Monday, December 6, 2010

Granularity




If we think back over the history of sociology and political science since 1800, one thing that is striking is the trend away from the most macro-level frameworks of thought in the direction of more situated and proximate social and political arrangements. There was a tendency among the founders -- Montesquieu, Spenser, Comte, for example -- to pose their questions at the civilizational level. The largest social constructs were identified as worthy of study -- religions, kinds of governments, cultures. By the time we get to writers like Tocqueville and John Stuart Mill in mid-nineteenth century, we have thinkers approaching modern social life with a more limited gaze. They are more interested in particular, differentiating social and political arrangements, and they are interested in the particulars of human behavior within those arrangements. And they are much more interested in the empirical details, rather than general philosophical theories of democracy or religion.

Twentieth century thinking in the social sciences has generally continued this trend. What makes social science "scientific" is rigorous attention to empirical characteristics of the social world.  There is generally even less patience with large philosophical theories of society and history.  A good theory isn't of much interest unless it can be closely tied to particular bodies of empirical observation. It is hard to think of a 20th-century sociologist who took up the Comtean project of comparing the course of civilizations. And there is a growing consensus that the focus of social research needs somehow to capture the behavior of situated actors within socially specific arrangements -- in other words, a refinement of focus towards the more particular arrangements.

A partial exception to this trend is the field of comparative politics (Lichbach and Zuckerman, eds., Comparative Politics: Rationality, Culture, and Structure). This is a paradigm that undertakes to discover similarities and differences in large political institutions across settings -- with the implication that there is some reason to expect that "developmental states" will have some important similarities whether in Brazil, Nigeria, or Syria. How does "electoral competition" work in the United States, Mexico, and Poland? Hw do anti-corruption regimes work in Indonesia, Singapore, and Hong Kong? Comparative politics specialists have accepted the need for careful empirical investigation, to be sure, and they are mindful of differences as well as similarities. But the field is more macro than many other areas of contemporary social science.

Another exception seems to be "world systems theory", which locates social-historical analysis at the global level.  But as argued earlier (link), Wallerstein was in fact sensitive to historically specific "granularities" within the world system.  Here's what I concluded there:
But Wallerstein's practice as a sociologist is far more defensible than this language would suggest. He was in fact sensitive to causal heterogeneity, contingency, and variation in the systemic relations he meant to capture -- particularity as well as universality. So he doesn't actually treat the modern world system as if it were analogous to a set of gravitational objects governed by fixed laws of nature.
So we might say that two of the large features of modern social science research paradigms are these: the conviction that a researcher needs to formulate questions that permit careful empirical observation and assessment; and the idea that the scope of research should be tailored to capture social behavior and organization at an intermediate level. We need to formulate research questions that capture enough social behavior to be interesting, while at the same time permitting careful, rigorous empirical investigation and assessment.

This is where the metaphor of granularity comes in. It suggests the idea of a deliberate choice of levels of analysis, a deliberate selection of the zone of focus we choose for the investigation. Go too fine and you lose sight of the structures that will turn out to be interesting. Go too gross and, once again, those structures disappear. Then there is the important idea of heterogeneity -- the idea that stuff is not uniform in structure all the way down (link); so it matters what level we choose to focus on.

The scanning electron microscope image above (center) is illustrative of this fact with regard to the natural world. Carbon appears uniform at a gross level (anthracite), and it is also uniform at an atomic level.  But there is an intermediate level of resolution where there are structures that are very important for the physical properties of the stuff; and we won't find these structures unless we look at this intermediate level.

Something very much like this seems to be true about the social realm as well.  If we ask questions about "citizens" or "consumers", we have characterized the actor in a highly abstract way, with no indication or recognition of the more specific circumstances in which the actor conducts his/her life.  If we provide more institutional, organizational, and motivational context, however -- that is, if we increase the level of resolution in our depiction of the social actor -- we are likely to find more illuminating features of behavior, and we are likely to find discriminating explanations for differences in behavior.

Perhaps an illustration is found in the kinds of social action that Erving Goffman investigates -- the stylized behaviors and interactions of restaurant workers, for example (link). Go too deeply into the personal stories of the persons involved and you lose what is distinctive about the social context of the restaurant. Aggregate facts about service workers everywhere, and once again, the distinctive behaviors disappear.  And, if we abstract too severely from the specifics of the social setting of the restaurant worker -- that is, if we zoom our microscope out to the level of pure rational choice theory -- then we will overlook the differentiating social circumstances that influence behavior in real hotels and restaurants.

We might say that all of this is pointing towards a certain kind of sociological research that focuses on the proximate social settings within which people act and interact. This takes us towards what I call methodological localism (link). But it also conforms to the sociological frameworks of the new institutionalism (linklink) and even with the emerging emphasis on explanations based on social causal mechanisms (link).  

In seeking out the concept of granularity here, I've actually invoked several different contrasts: abstract to concrete, apriori to empirical, and macroscopic to microscopic.  If we put all these together, we might say that a "granular" approach to a social phenomenon is one that achieves a degree of specificity in conceptual definition, a robust degree of empirical detail, and an intermediate analytical and representational perspective that permits discovery of the mid-sized structures that influence the behavior of the whole.

(This issue has come up previously under the topic of "macro, meso, micro" (link).)

1 comment:

russell1200 said...

http://en.wikipedia.org/wiki/Cliodynamics

Various cyclical theories go against this trend. But the fact that so many people hate cyclical theories probably proves your point.