Party, policy – or both? Partisan-biased processing of policy arguments in direct democracy

ABSTRACT

How do party cues and policy information affect citizens’ political opinions? In direct democratic settings, this question is particularly relevant. Direct democratic campaigns are information-rich events which offer citizens the opportunity to learn detailed information about a policy. At the same time parties try to influence citizens’ decision procedure by publishing their own positions on the issue. The current debate on whether “party” or “policy” has more impact on political opinions has not yet yielded conclusive results. We examine the effect of policy arguments and party cues on vote intention in two Swiss referendum votes using panel survey data. To the simple dichotomous question of “party cues or policy information” we add an additional twist in asking how party cues affect the processing of policy information through processes of motivated reasoning. We find first that both, policy arguments and party cues, have an independent effect on vote intention. However, in a second part of the analysis, we find strong evidence for partisan-biased processing of policy arguments – that is, voters tend to align their arguments with their preferred party’s position. Our conclusions with regard to the democratic quality of these vote decisions are therefore ambivalent.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Céline Colombo is currently a postdoctoral researcher at the Political Science Department, University of Zurich, with a focus on Political Psychology and Behavior. She completed her Ph.D. at the European University Institute in Florence in May 2016. In her research, she studies citizen competence and political decision-making, mainly in direct democratic settings. More specifically, she is interested in citizens’ political knowledge, motivated reasoning, the functioning of elite-cues versus policy-arguments and deliberation in decision-making, integrative complexity of political thinking, as well as the link between deliberative and direct democracy.

Hanspeter Kriesi was born in 1949 in Bischofszell (Switzerland). He studied sociology at the Universities of Bern, Zurich and Chicago. He obtained his Ph.D. in sociology at the University of Zurich (1976), where he also did his Habilitation in sociology (1980). In 1984 he became a professor for collective political behaviour at the University of Amsterdam. In 1988, he went to the University of Geneva, where he taught as a professor of comparative and Swiss politics until 2002, when he took up the Chair for Comparative Politics at the University of Zurich. He was appointed to the Stein Rokkan Chair at the department in September 2012.

Notes

1. A range of other cues and heuristics might possibly play a role in political decisions, such as trust in government, status quo heuristic, or affective and likeability heuristics.

2. In Swiss direct democracy, parties are not the only elite actors which can provide information or cues, non-partisan information can come from initiative or referendum committees, civic society organizations or organized interest associations. However, as the final decision is always structured as a binary choice, where each side is supported by a coalition of actors, and at the same time the major parties issue a vote recommendation in every case, we decided to focus on party cues in this study. Party recommendations also figure much more prominently than cues from other organizations among the reasons Swiss voters give for their decisions in post-ballot surveys (own calculations, based on Vox data).

3. A detailed description of the two cases can be found in the online appendix.

4. The Italian-speaking part was left out of the analysis for cost reasons.

5. The results remain virtually identical when actual voting behavior reported in the second wave is used as the dependent variable.

6. The online appendix shows the key arguments of each campaign

7. As Table A3 in the online appendix shows low-knowledge scores decreased while high-knowledge scores increased from t1 to t2, which indicates learning effects during the campaign.

8. A Hausman test (Prob>χ² = 0.136) shows that the coefficients of our random-effects model are not systematically different from those of a fixed-effects model (for the time-varying covariates). Possible bias introduced through omitted individual-level time-invariant characteristic is thus negligible and the use of random-effects model is appropriate.

9. Table A4 in the online appendix presents the full models including an interaction effect between party orientation and party cue knowledge as an alternative way to display the moderating effect of cue knowledge. In both cases, the interaction effect between party orientation and party cue knowledge is statistically significant and positive. This means that the effect of party orientation on vote intention is significantly stronger for respondents who know their party’s position (marginal effect 0.17, p < .001) than for those who do not (marginal effect 0.06, n.s.). This underlines the robustness of the results presented in Table 1.

10. Furthermore, the participation rate is higher among respondents who know the party cue than among respondents who do not know it (Table A5 in the online appendix). This correlation is stronger in the complex corporate tax case. This suggests that party cues might facilitate participation in referendum votes.

11. Table A6 in the online appendix underlines these results: it presents the effect of party orientation on vote intention, excluding argument position. When not controlling for argument position, the effect of party orientation is substantially stronger. This indicates that a part of the effect of party orientation on vote intention acts through argument position, as a mediator variable.

12. Another possible limitation we would like to point out here however, is that the results of the referendums themselves may affect the partisan bias. As our dependent variable argument position at t2 is measured after the outcome of the referendum is known, we cannot exclude that the fact of being a winner or a looser of the referendum affects the estimated partisan bias. This is in itself an interesting question, which goes beyond the scope of this paper however.

Additional information

Funding

This work was supported by the Swiss National Center of Competence in Research (NCCR) Democracy.