ABSTRACT
The Manifesto Research Group (MRG) project is regarded as a landmark in empirical political science, not only in its own right, nor simply because of its rich, publicly available data set, but also because it provides a wellvalidated standard for parallel types of study. The project has shown that it is possible to create reliable time series estimates of party preferences across a broad range of strategic policy areas by content analysing election manifestos. This has allowed us, inter alia, to map changes and trends in what parties see as important over time. (Budge et al., 2001) In recent years, computerisation of content analysis has also been employed not only to speed up the actual process of coding but also to enhance reliability. Examples of different approaches attesting to greater or lesser degrees of success now abound. (See; inter alia, Bara, 2001a; Bara 2001b; Laver, 2001; Laver and Garry, 2000; Laver et al., 2003). Much of this is related to developing accurate measures, not only of issue salience, but also of direction/ideological positioning, using ‘leftist’ or ‘rightist’ language and discourse.
