CORDIS - EU research results

Industry and Doctoral Training in Vote Advice Applications

Final Report Summary - VOTEADVICE (Industry and Doctoral Training in Vote Advice Applications)

VOTEADVICE is a collaboration between the University of Exeter (academic partner), Kieskompas (industry partner), a developer of online voting advice applications (VAA), and, Koç University (associate partner). The scientific objectives of VOTEADVICE are to produce research related to how new technologies can be used to engage and inform citizens as well as how data generated by these technologies can be used to study social and political behaviour. We focus specifically on VAA as a civic technology application. VAA match voters to political parties based on issue positions. We help generate new insight into the use of these applications as well as develop techniques for analysing the data to answer questions of interest to social scientists. A major output of the ITN is training three early career researchers and their PhD theses.

Research Activities:
Our main research activities, with our industry partner and within the PhD theses of the ECRs, focused on the following: (1) data collection through the development and implementation of VAA; (2) data collection through experiments (including use of eye-tracking) and (3) analysis of secondary sources of data. These research activities and the related key findings are discussed below.

1) The VAA generated data resulted from our collaboration with industry partner Kieskompas whereby we developed VAA tools for elections and then utilized the data to better understand electoral behavior. Additionally, working with our industry partner, we have developed coding schemes for political party placements on issues across six electoral campaigns in 5 countries. For this coding scheme, it was necessary to identify the main policy issues that were salient in the campaigns and through coding manifestos and other party documents place the parties on these same issues. Together, we have a rich source of data on political party issue positions in each country and the policy preferences of a sample of citizens who used the VAA in each election. These data were used by ECR Popp in her thesis measuring the congruence between political parties and citizens across Europe. VAA data were also used in ECR Zucca’s in her thesis examining information effects in the new media ecosystem. VAA generated data around the time of the UK Referendum vote was the basis of a scientific publication.

These tools not only helped voters make decisions in the election (see impact section below) but also generated data to help us understand political preferences in elections. The VAA generated data are particularly important in this respect because they tend to capture, in larger numbers than national probability samples, supporters of small parties. Working with our industry partner, VOTEADVICE researchers also developed and implemented post-stratification weighting scheme to use with VAA generated data. This post-stratification weighting algorithm for VAA data sets to allow further analysis by the Kieskompas team.

2) We also conducted two types of experiments. First, eye tracking equipment was used to examine study participants reactions to visual stimuli (pictures of groups of elected politicians). These experiments formed the basis of ECR Horvath’s PhD theses. Second, VAA application users were also asked to participate in a follow up survey to capture their characteristics (e.g. age, gender, partisan attachments). In this follow-up survey, we also conducted question wording experiment used by ECRs Horvath and Zucca in their theses.

3) Finally, existing observational data source (British Election Study) was used to validate findings across the other data collection efforts.

Key Findings: Results
Key findings in other publications can be found on the webpage and below we give key findings from the PhD theses of the project researchers:

1) In ECR Horvath’s research an eye tracking study explored attentional engagement with candidate gender, whereas online surveys were used to understand how political engagement might depend on gender balance in political groups. He found that as the number of women politicians demonstrated a growing presence more women in the electorate were engaged in politics. The studies demonstrated that political role models are powerful in engaging more people with politics by improving citizens’ sense of confidence about their own political skills.

2) Using an online experiment on VAA Election Compass UK 2015, Zucca finds that measures the patterns of opinion change after the exposure to online content. Using panel data, re-interviewing Election Compass UK 2015 VAA users in a follow-up survey, she finds that the VAA produces useful information for voters but that this does not alter preferences.

3) Using VAA generated data from the 2009 and 2014 European Parliamentary Election VAA (EUVox), Popp finds that, because niche parties focus on specific issues, supporters of these parties tend to be more congruent with the parties than large, more centrist parties, and can easily identify the parties that stand closer to their own political preferences. For issues, the highest congruence between parties and voters was on the issue of European integration.

Social and Economic Impact:

We identify our main social and economic impact for two groups of beneficiaries: citizens and social science researchers. In terms of citizens, VAA are a class of civic technology that are meant to inform and engage citizens in the electoral campaigns. We informed and engaged citizens through the design, development and implementation of six VAA: Portugal ( ), Spain ( ), Romania ( , USA ( ) and United Kingdom ( We also supported the creation of Election Compass Wales ( Across these six VAA tools we helped over 240,000 citizens used the VAA tools to inform their electoral decisions. Our research thus contributes to enhancing the quality of representative democracy through the implementation of civic technologies that are intended to allow citizens to access high quality information. Our impact also was extended through our work with media organisation Localworld to promote these tools in UK elections.

In terms of researchers, one of the major outputs of VOTEADVICE has been the application of techniques that will allow VAA data to be used for research purposes. Basic modes for sampling and data collection have been challenged by the evolution of mobile communication devices, which includes new analogue as well as the virtual modes of communication. Survey researchers needs to discuss the future of survey research and the development of new approaches regarding the basics of sampling and survey methodology, as well as innovations in statistical theories and the standardization of practices. With our industry partner, we have developed and applied novel weighting methodologies to adjust online and voluntary samples. These weights can then be used for the analyses of these data to reflect representative samples of the population. In addition to the tools for analysis, one of our project priorities is to make Voting Advice Application-generated data accessible for social science research. This is currently being developed in partnership with Kieskompas. Details can be found at