![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
This article investigates the mobilization potential of online voter information tools known as “Voting Advice Applications†(VAAs). We argue that an observational approach utilizing survey data constitutes the best available method for causal inference where VAAs are popular—and we are thus most interested in VAA turnout effects—because randomized experiments are likely to run into double cross-over problems. We suggest that matching offers key improvements over existing methods to tackle self-selection into VAA use in observational studies. To improve confidence in selection on observables, we complement matching estimates with an extensive sensitivity analysis, including a placebo test. Empirically, we study the effect of smartvote, a popular VAA from Switzerland, on turnout in the 2007 Swiss federal election. We find that smartvote usage significantly increased the individual-level probability to vote. Our results suggest that smartvote was, on the aggregate, responsible for about 1.2 % points of the total tally with an estimated cost of nine Swiss Francs (7.5 U.S. dollars or 1.4 “Big Macsâ€) per additional vote. Promising as well, we find that the mobilization effect was more pronounced among younger voters. Our findings point to the value of VAAs compared to traditional get out the vote tactics.
Acknowledgments
We thank Jens Olav Dahlgaard, Jonas Israel, Jan Fivaz, Silke Goubin, Marc Hooghe, Georg Lutz, Fernando Mendez, Martin Rosema, Uwe Serdült, Jonathan Wheatley, Yiqing Xu, the journal editor, as well as the anonymous reviewers for valuable comments at various stages of this project. Earlier versions of this paper were presented at a 2018 seminar at KU Leuven, the 2015 EPSA Annual Conference, the 2014 ECPR General Conference, the 2014 ECPR Graduate Student Conference, and the 2014 Annual Convention of the Swiss Political Science Association.
Supplemental Material
Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2018.1526237.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1. A second problem is that the left-right self-placement (which is measured post-treatment) might be affected by post-treatment bias: VAA users might have placed themselves differently in the absence of VAA usage.
2. Formal sensitivity analysis constitutes another useful strategy to probe the plausibility of selection on observables (see, e.g., Rosenbaum [Citation2002]).
3. smartvote first appeared before the 2003 election, but the 2003 election study does not include an item on VAA usage.
4. Turnout in past elections and referendums constitutes a strong proxy for almost any concept linked to turnout and VAA use. Controlling for past turnout also helps to compensate for possible deficiencies resulting from the necessary adjustments to the measurement of political interest and political knowledge (see later discussion).
5. We consider VAA use unlikely to affect party identification, at least in the short run.
6. Polychoric correlations range from .14 (signature collection) to .49 (active party member).
7. Argovia, the two Appenzells, Basel-City, Basel-Country, Glarus, Grisons, Jura, Neuchâtel, Obwalden, Schaffhausen, Schwyz, Solothurn, Uri, Valais, and Zug.
8. The same argument holds if self-reported VAA use were affected by social desirability.
9. Refer to Online Supplemental Appendix E for the regression output and covariate balance statistics.
10. It should be noted that the model does not include 18-to-21-year-olds due to the accounting for turnout in the 2003 federal election.
11. None of the second differences of the VAA mobilization effect at different levels of education even approaches statistical significance (see Online Supplemental Appendix H).
12. Several of the second differences are statistically significant, including the differences between voters aged 45 and voters aged 60 above and (at the 10% level) between voters aged 30 and voters aged 75 and above (see Online Supplemental Appendix H).
Additional information
Funding
Notes on contributors
Micha Germann
Micha Germann, Postdoctoral Prize Fellow, Department of Politics, Languages & International Studies, University of Bath.
Kostas Gemenis
Kostas Gemenis, Senior Researcher, Max Planck Institute for the Study of Societies.