Putting voters in their places: Local context and voting in England and wales, 1997

Most analyses of voting behavior in the United Kingdom treat the electorate as so many atomized individuals whose decisions are made apart from any social context. With few exceptions, for example, there is no treatment of the household context within which most people are continuously socialized (but see Zuckerman, Fitzgerald, and Dasovíc, in this volume). And although there has been more attention to neighborhood and regional contexts, much written about these has been contested, and there is no consensus that people are influenced by such contexts. Indeed, Dunleavy (1979) challenged the notion of neighborhood effects on both procedural and theoretical grounds, and McAllister and Studlar (1992) criticized studies of regional variations in voting during the 1980s on the grounds that the models tested were under-specified. In better-specified models, they claimed, geographical-spatial differences become insignificant. In the view of some, circumstantial evidence strongly suggests that local contexts (at a variety of scales) influence voter attitudes and behavior, and that many individual characteristics associated with such behavior are themselves locally stimulated, if not created (see M. Johnson et al. 2002; Marsh 2002). Most models treat independent variables such as social class as universal characteristics, however- as meaning the same thing to all people in all places, whatever their backgrounds and local situations. Social class is not a pre-given characteristic, however. Rather, its meaning is learned in context-in the home, neighborhood, school, workplace, and so on. Class positions are interpreted according to how we learn about them and from whom, and many of the sources live in or close to our homes. There is a probable issue of simultaneity involved in local contexts at some scales, as some people may choose where to live (though not their parents!) on the basis of their perception of neighborhood characteristics and quality. In such cases, neighborhood effect will not influence their attitudes and behavior; rather, neighborhoods are selected that are consonant with those perceptions. Behavior precedes residential choice rather than vice versa. Cross-sectional analyses here cannot separate out the relative importance of the two processes. If the expected pattern emerges, however, it is for later studies to evaluate why. One component of local effects, therefore, consists of home and neighborhood and the people we interact with both regularly and frequently. Such neighborhood effects are just one of the ways in which local contexts can influence voter behavior. Books and Prysby (1991) identified four such mechanisms: individuals' observation of local conditions; their contacts with neighbors and others in local social networks; exposure to information flows through local media; and contact with the mobilization and campaigning strategies of political parties and other inter ested actors. All four suggest that how we think politically, and how we vote, can be influenced by our local context, by the material circumstances that we observe there, and by the people, organizations, and institutions we interact with. To evaluate these arguments it is necessary to identify the processes at work, which is more feasible for some than others. For example, if we hypothesize that people are influenced through their interpersonal contacts, then studies of the flow of information through social networks-such as those by Huckfeldt and his collaborators (e.g., Huckfeldt and Sprague 1995; Huckfeldt, Johnson, and Sprague, in this volume)-may provide strong evidence of links between contacts, information flows, and voting decisions. For some of the other processes, however, uncovering direct links is more difficult-as with those that affect people's voting decisions are influenced by their observations of local material circumstances, such as local levels of unemployment. When determining which party or candidate to support, some voters may make an immediate decision based on habitual behavior patterns ("I always vote Labour"; "The Conservatives are the party for people like us"). Others may be subject to a range of influences, and no one cause may outweigh the others. Even if one does predominate, it may be difficult to discover it-assuming that the voter is aware of it her- or himself! In such cases, strong circumstantial evidence, set in a clear theoretical framework, may be the best we can get. Whatever the difficulties of identifying the reasons why people vote as they do, circumstantial evidence can set the context for more detailed investigations. Such evidence is usually found in the aggregate rather than at the individual level. It may introduce problems of both underspecified models and ecological fallacies, but it can provide compelling arguments that local contexts do matter, that people are not just members of universally defined categories, such as social classes, but are also socialized and politically mobilized in particular geographical- spatial milieus with their own characteristics and influences on how members behave: that is, people act differently according to the type of place they live in. That approach is adopted here with regard to voting at the 1997 general election in England and Wales, for which specially constructed information on local contexts was added to the data produced by the British Election Study (BES) postelection survey.1 The goal is to establish whether there were spatial variations in voting behavior that were apparently unconnected to individual voters' characteristics but were related to the type of neighborhood in which they lived. In this context, we are referring to geographical space, not social space, which is the focus of other chapters in this book. (See particularly the chapter by Huckfeldt, Johnson, and Sprague; also Baybeck and Huckfeldt 2002a, and Kotler-Berkowitz in this volume. Gimpel and Lay's chapter in this volume looks at local contextual effects for one population subset.) The analyses reported here place survey respondents in their local neighborhood context. This may be the same as their social space- that is, their social networks are contiguous with their immediate neighborhoods. Our argument, however, is that people may be influenced not only by people who live locally but also by other characteristics of their local milieu, such as the state of the local economy.