Visiting the question of communicating uncertainty in Marissa’s lab meeting, focusing on this excellent Science article (Spiegelhalter et. al. 2011). Noam and Jamie are leading the discussion, and suggested we pick a few favorite examples. None on my list really address the challenge of communicating uncertainty directly, though the first one probably comes closest. I think they do highlight some of the potential and the challenges of good data visualization though, which is certainly one of the essential building blocks to visual communication of uncertainty.
- Porcupine plot
- Data journalism video from Stanford, the best introduction I’ve seen, which hits upon some of the best-known examples.
- Success and failure of data storytelling Shan Carter, NY Times visualizations expert, gives a thought-provoking talk that leaves me wondering if we are asking too much of the visualization.
- Animated graphics in R by SVG, because easy and platform-independent interactive graphics are essential.
- ggplot2, because it’s not a revolution until it’s democratized.
- Tufte’s first book, which sits above my desk.
the missing second block
(Spiegelhalter et. al. 2011) does a less excellent job addressing the some of the more basic problems in communicating uncertainty (though the conditional probability example comes close). Uncertainty can be immensely counter-intuitive, as Mlodinow illustrates in one of my favorite popular non-fiction books. He doesn’t tackle visuals as a way of improving our intuition on the topic. I feel a proper treatment of the topic would start with some of the easily demonstrated fallacies around uncertainty Mlodinow discusses, and illustrate if and when visualizations can improve our accuracy.
References
- Spiegelhalter D, Pearson M and Short I (2011). “Visualizing Uncertainty About The Future.” Science, 333. ISSN 0036-8075, https://dx.doi.org/10.1126/science.1191181.