Agenda building : a time-series analysis of the challenging issue priorities of parties and media during the 1997 General Election campaign in the UK

Abstract

This is a study of agenda building during an election campaign, which is a dynamic process and hence calls for the application of time-series analysis. The data, from the 1997 General Election campaign in the UK., consist of repeated observations at regular time intervals. The variables are political parties and media actors (television and the press), whose daily attention to different issues and policy dimensions was content analysed and then quantified. While some simple methods can be applied to the quantification of changing attention to individual issues, a more sophisticated time series model can be applied only when data are aggregated to the level of policy dimensions. For the 1997 campaign, issues were grouped into nine separate policy dimensions. Instead of treating policy dimensions separately, a time-series cross-section (TSCS) design was applied, with policy dimensions as cross-sections and campaign days as time units. The main purpose of time series analysis is to establish Granger-causality between different variables, which have repeated observations on each cross-section. In order to accomplish this, separate regressions, with either parties or media as dependent variables, were carried out.