‘Visualisation Choices about Refugees Have Delimited Impacts Along Partisan Lines’ with William Allen (Southampton) 

4 December – 12:00 to 13:00

Data visualizations are indispensable to journalists, lobbyists, and scientists for communicating information. Growing interest in story-telling with data, as well as received wisdom about these outputs’ power, suggests that visualizations can change what viewers think. Yet there is little large-scale evidence for this, particularly on political issues, that also considers whether specific design features are responsible. I conducted a full-factorial experiment among 3,082 British respondents who viewed a visualization about refugee inflows to the UK between 2001-2020. While the underlying UK Home Office data remained identical, the visualizations varied in terms of chart type, dominant color, editorial framing, and disclosure of the source. Overall, realistic choices about these key dimensions had limited effects on attitudes and preferences. This raises questions about the perceived effectiveness of visualization—particularly on politically salient issues, and involving relatively uncontentious data—while opening further avenues for examining how people engage with information conveyed visually.