This investigation addresses the issue of how the social context in which citizens live, work, and play is transformed into social influence on voting behavior in an election campaign. Questions that follow from this include: What mechanisms promote homogeneity or heterogeneity in political choice in neighborhoods or work groups? In what ways do the social contexts of work, recreation, or neighborhood life filter the perception of politically relevant information coming from the mass media or other persons? What mechanisms of social interaction enhance or inhibit social influences on political choice? In the highly dynamic context of a political campaign, how does the political influence of social context evolve over time? What mechanisms govern the changing course of social influence on political choice during a campaign? To address these questions the investigators have undertaken a three wave panel survey of 1,500 voters in South Bend, Indiana, before, during, and after the 1984 election campaign. A supplemental survey of almost 1,000 persons identified by respondents as political discussion partners was also completed. This survey employed a research design and sampling technique that differed markedly from traditional practice. Extensive scientific study of mass political behavior, especially voting behavior, has been undertaken over the past four decades utilizing national random sample survey techniques. The current investigation differs from this methodology in two important respects. First, the research site is a particular urban community. Second, the sampling design randomizes respondent selection, but within purposefully chosen neighborhoods that differ systematically in social and economic composition. A high sampling rate within each neighborhood allows the investigators to characterize each neighborhood environment. This design is believed to be unique in the study of voting behavior in that it provides for direct, comprehensive measurement of the social, attitudinal, economic, and political characteristics of the voter's immediate social milieu. The data set resulting fom these procedures provides analytic opportunities unavailable in the usual national election studies. In the determinants of voter choice or policy attitudes, measurements of respondents' social contexts may be introduced directly into the analysis. The relative importance of individual properties compared with the political influence arising from the social composition of the respondent's neighborhood can be assessed. For example, a body of evidence suggests that vote choice by working class respondents is heavily influenced not only by their working class membership but also by the concentration of working class families in their living areas. This classic effect of social context on voting behavior can be specified in extraordinary detail with the data set now available from South Bend. What mechanisms of social interaction drive this observed amplification of working class influence on individual members of the working class? Is it the composition of their political discussion networks? Is it the composition of their work groups? Is it the homogeneity of other social cleavages within their neighborhoods - - race, religion, ethnicity? Such questions can be raised, and given answers in first approximation, with these data.
