Testers/Testees

CSGnet Archive CSG1993
1

[From Rick Marken (930106.0830)]

Martin Taylor and Bill Powers are right on target about tests:

Martin Taylor (930105.1800)says:

... it seems to me that the testers are justified in being
happy that (more people survive the cancer longer| more
students can read the instructions in a manual| more people are
happy using the isntrument), even if the differences are
measurable only statistically. It's a question of whose
control systems are you worrying about, the tester's or the
testees'.

and Bill Powers (930105.2130) says:

Precisely. The insurance company cares earnestly about the
statistical effects of hazardous substances or conditions. The teacher cares
(or if certain policies are carried out, will come
to care) about turning out class after class that scores high, as
a population, on national tests. Mass statistics apply to masses.

The individual, however, is well-advised to avoid being part of
such mass systems, because the tradeoff is never in the
individual's favor.

This is really what most social science is about -- producing
average effects on groups. Psychology is really the science
of groups; unfortunately, it thinks it is about individuals,
and this is mainly self-deception. Now that I think of it, the
only places PCT has been able to have ANY impact on
conventional psychology is where this psychology is REALLY
about individuals -- as in much of perceptual psychology,
operant psychology and clinical psychology. PCT is
a model of individuals -- so right off the bat, a hugh
amount of psychological data, based on averaging over
individuals (ie. mass data based on sample measurements)
becomes irrelevant. Of course, it would not be irrelevant
if there were almost no variance across individuals and/or
if the reliabilities of the measurements were on the order of
.99. So Martin's point is well taken -- high reliabilities are
most important if you care about individuals (testees); you
can be a lot more sloppy if you just care about getting
mass data results that satisfy yourself (the tester) -- and
so the question to the people who want to do the testing
at my company is "are you doing this for the individuals
in the company, or for yourselves"? I'm sure that they im-
agine that they are doing it for the former; my job will be
to gently point out that they are really doing it for the
latter.

By the way Martin, thanks for the wonderful "loose
canon [sic]" article from your paper; I am having it framed.

Best

Rick

2

[Martin Taylor 930106 13:10]
(Bill Powers 930105.2130)

The individual, however, is well-advised to avoid being part of
such mass systems [tests. MMT], because the tradeoff is never in the
individual's favor.

What, never? Well, hardly ever! (Gilbert & Sullivan)

I would have thought that it depended on the likely outcome whether it was
in an individual's interest to submit to tests, be they educational, medical,
or otherwise. If I want to live longer, I have a better chance if the
cancer is detected earlier (by a test that is not guaranteed to give a
correct result, but which does so more often than not). If I want to
get into a university, and I know they go by exam results, I will submit
to a test if I think I will get a good mark, though the good mark is not
guaranteed. There are lots of reasons why an individual might want to
be part of mass tests, even if the results would not be applied directly
to that person. The indirect benefits resulting from the statistical
findings, which alter public policy in a way that generally improves
matters (though not for everybody), might make it worthwhile to submit to
the test.

You can't demand perfection from the results of your actions on the world.
You have reorganized so that your actions tend to move your perceptions in
the direction of their reference levels, but it is not only for reasons of
disturbance that the actions sometimes fail. It is more often (at higher
levels particularly) that you have not had the statistical opportunity to
reorganize to match the way Boss Reality works. Situations never recur
exactly, and at higher levels, by their very nature, there are more important
interactions that easily take you into situations where the results of actions
are different from what you expect, working from memory of apparently
related situations.

Reorganization works by statistics: you really can't do better when dealing
with the real world. The nature of reorganization, by whichever of the 12
different mechanisms, ensures that you do as well as you possibly can. But
you can't get perfection, so you have to go with probabilities. If the
probabilities seem likely to lead you into a perceptual state further from
your references than you now are, don't do whatever it is, whether it is
submitting to a test (that might get you fired from your job unfairly), or
running in front of an oncoming car.

Martin