File spoon-archives/bhaskar.archive/bhaskar_1998/bhaskar.9803, message 8

Date: Thu, 5 Mar 1998 17:55:34 -0400
Subject: BHA: Realist Methods

I had been too bogged down to respond to this topic of realist methods
before, but as I have been contemplating a paper on this subject for the
realist conference, I am grateful to Andy for another opportunity to jump

Andy, I agree with almost everything you say on methods except your ruling
out simple correlations and regressions.  Actually, there is no real
ontological difference between what is presupposed by, say, chi-square on
the one hand and zero-order regression or correlation on the other.  Each
is measuring covariation between two variables -- event regularities. Thus,
if regression is to be ruled out on that basis, so, too, should chi-square.

I think, however, that realists have no reason to reject any such
statistical techniques.  What makes realist use of statistics distinct is
our interpretation of what we are doing when we use them.  Certainly, we
are not looking to uncover invariate social laws or even statistical laws.
Thus, we would not attempt to identify contextless generalizations, say
between size and differentiation.

The crucial difference between our use of correlations and positivism's is
that for us, the statistical relations are just evidence for an explanation
and not the explanation itself.  For us, a zero-order regression or
correlation is a first piece of evidence that a particular mechanism is
operative in a particular context.  It is, of course, as all here seem to
agree, not conclusive evidence.  We still would like to see if the
covariation remains after controlling for the operation of other
mechanisms.  And the statistical verification that a mechanism is operating
in one socio-historical context does not license the lifting of that
correlation out of context and making it into the kind of generalization
that would support the covering law model of explanation.

With that caveat, however, even mere correlation is often a good, first
piece of evidence as to whether, for example, banks are redlining, whether
people who have lost welfare coverage are doing worse, or whether
international inequality is a consequence of capitalist penetration.

Another consequence of realism, as, I think, some have already said, is
that statistical analysis becomes deprivileged as just one of multiple
methodologies, all of which have their appropriate place.

These, at any rate, are my thoughts on the subject. (A copy of Sayer is
finally on its way to me from Routledge so I have no idea whether I have
just repeated what has already been said.  If so, sorry.)

Note to Caroline:  Re this surprise virtual reappearance, I had no idea i
would be able to de-lurk so soon.


doug porpora
dept of psych and sociology
drexel university
phila pa 19104

     --- from list ---


Driftline Main Page


Display software: ArchTracker © Malgosia Askanas, 2000-2005