was (and apparently still is) to provide the subjects
with the same instructions, as written down before ever the first subjects
showed up.
How is it possible that subjects with different backgrounds, experiences, and
capabilities can be expected to treat the experiment similarly if you give
them all the same instructions? Seen from the viewpoint of PCT, it is the
purest example of S-R thinking I have come across.
One of the objectives of precision in instructions is to avoid "biasing"
the subject into giving you what you want to see in the results. That's
a legitimate objective, but specifying the instructions beforehand is not
going to prevent the subjects from being biased. You have to go a lot
further--double blind conditions and stuff like that. When you are doing
an experiment in which the subjects CAN bias the results, you have to
wonder anyway whether your interpretations of the results will later
stand up. When the subjects can please you (other than by doing their best),
they are probably going to be figuring out how they can best please you,
and the way YOUR subjects please YOU is not necessarily the way another
experimenter's subjects would please her.
Many means to the same end--many different instructions to the same
understanding. Easy to say, hard to do. But an open-loop approach to
the giving of instructions is worse, just as is tracking based on pure
planning.
Martin