7
The first test of the claim that governments are easier to dismiss under plurality rules is
found in a study of the effects of incumbency on the electoral fates of governments (Rose
and Mackie 1983). It employed data from 1948 to 1979, from 18 countries. The analysis
relied only on bivariate comparison. Electoral system effects were not estimated directly,
but rather by proxy in three cultural categories. The first was made up of the Anglo-
American democracies including the United States. The others were Scandinavia and
Continental Europe (including France, which had a proportional system for only part of the
period). Coalitions under proportional representation were less likely to lose votes overall
because of vote ‘jostling’: that is, while one party might lose votes, another coalition
partner could gain votes to compensate.
In Continental Europe, over half of the elections in the period saw some government
parties gain, and others lose votes. Rose and Mackie distinguished between three scenarios.
The first, where an entire government remains in office, whether a single party or every
party in a coalition, made up just over 61 per cent of Anglo-American elections, 51 per cent
in Scandinavia, and 46 per cent in continental Europe. Defined in this way, Anglo-
American governments in mainly plurality systems were more likely to remain in office.
However the opposite scenario of complete dismissal of every single party applied to 39 per
cent of Anglo-American democracies, 25 per cent of Scandinavian, and only 5 per cent of
Continental democracies. In the middle category of ‘reshuffle’, one or more incumbent
parties remained in office as part of a new coalition (Rose and Mackie 1983, 134). But the
authors made to attempt to assess the extent of ‘reshuffle’: they therefore could treat a case
in which 90 per cent of a government might be replaced as the same as one in which only
10 per cent of the composition of a government changed between parties.
Convention