Spatial distribution of candidates and users in VAAs

VAAs are tools primarily oriented on their users. However, they also may provide unique data for scientific research. We use the answers of candidates and parties in VAAs together with the VAAs users' answers to assess the positions of the political actors and voters in a political space. We deploy several methods to model the political space: our research is based on distance matrices, which we later project into lower dimensional space using (weighted) principal component analysis. We chose these methods as, among other things, they are useful because they allow modelling of several similar VAAs (elections) together (using common model, or by projection of a VAA – its candidates, parties and users – into a model created from another VAA). We also suggest a couple of charts that show the models well. Our method is primarily data-driven (exploratory). We start with the data and create the political space based on it (as an opposite to model-driven approach when the model is set first). There are several practical implications of our research, too. For political actors, the models show closeness of the candidates, thus revealing ones that are competing for the users of the VAA. For VAAs authors, the insight is useful for construction of better tools (e.g., a tool that differentiate well among the candidates or parties). Our research uses data from several dozens of VAAs covering several countries (the Czech Republic, Slovakia, Hungary, etc.) and several different types of elections (parliamentarian, presidential, regional, municipal).