[Martin Taylor 951212 16:05]
CHUCK TUCKER 951212
The comments on "instructions" MMT 1211 & RSM 1218 are
getting rather absurd. Are you seriously saying that
it makes not difference WHAT you ask another to do
when you do research?
Of course not. What has either of us said that would lead you to say such
a thing? You have been talking about specifying instructions precisely,
and since this seems to be a kind of shibboleth for journal editors, it
seemed worth comment. It's rather like quoting "significance levels" as
if they had some significance. Your comment is akin to suggesting that
someone who proposes that it might be a good idea for a car driver to
look out of the windscreen rather than applying a preprogrammed series
of twists to the steering wheel, is proposing that the steering wheel should
be twisted at random.
Every report that I know of
has a 'method' section with some instructions; if they
don't they probably don't get published. If they don't
have some instructions they can't be replicated.
I would argue that if the instructions are reported in the form: "These
are the words presented to the subjects" then the experiment is going to
be hard to replicate. Instead, if the instructions are reported in the
form: "This is what the subjects were given to understand that they
should try to do" then the experiment may be more replicable. The
report may be much longer than was needed for some subjects to gain the
proper understanding, and much shorter than for others. I've never had
much trouble getting my experimental papers published, and in none of
them have I listed a formal text of instructions. In all of them I have
explained what the subjects were asked to try to do.
You don't have to ask the other
to "Place the know over the spot" to do the rubberband
demostration or that Bill didn't have to write any
instructions for DEMO!
Let's be absurd, here. Suppose you want to do the rubber band demo,
using a knot in the table surface as the target location for the knot in
the rubber band. You specify the wording of the instructions, including
the sentence: "Your job is to keep the knot in the rubber band over the
knot in the table top." The subject responds: "What do you mean by the
'knot' in the table surface." You, the experimenter, remain mute, because
the written instructions don't include giving that answer. Do you think
the results of that session of the rubber-band experiment will be like
most of the others?
If Martin is correct (and he might be) that ALL research
using other persons where you ask them to volunteer for
money, class credit or just to please the researcher has
REINFORCERS by design....
Huh? Where did I say this?
....then it should be easy to "test"
RT. On the other hand, even PCT research has involved
reinforcers to the extent that you have asked others
to assist you with your research. ALL RESEARCH WITH
HUMANS USES REINFORCERS.
I'm lost here. As far as I can see, all that's happening is that the
subjects are controlling some of their perceptions, one of which is
the experimenter's state of satisfaction with them, for which they
have a reference value that the experimenter be reasonably well
satisfied.
There is no need to comment on the above, if you don't
wish to use instructions with others that are somewhat
similar or you don't care to make a record of the
activities you engage in during your "research" then just
don't do it.
Hang on, a minute. Obviously the instructions will be similar for most of
the different subjects, since you want to perceive the results of giving
the instructions to be similar. But they won't be (or shouldn't be) the
same, because then you'll get different understandings by the different
subejcts.
Likewise, you should record your behaviours, not necessarily your activities.
It's the results of what you did that matter, not the means whereby you
achieved those results. Yes, it's a good idea to record your activities,
insofar as you are able to perceive them, since it is your activities
that your subjects can perceive. But to assume that those activities have
the same effects on every subject, and are therefore worth reporting, is
to take a pure S-R view.
In doing an experiment, one tries to ensure that there is no direct conflict
between subject and experimenter, though there may be conflicts within
a subject. One assumes either that the subject actively collaborates,
by trying to do what the subject perceives the experimenter to want, or
that the subject is not doing anything deliberately opposed to what the
experimenter wants. The experimenter in either case can control his/her
perception of the subject without affecting the subject's ability to
control the perceptions that are at issue in the experiment. To use
fixed and formal instructions is to eliminate the experimenter's ability
to control his/her perception of the subject, and therefor it is to
introduce unnecessary variability into the actual conduct and results
of the experiment.
I think that is just being honest.
Although this quote is taken in a reversed context, I think it applies
to what I did say and now say.
Martin