Electoral Geography

Geographical Effects on Electoral Behavior

There is a long history of debate about whether such geographical effects on electoral behavior do in fact exist. Opponents of the electoral geography hypothesis have contended that once all the variables affecting individual electoral behavior have been fully identified and properly specified, there is no role for geographical and contextual variables to have any independent effect. Furthermore, research on the existence and nature of neighborhood effects depends on the existence of usable neighborhood data: As a result, it is not necessarily easy to investigate the various levels at which they may exist.

Neighborhood effects in voting patterns were first demonstrated in 1937 in Sweden by Herbert Tingsten, who showed that the more working class the area, the more likely the working-class voter was to cast a socialist vote. Much subsequent work in this area has focused on the United States and the United Kingdom, perhaps because the political effects of the links between electoral geography and political representation are more immediately evident under majoritarian electoral systems. Support for the electoral geography hypothesis has been more extensively demonstrated in the work in the UK, over many years, of Ron Johnston and Charles Pattie, demonstrating that, in the words of William Miller (1977), “people who talk together vote together.” These effects are not, however, uniform. They are, for example, more powerful in “working-class” neighborhoods than in “middle-class” neighborhoods, probably because the social circles of salaried workers are likely to be wider and more geographically varied than those of residents of poorer neighborhoods.

The impact of neighborhood effects on patterns of political representation is affected not only by the formation and evolution of social communities but also by the geographical definition of electoral communities achieved by the delimitation of electoral boundaries. The legislation, which defines how delimitation and redistribution takes place; the independence or otherwise of the authority that undertakes it; and the attempts of political parties and movements to influence it in their favor are all relevant factors. It is not possible for redistribution to produce outcomes without a bias of some form. The inherence of geographic factors in redistribution is shown in the analysis of Bernard Grofman and others (1997), breaking down bias into three component parts: (1) bias resulting from malapportionment, (2) bias resulting from different levels [Page 748]of turnout, and (3) bias that results from the geographic distribution of party vote shares. Although requirements on a delimitation authority to control or eliminate malapportionment through a stipulation that the electorates of all seats be equal within a stated small tolerance level are now common, statutory exceptions exist for sparsely populated areas, for example, Norway and Western Australia. In South Australia, boundary revisions are also required to take into account the geography of party or grouping vote shares and ensure that the party or grouping with more than half the two-party preferred vote sees this reflected by an elected majority in the state legislative council.

...