Friday, August 30, 2019

Why we don't visualize uncertainty

There's a whole chapter about uncertainty in How Charts Lie, —I've just adapted part of it for The New York Times—and I mentioned techniques to visualize it in The Truthful Art, so it should be clear I've been a bit obsessed with the topic for a while.

That's why I was so happy when Jessica Hullman told me she's written a paper to explore why designers, journalists, and even scientists, are often reluctant to reveal uncertainty in their graphics.

The results of her survey are relevant, but I found the responses themselves even more interesting:
Multiple interviewees and survey respondents implied that most viewers who they created visualizations for did not require specific information about process or uncertainty to trust that a signal is valid [...] Some authors described trust as a pervasive default in visualization-based communication. As one industry interviewee described it “There’s a participation in trust with the system produced by the information, so wherever that comes from. Most people will trust the doctor, not necessarily because the information itself was trustworthy, but because the doctor was” (I8). In contrast to the seemingly rational expectation that uncertainty would play a role in fostering trust, the same interviewee described how a priori trust is instead a necessary precondition to presenting uncertainty: “I would say that you want trust established before you show uncertainty... My hypothesis would be that it [uncertainty communication] may have no effect for trust development.”
Showing uncertainty “may have no effect for trust development”. That's something waiting to be empirically tested.

Anyway, please do read Jessica's paper, and don't miss her talk at Chi Data Viz (YouTube). Also, Jessica will be the main speaker in my own VizUM Symposium, on December 12. Attendance is free!