[OPE-L:2162] Empirical method

Alan Freeman (100042.617@compuserve.com)
Sat, 11 May 1996 06:00:33 -0700

[ show plain text ]

What role for empirical verification?
Re Paul's #2135:

There is then a second issue which is the programme of enquiry which
Paul himself proposes. Since he sets it up as the alternative to
Scholasticism, an examination of what he proposes should help assess
what Scholasticism actually is.

Why study anything other people say? One can take the view, which
seems to be Paul's, that it is not really worth doing. All one does
is construct ab initio a theory of the world that can be tested
(using the methods that the theory itself proposes to test itself
with...), test it, and bingo, we have science.

This leads to a certain method of debate which I don't think is
likely to be productive. If we think that the only issue to be
settled is which theory is 'correct', and the only test to which it
can be subjected is the test of explaining data, and moreover we
define that the test to be applied is the test which the theory
itself prescribes, then

(a)the result is hermetically-sealed. It is bound to confirm itself,
because it supplies its own methods of verification.

If, whenever I argue with someone about theory, I am told this is
illegitimate because we must go to the data; and if when I
challenge the data I am told this is illegitimate because I am
using the wrong theory (I seem to recall that at one point Paul
objected to using numbers because of all the presuppositions they
contained) then we should not be surprised if the debate is not
very constructive.

(b)it is bound to generate a great deal of heat, because the demand
which is placed on theory A by author B, is that theory A should
be tested using theory B's methods.

What is under discussion (the theory including its methods of
testing) is presupposed by the test. Theory B has written author A
out of the discussion before the discussion even begins.

Theory B has set itself up as a complete judge and jury of all
other theory: as a source of absolute authority.

I believe that this is what Bacon objected to in the methods of the

In OPE 2135 Paul says that portraying reality is

"a theory of prices and profits in a capitalist economy that can be
put to statistical tests against real data from such economies,
and come up with good predictions."

But two paragraphs earlier he states:

"As a theory, I have found that, as one would expect, your version
of prices of production theory gives more accurate predictions
than the orthodox Sraffian theory. I say, as one would expect,
since it is actually a weaker theory - closer to surface
appearances than either the labour theory of value or other prices
of production theories."

Well come on then, which criterion applies? Is it the criterion of
accurate prediction, or the criterion of being close to the surface?
And if 'the' labour theory of value (a choice of words that begs a
rather large number of questions for reasons given earlier) is
further from the surface, how is it a good empirical predictor?

This whole argument looks perilously to me as if Paul chooses the
terrain of figures when he has trouble with theory, and theory when
he has trouble with figures. I think the most charitable
characterisation of this method is Eclecticism.

However there is an even more fundamental issue. How can Paul's
theory ever be refuted? In #2135 he puts forward a quite reasonable
description of a limited range of issues to be investigated with this
theory and I have no problem with that. But elsewhere in the debate
(nearly everywhere until now) he has proposed this empirical method
as confirmation of the *theory itself*. Thus the theory is confirmed
by the data and the data is confirmed by the theory. This I think is
wholly illegitimate.

If we really wish to pursue the Inductive and the method
systematically, if we really want to do away with Scholasticism, then
we should pay attention to Bacon's idea of seeking negative
refutation as well as positive affirmation. An obvious question to
Paul flows: is there any possible empirical basis for *rejecting* the
hypothesis that v=va+l portrays reality?

Being a loyal Baconian, one of the first experiments I conducted with
the i/o technique of verifying correlations between VILE magnitudes
and observed prices, was to repeat the calculations with a set of
data constructed from random numbers.

I would recommend Paul and Allin to do the same before pursuing the
argument from empirical verification much further. As far as I can
see they have a theory which can never be refuted.

This applies, of course, only if they use their calculations as a
basis to support the theory that v=va+l is an accurate portrayal of
reality. If they use these calculations to support the substantially
more limited assertions made in #2135, none of which confirm or deny
that v=va+l is an accurate portrayal of reality, then I have no
problems with that.