Political scientists should reject Richard Rose's advice in PS (1991) to abandon the use of swing, in favor of his own alternative measures of electoral change, in three or multiparty electoral contests. Given the ubiquity of such electoral contests, this advice has a great potential influence on future election studies. However, a balanced judgment requires subjecting alternative measures to equal scrutiny. This note will do this by bringing relevant evidence to bear on the actual performance of swing and Rose's alternative, as a measure of variations in electoral movements between different areas. The findings are most revealing and in the final section of this note reasons for them will be put forward.

Rose maintains, in his critique of David Butler and Stephen D. Van Beek (1990), that swing can only be meaningfully used as a measure of electoral change between elections where virtually the entire vote is split between two parties. On this basis he claims that “swing is not applicable in nine-tenths of competitive party systems because two parties do not monopolize the vote.” As Rose shows, reliance on swing can ignore important changes in third party votes, which may often be an important determinant of the swing between the two main parties.