How should we visualize data? Lessons learned from reviewing high quality graphs in the Socius Special Collection Data Visualization

The Socius Special Collection: Data Visualization aims to publish visually appealing displays of information that illustrate interesting findings and are self-explanatory without reference to the text. We reviewed the corresponding peer-reviewed, high-quality visualizations pursuing three research questions: What types of graphs are used? What errors do people make when visualizing data? What variation can we find across the collection in terms of attractiveness, complexity, innovation and understandability? In total, we analyzed and rated 137 data visualizations published between November 2, 2017 and March 26, 2024. We find that authors most commonly use standard plot types, that flaws are surprisingly frequent and that there is a intriguing variation in terms of perceived attractiveness, understandability, and innovativeness across visualizations. We provide key takeaways and a checklist that may help to improve visualizations in the future.