Visualizing Uncertainty

Design Insights

Dr. Loni Besancon emphasizes the importance of visualization design in data interpretation. She highlights the need to address the "cliff effect," where confidence drops sharply around the 0.05 mark, by creating visually appealing and space-efficient designs 1. Aesthetic appeal is crucial, as visually pleasing designs are more likely to be used. Dr. Besancon describes various visual styles, such as gradient and violin plots, which aim to convey uncertainty effectively 2.

We used a line to show the mean, and then in this rectangle, we would see a big chunk of green color that would actually represent the classical 95% confidence interval.

--- Dr. Loni Besancon

These designs are intended to reduce the cliff effect and enhance user engagement.

   

Data Interpretation

Interpreting visual data presents unique challenges, especially in avoiding binary conclusions. Dr. Loni Besancon advocates for visualizations that encourage nuanced interpretations, moving away from dichotomous thinking 3. She suggests that visual data can help people remember and compare information more effectively, fostering cautious conclusions.

We want them to not interpret all of this data that they have in a dichotomous fashion anymore.

--- Dr. Loni Besancon

Dr. Besancon also notes the potential for these visualizations in education, helping students understand complex data concepts better 4.