[OPE-L:2196] Statistical regularities

From: P.J.Wells@open.ac.uk
Date: Mon Jan 17 2000 - 13:24:31 EST

[ show plain text ]

I entirely agree with the first clause of the second sentence from Nicola,

> If a statistical regularity is discovered between, say, eating ice-cream
> and going to church, what inferences would Julian draw from the
> regularity?
> I would say none, since no causative inferences about either structure or
> free will can be drawn from spontaneously occurring (unrestricted) social
> regularities.
        But this isn't the sort of thing that I, or the 19th-century
debaters, had in mind: rather, such "facts" as that the annual average
number of murders or suicides in a particular jurisdiction was normally
distributed about a stable mean.

        It not unreasonably occurred to those working with social statistics
to wonder what this implied about the sources of human behaviour.

        Since murder -- and especially self-murder (as many thought of it in
those days) seem on the face of it the most extreme outcomes of human will
and intentionality, one might expect the annual rates of these to be very
erratic, whereas they are (or were) apparently rather regularly distributed.

        Quetelet was an outright statistical fatalist: he fully believed
that the regularities which he discovered implied that the agents were under
compulsion to carry out the acts involved: "society prepares the crimes", he
said. Thus he argued that responsibility and punishment were inappropriate
categories in this connection.

        The interesting thing is that support for statistical fatalism was
most marked among those who adhered to Manchester-type liberalism; among the
most prominent on the other side (i.e., who claimed that social regularities
of this kind could still be the outcomes of voluntary acts) were those
around the Prussian statistical bureau and supporters of "national economy"
generally (see Ian Hacking's book "The taming of chance" for an account of
all this).

        BUT I don't go along with the second clause -- it seems to be a
tautology: if events are genuinely "spontaneously occurring", then naturally
it would be inappropriate to derive causative inferences from them. The
point is HOW one distinguishes (if one can) causally-determined actions from
those resulting from free will, faced with a bare regularity, such as the
distribution of suicide rates.

        The relevance of all this to F&M is that their theory is NOT about
regularities of the kind mentioned in Nicola's post (Machover's day job is
as a philosopher of science, so he is unlikely to come within range of
Hume's strictures on this subject).

        Instead, their claims are simply about the *kind* of distribution
various quantities (including the rate of profit and the rate of surplus
value) must have, given very simple assumptions about firms' behaviour --
essentially that the firms' actions with regard to inputs and outputs are
unco-ordinated ("independent"), and that their interactions are solely
through exchange in a competitive market.

        N.B., in response to a possible misunderstanding on Mike W's part,
"competitive market" here is not the same thing as in neo-classical
shorthand: i.e. it doesn't mean "perfectly competitive". F&M argue that
market power, collusion, etc. undoubtedly reduce the number of degrees of
freedom in the system (increase the amount of co-ordination among agents),
but in practice not by enough to significantly reduce the number of degrees
of freedom. Note also that for F&M a "commodity" is the bag of flour I
bought yesterday: the two bags of flour which you may buy tomorrow (in the
same transaction) constitute a *different* commodity; both are examples of
the same "commodity-type" (F&M, p. 76).

        Since not only are there millions of distinct commodity-types in a
real capitalist economy, but billions of daily transactions, market
intervention would have to be pretty thorough to undermine the applicability
of statistical mechanical formalism.

        (There is also the technical point that many of the constraints
introduced by government intervention, etc., are in fact *inequalities*
(e.g. maximum and minimum prices or output levels): these do not reduce the
number of degrees of freedom (F&M, p59).)

This archive was generated by hypermail 2b29 : Mon Jan 31 2000 - 07:00:07 EST