Data Videos: Human Stories Beat Statistics for Donations, Study Finds

In a paper published on arXiv (2607.03445), researchers Elsie Lee-Robbins and Eytan Adar from the University of Michigan explore how data visualization design affects emotional engagement and prosocial behavior. They tested three narrative strategies in short data videos about a humanitarian crisis: pure data-driven (charts and statistics), pure human-driven (photographs and individual stories), and a mixed narrative combining both. Participants were given the option to donate real money as part of the study, providing a concrete behavioral measure beyond self-reported feelings.

The results were striking: human-driven narratives elicited the highest donations, while the mixed narrative unexpectedly led to the lowest donations, even below the data-only condition. This suggests that blending statistical information with personal stories may trigger statistical numbing—where numbers dampen emotional response—rather than reinforcing empathy. The authors caution that common design instincts to "add data for credibility" may backfire when the goal is to inspire care or action. The study has direct implications for journalists, NGOs, and designers communicating crises like natural disasters, war, or pandemics.