How party‒issue linkages vary between election manifestos and media debates

Abstract

Existing research on party behaviour has largely focused on the drivers of issue salience in direct party communication. However, less is known about party‒issue linkages in election campaigns covered by the mass media, from which most voters get their information about party positions. Against this background, this article explores how two important drivers of issue salience in direct party communication – issue ownership and systemic salience – play out in the media. Based on considerations about the news value of specific party‒issue associations, one would expect both relationships to be particularly important in the media. Despite substantial similarities in party‒issue linkages across platforms, a comparison of manifestos and newspaper content reveals evidence for this claim. In particular, smaller parties are hardly covered in the news on issues they do not own, while large parties are especially covered on salient topics. These findings contribute to our understanding of issue competition in mediated environments.

Keywords:

• Salience
• issue ownership
• systemic salience
• election campaigns
• party manifestos

Acknowledgements

We have presented earlier versions of this manuscript in workshops at the Chair of Comparative Politics at the LMU Munich and the Centre for Civil Society Research, WZB in Berlin. We thank the participants at the workshops and the anonymous reviewers for their valuable comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

Swen Hutter gratefully acknowledges funding by the ERC project Political Conflict in the Shadow of the Great Recession and by the Volkswagen Foundation.

Notes on contributors

Tobias Schwarzbözl is a doctoral researcher at the Geschwister-Scholl-Institute for Political Science, LMU Munich. He studies party competition and political representation. [[email protected]]

Matthias Fatke was a postdoctoral researcher at the Geschwister-Scholl-Institute for Political Science, LMU Munich. He studied party competition, political participation and political trust before joining the city of Stuttgart. [[email protected]]

Swen Hutter is Lichtenberg-Professor in Political Sociology at the Freie Universität Berlin and Vice Director of the Centre for Civil Society Research, WZB in Berlin. His main research interests include party competition, civil society, political participation and cleavage structures. [[email protected]]

Notes

1 The data used in this article originated from the research project ‘National Political Change in a Globalizing World (NPW)’ (Kriesi et al. 2008, 2012). The data can be obtained from the research project’s authors.

2 Overall, these six issue categories account for 69% of the issues reported in the media and 61% in party manifestos.

3 The study includes all parties that are also coded in the Comparative Manifesto Project that fulfil two additional requirements. First, parties must obtain more than 5% of the votes in the election under study. This criterion guarantees that the overall results are not affected by very small parties, which are largely irrelevant to the logic of party competition. Second, a party is only included when it is to some degree visible in the public debate. Therefore, the mass media data set must contain at least 20 core sentences for a party related to any kind of political issue. This threshold is necessary to secure the accuracy of the salience measure; extremely low overall numbers of core sentences do not allow us to infer a party’s emphasis on a specific issue. It is important to note that most parties either meet both criteria or fail to meet even one of the two.

4 In fact, the significance of the interaction term in Model 4 does not hold up when we dichotomise the variable.

5 Marginal differences for systemic salience are smaller than those for ownership because the variable, when standardised, ranges from ‒1.33 to 3.99. Rescaling the variable to the range of the ownership variable (from 0 to 1) results in estimates for the marginal differences (0.12) similar in size to those for ownership.