Use low-density quantile dotplots when viewers must estimate probabilities precisely

Encode predictive uncertainty with a small, countable set of evenly spaced quantile dots when users must estimate interval probabilities from a static view. For example, replace a static density plot with a quantile dotplot of about 20 dots on each timeline row so viewers can judge the chance that an event happens before a threshold by counting or quickly grouping dots.

Low-density quantile dotplots turn a continuous predictive distribution into a small set of discrete outcomes that people can treat like frequencies. That makes interval judgments more concrete, especially near the tails, without requiring animated sampling or a large display.

Mechanism: A small number of dots lets viewers estimate risk by counting or subitizing small groups instead of inferring area from a smooth shape. This reduces response variability and supports better self-assessment of confidence.

Evidence: In the controlled experiment, the 20-dot quantile dotplot produced the lowest variance in probability estimates, about 1.15× more precise than the density plot, and also yielded the highest confidence; the 100-dot version performed similarly to density, and the stripeplot performed worse (Kay et al., 2016).

Notes: The density plot was rated more visually appealing, so the precision gain comes with an aesthetic tradeoff.

• User Goal: Make a quick decision based on the chance that an event occurs before or within a time threshold.
• Task: Estimate a cumulative or interval probability from a predictive distribution.
• Data: Continuous predictive uncertainty for a single future event, often with skew.
• Chart Setting: Static small-screen timeline rows with limited space and little time for interaction.
• Audience: Everyday non-expert users with limited statistical training.
• Success Criterion: Probability estimates are more precise and users feel confident about them.

Break it when: Visual appeal is the primary success criterion and a small loss in estimation precision is acceptable. Why: The density plot was more visually appealing while being only slightly less precise.

Sacrifice: You give up some visual smoothness and some fine-grained density detail. Risk: If you add too many dots, users stop reading the display as countable outcomes and the precision advantage disappears. Mitigation: Keep the dot count low enough that the display still supports quick grouping and selective counting.

Mistake: Packing the display with many dots or stripe-like marks. Why it fails: The display starts behaving like a continuous area or opacity encoding, so viewers no longer get the countability benefit.

Failure Sign: Users give widely scattered probability estimates or say they are judging the filled shape instead of the dot counts. Quick Check: Show the same predictive scenario in your current density plot and in a low-density quantile dotplot, then ask several target users for the chance of arriving before a threshold; keep the version with tighter estimates and higher confidence. Stronger Test: Inspect the stacks of dots; if vertical groups often exceed about five dots, the plot is too dense to support fast subitizing.

• Reduce the number of displayed outcomes to a small set, around 20 rather than 100.
• Space the dots so most vertical groups remain small enough to be quickly grouped.
• Replace stripe-like or very dense discrete encodings with a low-density quantile dotplot.
• Generate the dots from evenly spaced quantiles rather than random draws so the display is stable from one instance to the next.