RE: BHA: RE: Predictions

["Marshall M. A. Feldman" <Marsh@xxxxxxx>]

Hi,

I agree with John that this is an important topic. However, I want to raise
two points.

1. Doesn't Tony Lawson's notion of a "demireg" make prediction somewhat more
sensible, even within open systems?

2. I think the positivist literature, including Uncle Milty, tends to use
"prediction" ambiguously -- often I think for political purposes.
"Prediction" can mean one of two things: (1) in the future such and such
will happen; (2) is much more restricted and is much more along the lines of
Hempel's argument, "if we do this and that, then such and such will happen."
The logic of positivist accounts of scientific method only warrants the
second meaning. Hence, we can say, "If we examine the data on 19th century
Britain, then we'll find a correlation between rising consumer incomes and
decreasing infant mortality." Embedded in the latter statement are two
implicit statements. One is that if our theory is correct, then we'll
observe the correlation -- this is Hempel and an epistemological claim. The
other is, "if consumer incomes increase, then infant mortality will
decrease" -- this is an ontological claim, mirroring the theoretical
propositions presumably embedded in the theory being tested. As is common
with positivism, the observables are thought of as "variables" and the
relation between theory and observation is thought to be relatively
unproblematic, with 1:1 correspondences between theoretical concepts and
observables.

Notice that nothing in either logic implies our ability to predict in the
sense of (1). At best, what might be called genuinely future predictions can
take the form, "if A happens, then B will happen." But we can't predict that
A will happen without running the risk of infinite regress. I refer to this
as a "genuinely future prediction" to differentiate it from predictions
about hypothesis tests as in my example, which deliberately refers to the
past (the hypothesis will be tested in the future, hence the outcome of the
test is predicted; the phenomenon under study is not). Hence, even within a
positivist framework our ability to predict does not follow from the logic
of the positivist argument.

There is, of course, an other positivist argument which makes much stronger
ontological claims. It claims the world is ordered and structured, changes
slowly if at all, that the empirical level exhausts reality, that causes are
empirical regularities, that change is generally about the levels of
variables rather than structures, and that contingency is merely a matter of
probabilities (which can be known). [I once had to teach from a research
method text written by McGaw that presented this trash as if it were the
foundation of "science" (with religion and witchcraft being the only two
alternatives, if I recall correctly). As is common with this genre of
textbook, students are force-fed this stuff, and it's on the exam.] Within
the scope of this argument, predictions are possible.

I say that I suspect positivists are motivated to be ambiguous for political
reasons (whether they know it or not) because the ability to make "genuinely
future predictions" opens up the doors to the halls of power. Technocrats
are hired to predict what will happen. Moreover, the second implicit
statement, which presents the historically, geographically, and contextually
specific correlation (statistical correlations are always historical,
geographical, and contextual) as a universal law (again, usually via tacit
assumption rather than explicit argument), gives one further entry into the
halls of power, because now one can advise on policy: not only because of an
alleged ability to predict the future, but also by being able to predict the
consequences of action.

	Marsh Feldman
	Department of Community PLANNING
	The University of Rhode Island

> -----Original Message-----
> From: owner-bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx
> [mailto:owner-bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx]On Behalf Of Andrew
> Mearman
> Sent: Friday, August 15, 2003 5:33 AM
> To: bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx
> Subject: Re: BHA: RE: Predictions
>
>
> Just to add to this: of course there is nothing to stop a person
> making a prediction, but in open systems, it is much less likely
> that the prediction will be correct. One important epitemological
> implication of this is that positions such as Milton Friedman's -
> that theories should be assessed solely on their predictive power
> - are untenable in open systems.
>
> It might also be useful to distinguish between predictions of
> events and predictions of other types, perhaps of structural
> changes, processes, etc. The latter are clearly easier in open
> systems. For instance, we can predict very well that kittens
> change into cats. That is part of their nature. Of course, this
> still occurs in open systems and their are other factors which
> can prevent the kitten becoming a cat.
>
> Thanks for the question. It's an important one, because if
> theoretical bodies derived from/consistent with CR wish to be
> relevant and to help shape policy, they (at least in the current
> institutional framework) have to make predictions, albeit more
> cautiously than perhaps currently occurs.
>
> andrew
>
> John Ridgway <John.Ridgway@xxxxxxxxxxxxx> wrote:
>
>
>
> Mint
> Just a quick response. As I recall the notion of Rhythmic in DPF
> (repeated patterns of social behaviour which is the foundation of a
> social institution) does mean that predictions are possible or useful
> (necessary to operate in society).
>
> The fact that the system is open means that the predication will not
> necessarily be perfect, that a person as a casually efficacious being
> may change therefore social institutions and society can change.
>
>
> John r
> -----Original Message-----
> From: Ann Bergman [mailto:Ann.Bergman@xxxxxx]
> Sent: Thursday, 14 August 2003 9:58 PM
> To: bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx
> Subject: BHA: Predictions
>
>
> Hi,
>
> I have a question about the use and relevance of predictions in social
> science. What I can understand of critical realism predictions are not
> useful since society is an open system. At the same time critical
> realist
> argue for the importance of identifying structures and mechanisms which
> makes society to what it is and how it works. Is not the way it is and
> how
> it works is in many ways predictable? We often know what to expect in
> certain situations or after a certain action is taken. Is it possible
> for a
> critical realist to argue that the structured society and its durable
> patterns are in some way possible to predict. How do you think about
> these
> issues?
>
> Sorry for my terrible english
> Mint
>
>
>
> --- from list bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx ---
>
> ------------------------------------------------------------------
> ----------
> This email, together with any attachments, is intended for the named
> recipient(s) only and may contain privileged and confidential
> information. If
> received in error, please inform the sender as quickly as
> possible and delete
> this email and any copies from your computer system network.
>
> If not an intended recipient of this email, you must not copy,
> distribute or
> rely upon it and any form of disclosure, modification,
> distribution and/or
> publication of this email is prohibited.
>
> Unless stated otherwise, this email represents only the views of
> the sender and
> not the views of the Queensland Government.
> ------------------------------------------------------------------
> ----------
>
>
> --- from list bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx ---
>
> ---------------------------------
> Do you Yahoo!?
> Yahoo! SiteBuilder - Free, easy-to-use web site design software
>
> --- StripMime Warning --  MIME attachments removed ---
> This message may have contained attachments which were removed.
>
> Sorry, we do not allow attachments on this list.
>
> --- StripMime Report -- processed MIME parts ---
> multipart/alternative
>   text/plain (text body -- kept)
>   text/html
> ---
>
>
>      --- from list bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx ---
>



     --- from list bhaskar@xxxxxxxxxxxxxxxxxxxxxxxxxx ---