[OPE-L] models with unequal turnover periods

From: Jurriaan Bendien (adsl675281@TISCALI.NL)
Date: Tue Sep 11 2007 - 16:07:01 EDT

Paul Z wrote:

"I must say it is rather disconcerting: there is so much movement in your
description that we'd almost seem to have to thrown in the towel."

I do need my towel, I haven't thrown it in the ring yet. I worked in
statistics for some years, and there you learn - if you didn't know it
already - to draw a sharp distinction between a changeable reality that is
in motion, and what you can actually measure, in good time. What you can
measure, is only an aspect of reality, and you have to be able to separate
out constants and variables using categorisation schemes. Sometimes what you
use as variable, and what you regard as a constant, might not make much
scientific sense, but it makes practical sense, for the purpose of what you
try to measure.

Suppose you said, "I want to measure the kinship relations (or total
families by type) of the nation", which social statisticians in fact want to

At first sight, the research statistician observes you simply cannot do it,
it does not make any sense at all. Bunkum. Wrap the towel round your head or
something. Because where those relationship begin and end, is mighty
difficult to tell at the best of times. What is your measurement unit? You
get to very basic inquiries such as: what is a "family"? What is obvious to
everybody, suddenly is not obvious anymore. But the issues aren't
"philosophical" anymore either, because now you've got to count them.

Okay, grossly simplifying the abstractive process, step one: we measure
kinship relations within households only. Step two: define households (again
the boundaries can be tricky there). Step three: identify the universe of
possible kinship relations. Well there is an awful lot of those, and you
simply cannot measure all those. Question arise for example, do you include
same-sex relationships with children as families? There are, besides, all
sorts of intractable problems of kinship relations, where you don't know if
you have got one, or you haven't got one. When is a parent a parent? Step
four: design a classification of logically discrete, measurable kinship
relations. Step five: design a questionnaire format that will actually
collect your data reliably in a way you can process it. Step six: decide on
a reliably survey methodology (samples and frames), collect your data, and
code it (well really there's a lot of steps to go through here of course).
Step seven: apply math, programming and classification principles to
aggregate your data and then tabulate it. Step seven: study the distribution
obtained, to iron out logical, data and boundary problems, so that your
result looks sufficiently credible against what you know already. Step
eight: finally store and present your data in a way that is comprehensible
to the user of that data, and comparable with previous data, or other data.

That is, if you like, a (sort of vulgarised, abbreviated) description of how
you might do it. But suppose you have crunched the numbers in this sense,
then what is this artifact you have obtained? Can you really say you have
measured the kinship relations of the nation? Well, you cannot in truth say
that. All you can say is that you have measured some "aspects" or
"dimensions" of kinship relations. It is not the total story of kinship
relations, and you may not even be able to tell for sure how accurate your
measurement is, beyond comparing different data sets. Nothing is easier for
an scholar or researcher to come along and buy some data, in order to prove
or disprove a theory he entertains about "kinship relations in the nation",
but really even after all the work that has been done, there's still not a
lot you can validly say about the topic compared to what there really is out
there. But there are some things that you can validly say about the data,
and so the whole "art" then becomes one of knowing what you can validly
claim about the data.

Why say all this maybe dreary stuff? Because I tend to think that, if you
just accepted these capital compositions and movements as given, THEN you
have already thrown in the towel. You don't throw in the towel, if you study
what these capital movements actually mean in reality. But that could take
years and years of solid empirical and theoretical research, including the
question of how I can best go about getting to know this? Theory may provide
some quick shortcuts, but in the end, if you really want to understand what
you are dealing with, well, it takes a lot of factfinding. Anyway, I am now
going to take my towel and dry myself, I am starting to feel a bit wet here,
for some reason... :-)



PS - I often think these days that people have forgotten what theory is, and
what its purpose is; "theory" might be more a metaphor to choreograph their
lives with. Hell, if I had that salary, I'd be away laughing :-) Actually I
didn't not even mention previously that "intermediate consumption" also
includes some "faux frais of production" (incidental expenses) which are not
(in Marx's opinion) Cc but that is another story.

This archive was generated by hypermail 2.1.5 : Sun Sep 30 2007 - 00:00:05 EDT