From Information to Choice: A Critical Inquiry Into Visualization Tools for Decision Making
Emre Oral,Ria Chawla,Michel Wijkstra,Narges Mahyar,Evanthia Dimara
DOI: https://doi.org/10.1109/TVCG.2023.3326593
2023-08-02
Abstract:In the face of complex decisions, people often engage in a three-stage process that spans from (1) exploring and analyzing pertinent information (intelligence); (2) generating and exploring alternative options (design); and ultimately culminating in (3) selecting the optimal decision by evaluating discerning criteria (choice). We can fairly assume that all good visualizations aid in the intelligence stage by enabling data exploration and analysis. Yet, to what degree and how do visualization systems currently support the other decision making stages, namely design and choice? To explore this question, we conducted a comprehensive review of decision-focused visualization tools by examining publications in major visualization journals and conferences, including VIS, EuroVis, and CHI, spanning all available years. We employed a deductive coding method and in-depth analysis to assess if and how visualization tools support design and choice. Specifically, we examined each visualization tool by (i) its degree of visibility for displaying decision alternatives, criteria, and preferences, and (ii) its degree of flexibility for offering means to manipulate the decision alternatives, criteria, and preferences with interactions such as adding, modifying, changing mapping, and filtering. Our review highlights the opportunities and challenges and reveals a surprising scarcity of tools that support all stages, and while most tools excel in offering visibility for decision criteria and alternatives, the degree of flexibility to manipulate these elements is often limited, and the lack of tools that accommodate decision preferences and their elicitation is notable. Future research could explore enhancing flexibility levels and variety, exploring novel visualization paradigms, increasing algorithmic support, and ensuring that this automation is user-controlled via the enhanced flexibility levels.
Human-Computer Interaction