Our purpose in this chapter is two-fold. One objective follows on from the theme of the last chapter and it is to examine whether and how local voting can be used to make observations about the national electoral mood. As we saw, there are many instances where local voting is dominated by local issues and local responses to national influences. That still leaves a great deal of behaviour which is national in orientation, with local voters using local elections as a vehicle to express their views about the wider political situation (Curtice and Payne 1991). The largest number of local elections occur, of course, in the first week in May but though these are plentiful they are only partly suited to our purpose. Although these election results can be used to note annual rises and falls in the performance of the parties, they offer few clues for election forecasting unless, as in 1983, 1987 and, perhaps 1991, a general election is thought to be imminent. Moreover, because they only take place once a year, they cannot help in the frequent mapping of trends in behaviour in a manner similar to that undertaken by opinion polls. What we require for this task is local votes which are cast throughout the year. That, in effect, means using the results from local government by-elections.
That brings us to our second purpose. Before we can test the usefulness of local by-elections as surrogate indicators of wider political change more needs to be discovered about them. First, some of the circumstances giving rise to local by-elections need to be assessed. In most instances they are caused by much the same events as are parliamentary by-elections, primarily death and resignation. In some cases, however, local by-elections can occur for reasons which have no parliamentary equivalent and the circumstances of the contest can sometimes have a direct bearing on the outcome. A second consideration is when and how frequently by-elections occur. This is critical because if they are to be used as the basis of a model of electoral behaviour they need to be distributed fairly randomly throughout the year. If they were concentrated in time then they would suffer the same sorts of weakness as do the May elections. A third issue has to do with the spatial distribution of local by-elections. By-elections concentrated in