Empowering Citizens in the Digital Age : a systematic evaluation of voting advice : applications and best practices for their design

Voting Advice Applications (VAAs) are online tools that match voters and parties based on common positions on a series of issues. Starting as tools for political education, VAAs have nowadays become relevant political actors. Besides providing information, they also raise electoral turnout, improve political knowledge and influence party choice. The degree in which they do so depends on their design. In this thesis, I view the design of VAAs as a process and focus on two steps: the questionnaire and the visualizations. For the questionnaire I focus on two aspects: the scales used to position users on the political map and the formulation of the questions. I assess the scales using various data-reduction techniques and rank them on unidimensionality, quality and reliability. I find that most scales score insufficient, though the score depends on the construction method used for the scale. The cause of these problematic scales is that VAA users often apply simplification methods. This means they do not always understand the question or use the response categories as intended. This results in problematic scales, resulting in a political map that is difficult to interpret. Also, I find that altering the questions in the main questionnaire to have either a positive or negative formulation, not only influences the responses of the users but the match between the user and the party as well. For the visualization, I run an online experiment in which I ask users to answer questions related to various kinds of visualization. I find that not only do they have difficulty to finish some basic tasks, they also have diverging interpretations of popular VAA visualizations. My main conclusion is therefore that the design of VAAs cannot be neutral. Also, the underlying information used to calculate the match and visualize the political map are often troublesome. Yet, I also show designers can use simple methods to improve their VAAs. This is important as VAAs are likely to become even more popular than they are already at this moment.