Party responsiveness and voter confidence in Australia

ABSTRACT

Numerous studies have shown that Australians have little confidence in their political parties. This article presents the results of a study investigating whether the responsiveness of Australian parties to what their voters want drives this lack of confidence. It analyses two aspects of party responsiveness: programmatic responsiveness in electoral manifestos and perceived responsiveness that centres on Australian voters’ assessment of how well their parties meet their demands. The analysis finds that programmatic responsiveness has no significant influence. Instead, how Australians perceive their parties to be responsive has a modest effect on their confidence in those parties. The study suggests that, however, it is incumbency which has the most powerful effect on voter confidence.

许多研究显示，澳大利亚人对本国的政党没多少信任。本文探讨了澳大利亚政党对选民期望的反应是否造成了这种信任的缺失。作者分析了政党反应的两个方面：竞选宣言中的程序性反应，以及澳大利亚选民认为各政党是否对其要求做出了反应。研究发现，程序性反应的影响并不显著。而澳大利亚人觉得其政党是否对其需求做出反应则有一定的影响。作者认为，执政者的所作所为对选民的信任影响最大。

KEYWORDS:

• Political parties
• responsiveness
• confidence
• Australia

Acknowledgements

The author would like to thank Cosmo Howard, Onawa Promise Lacewell, and Duncan McDonnell as well as the anonymous reviewers and the editors for their helpful comments.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Annika Werner is a research fellow at the Centre for Governance and Public Policy at Griffith University. She received her PhD from Humboldt-University, Berlin, in 2015. Previously, she worked with the Manifesto Project (MARPOR) at the WZB Social Science Center Berlin.

ORCiD

Annika Werner http://orcid.org/0000-0001-7341-0551

Notes

1. In contrast, the role of party behaviour for citizens’ satisfaction with democracy has been established. Democratic satisfaction is, however, a separate part of the concept ‘support for democracy’ (Easton 1975; Norris 2011) and the results of these studies cannot be transferred directly to confidence in parties. They need to be tested again, as this study undertakes.

2. The following data sets are combined: Jones, McAllister, Denemark, and Gow (1993); Jones, McAllister, and Gow (1996); Bean, Gow, and McAllister (1999); Bean, Gow, and McAllister (2004); Bean, McAllister, Gibson, and Gow (2005); Bean, McAllister, and Gow (2008); McAllister, Bean, Gibson, and Pietsch (2011); Bean, McAllister, Pietsch, and Gibson (2014).

3. The most common survey question used to measure confidence in parties is the question of respondents’ trust in parties (Bean 2015; Dalton and Weldon 2005; Bean et al. 2014). The problem with this question is that ‘trust’ is an ambiguous term (Delhey et al. 2011). Furthermore, trust belongs to the private sphere and ‘is a feature of personal relationships based upon first-hand experience and knowledge’ (Zmerli et al. 2007: 40). Confidence, on the other hand, is part of the public sphere and is built upon second-hand sources, e.g. the media. Thus, I do not use the term ‘trust’ but ‘confidence’.

4. These differences are often discussed in a framework of data quality and which source for party positions are ‘better’ (e.g. Delhey et al. 2011: 121). An alternative view is to see these sources as different conceptualisations of party positions. Party platforms, speeches and other sources produced by parties and politicians are possibly best understood as messages ‘sent’, the perceptions of voters and experts are like messages ‘received’. Interpreted this way, the discrepancy between the message and the perception of these messages are not indicative of data quality issues but can be appreciated as theoretically meaningful.

5. See Appendix for exact question.

6. Australia is a special case as party programs were not issued before 2010. Before this time point, the central party leaders’ speeches at the beginning of the election campaign had the equivalent function (Merz and Regel 2013).

7. The data availability of the Manifesto Data limits the parties that can be included into the analysis to the Liberals, Nationals, Labor, Greens and Democrats. Smaller parties like One Nation are not included in the data collection.

8. The resulting category attributes can be found in Table A1.1 in the Appendix (see also Franzmann 2009).

9. A Brant test provides evidence that the parallel regression assumption is violated for nine of 15 explanatory variables.

10. One could assume that there is a close relationship between the party a voter supports and the effect of the government status, as the Liberals have held the government for the majority of the time frame. However, a test for collinearity shows that all independent variables have a Variance Inflation Factor (VIF, which shows how much of the inflation of the standard error could be caused by collinearity) and Tolerance around 1. Thus, there is no indication of such problems. See Table A1.3 in the Appendix.

11. Democrats are not shown.

12. This was confirmed by tests including interaction effects between perceived responsiveness and party.