social science, capacities and machines


One of the best things that I have come across recently is the work of philosopher Nancy Cartwright. I read How the Laws of Physics Lie (1983) a few months back, and liked it so much that I ordered her 1999 book entitled The Dappled World, which certainly did not disappoint. In this latest work Cartwright further develops and extends most of the arguments first laid out in her earlier book. In particular a realist broadside against the deductive-nomological model of explanation and the view that absolute universal laws are invariant features of the world.

Instead Cartwright argues that most “laws” are simply ceteris paribus laws and that what is real are the capacities of the real world objects and processes, that only when arranged in specifiable ways and only when protected from the disturbing influences of other casually relevant objects and processes, produce the regularities that are described by abstract laws expressed in mathematical form. Thus, things and their power to make a difference are real, while “laws” are simply a short-hand way of analytically describing what certain things do in very delimited and exceptional contexts (i.e. laboratories).

In The Dappled World, Cartwright extends this framework to social science (most of her examples are from econometrics). As you would expect, she is very skeptical of the view that the parameters estimated by statistical models are in fact pointing to nomological regularities that actually exist in the world. She has a nice critique of the standard view which holds that if the data show some empirical patterns of association, then there must be some recoverable probability model that can be thought of as having “generated” the data.

For Cartwright, while this situation is possible, it is highly unlikely since for some real world process to be able to “generate” regularities it must be arranged in a very specific way, which she refers to as a (socioeconomic) nomological machine. From her point of view “Regular behavior derives from the repeated triggering of determinate systems whose nature stay fixed long enough to manifest themselves in the resulting regularity.” For Cartwright, in contrast sometimes (more commonly than we realize) “there are simply no probabilities to be had.”

I find this idea of a nomological machine fairly attractive, since it points to the need to think careful about, before we begin to write down mathematical descriptions or specify linear regression models of some outcome, the actual institutional arrangement that might be responsible for a specific outcome of interest (which sounds like SOC 101, but is pretty rare in most empirical works which resort to high powered statistical modelling). Furthermore, we can think of a few very basic social mechanisms, from the level of the small group to entire institutions that do function as such nomological engines productive of predictable regularities (a view that Stinchcombe–in the case of organizations–defends as a virtue of the “old institutionalism”).

In Cartwright’s view most outcomes in economics (and by extension in all social science) are produced by such historically specific nomological machines, which serve to channel, modify and reroute the capacities of individuals and other institutional arrangements so that certain regularities are produced with some degree of predictability. By implication, “unpredictability,” randomness and chaos can be thought of as caused by the breakdown of these socioeconomic nomological machines (another name for institutions with a big “I”?). This appears to be a nice way of reconciling concerns with historicity with the scientific attempt to explain measurable regularities.

Written by Omar

October 31, 2006 at 2:21 am

5 Responses

Subscribe to comments with RSS.

  1. This book sounds fascinating. Something else to go on my reading pile. Based on your description, she is almost proposing a return to Weberian institutionalism.



    October 31, 2006 at 4:23 am

  2. You can find many of these ideas in Roy Bhaskar’s book from 1975, A Realist Theory of Science, or in the earlier work of Rom Harré. There are quite a number of discussions of social science methodology from a realist perspective. Some of the best stuff is in political science. Economics has seen rather few applications of these ideas, although a number of Bhaskar disciples (mainly Tony Lawson) have been pretty active. (I have an old paper (1994) on realism in economics in an obscure journal called Journal of Social and Biological Systems). Peter Abell of LSE defends a realist program in sociology.


    Nicolai Foss

    October 31, 2006 at 8:11 am

  3. You might be interested in this extended review of the book, by A Certain Philosopher. The Bhaskar connection is there, but Bhaskar’s view is much more reified than Cartwright’s — his separation of the various “layers” is pretty ridiculous.



    November 2, 2006 at 12:40 am

  4. […] social science, capacities and machines […]


  5. […] Following on from Omar’s post about Nancy Cartwright’s The Dappled World, I will abuse my right to post here to point to this review of the book by Laurie Paul. Paul is a very intelligent philosopher (her unwise decision to marry me was an uncharacteristic lapse), and her review gives a good sense of the response to Cartwright’s position within philosophy. I’ve found that social scientists who read Cartwright’s book are excited and interested by it. One reason for this is that her approach resonates nicely with how many researchers (and especially social scientists) feel like they do their work. This is not accidental—unlike many philosophers of science, Cartwright pays attention to the social sciences as well as the natural sciences, and has a much better feel than many philosophers of science for how work in various fields really gets done. […]


Comments are closed.

%d bloggers like this: