New Retrospective Approach for Voting Advice Applications’ Design Methodology

Voting Advice Applications (VAA) are tools designed to help voters with their decision in an electoral process. These applications are generally based on a questionnaire structure and user responses. As a result, they provide a ranked list of parties or candidates that are closer to the voter’s political positions. In this paper we identify the two most used design methodologies for this kind of applications and discuss their limitations. Namely, the dependence that these approaches have on the prior existence and specialized analysis of proposals and manifestos of political candidates or parties. We conclude that these methodologies, focusing on future projections made by political agents, discard their past performance and, as such, are not suitable for countries where there are inconsistencies between political practice and the ideology propagated in electoral programs and manifestos. Therefore, this paper presents a VAA, Bom Partido (“Good Party”), which follows a retrospective methodological approach which is easy and accessible to replicate. Aimed at the Brazilian party-political context, this project is based on replacing the question “Which party/candidate proposals best correspond to my political positions?” by the question “Which party/candidate acted politically in the way that best corresponds to my political positions?”. Likewise, in addition to contributing to the critical study of VAA design methodologies, this project suggests a new design approach for these applications that potentially increases the reliability of its results and whose development method can be replicated step by step.