byproducts, of course. To the extent that citizens exercise such control, they are more likely to extend their search for information, to be unable to locate reliable information sources, and thus to be politically isolated ( Huckfeldt 1983).
Alternatively, to the extent that informational choice is probabilistic, the incoming stream of information becomes a crucial consideration. What affects the content and composition of this incoming stream? The answer lies in media exposure and in the multiple bases of social experience to which citizens are exposed, and a steadily accumulating body of evidence points to the importance of these various bases of experience for political information and the formation of political preference.
The political influence of social experience is best understood relative to the various bases of this experience. At one and the same time, citizens live in households, among immediate neighbors, located in the middle of larger neighborhoods, surrounded by a city, a county, a state, and a region. Coupled with these relatively inescapable geographically based environments are a whole series of less geographically dependent environments: workplaces, churches, taverns, bowling leagues, little leagues, health clubs, and so on. And each of these environments, whether it be geographically or nongeographically based, serves to establish constraints and opportunities acting upon social interaction.
None of these environments is necessarily more important than any other in influencing citizen behavior. In particular, it is a mistake to believe that more intimately defined environments are more important than environments that are larger and more impersonal. Indeed, the work of Erikson, Wright, and McIver ( 1989) and Wright and Berkman ( 1986) demonstrates quite persuasively that the state is the most appropriate environmental unit for many purposes of political analysis. The important point is that opportunities for social interaction are circumscribed by availability, availability is influenced by a range of environments defined at various levels, and thus social experience arises in a particular place and time.
Just as individual choice is not determinant in the selection of information sources, neither is the environment. Citizens do not simply roll over and accept whatever comes along in terms of social interaction opportunities. All of us avoid association with some individuals while we pursue it with others. And thus individual social experience is best seen as the end result of a complex interplay between individual choice and environmental supply. Just as the environment is composed of multiple and intersecting dimensions of experience (work, neighborhood, church, tavern), so also is individual associational preference multi-dimensional, responding to a range of different goals and objectives. To the extent that social experience carries political content, these life domains become important to the diffusion of political information and to the resulting preferences and choices of citizens.
In short, individual discretion plays an important role in defining social space and thereby determining social exposure. Learning theory points toward the importance of exposure, and a range of empirical findings show that, for example, organizational involvement can serve both to shield individuals from and expose individuals to contextual influence ( Putnam 1966; Segal and Meyer 1974; Cox 1974). In point of fact, social contexts are created at least in part by an individual's construction of a social space. And thus the social context reflects a series of socially structured decisions regarding where to live, work, worship, drink beer, bowl, and so on. The social context experienced by any individual is the point of intersection between all these environments. As an empirical matter, we will be fortunate if we can obtain contextual measures on one or two environments at a time ( Wald, Owen, and Hill 1988, 1990), but the inconveniences of measurement should not obscure the underlying theoretical issues.
Of what value is a micro theory of contextual influence? We have already denied that its utility must be evaluated in terms of explaining individual behavior, but rather that it lies in the specification of interdependence. Of what value is the specification of interdependence?
An important dimension of contextual analysis is the effort to deduce the aggregate consequences of interdependence. These efforts are typically modeling enterprises that ask a series of "what if" questions. If the mechanism of contextual influence is social learning or behavioral contagion or social reference groups, what are the consequences for the form and dynamic of public opinion ( Sprague 1982; Huckfeldt 1983; MacKuen 1990; MacKuen and C. Brown 1987)? What difference does it make if the source of working class consciousness is to be located in exposure to the working class or in exposure to class-conscious working class behavior ( Przeworski and Soares 1971; Przeworski 1974)? If citizens are affected both by the groups to which they belong and by the groups to which they do not belong,