Guidelines
Suggest edit

Keep hypothetical outcome plot playback slow enough to perceive individual samples

For trend judgment in ordered-time displays, use a cognitively trackable animation rate on hypothetical outcome plots to improve fidelity and mitigate loss of sensitivity from over-fast playback for novice public audiences.

  • purpose:refine
  • basis:empirical
  • task:trend
  • time:ordered-time
  • quality:fidelity
  • lever:encoding
  • operator:uncertainty
  • temporal-pattern:dynamic
  • literacy:novice

advice

Slow the HOP to a trackable pace

Set hypothetical outcome plots to a rate that still lets viewers cognize individual samples. For example, use a regular pace around 400 ms per sample rather than a very fast pace around 100 ms per sample when animating uncertain lines for ambiguous trend judgments.

reason

Preserve the benefit of animated sampling

Very fast playback can turn the animation into a blur of change rather than a readable stream of sampled outcomes. When viewers can still pick up individual samples, the animation more reliably supports the uncertainty judgment.

Mechanism: A trackable rate preserves viewers’ ability to integrate repeated outcomes over time, while over-fast playback weakens that readout and reduces the consistency of the benefit.

Evidence: Regular-rate line HOPs with 400 ms per sample produced lower JNDs than static line ensembles, while fast HOPs with 100 ms per sample showed only a smaller and unreliable advantage. The paper interprets this attenuation as a sign that very fast playback can outrun viewers’ ability to process the animated samples (Kale et al., 2019).

Notes: The paper treats 400 ms as a supported starting point, not a settled universal optimum.

context

Use when the chart already relies on animated samples

  • User Goal: Infer which trend is more likely from a noisy time series.
  • Task: Integrate multiple hypothetical outcomes over time.
  • Data: Quantitative ordered-time series with uncertainty shown as repeated sampled outcomes.
  • Chart Setting: The chart already uses an animated HOP instead of a static uncertainty view.
  • Audience: Untrained readers who need to notice variation without training.
  • Success Criterion: The HOP improves judgments consistently across viewers.

exceptions

Do not use when viewers will not watch the animation

Break it when: The setting cannot rely on viewers to watch several frames of animation. Why: Even a well-paced HOP depends on temporal attention, so the benefit disappears if the animation is not actually seen.

costs

Trade off speed for perceptibility

Sacrifice: Slower playback shows fewer samples per second. Risk: If the animation is slowed too much, viewers may become impatient. Mitigation: Slow the HOP only enough to preserve recognition of individual samples.

mistakes

Do not compress frames just to show more samples faster

Mistake: Shortening HOP playback to a very fast rate so more frames fit into less time. Why it fails: The performance gain weakened at 100 ms per sample, so extra speed can erase the benefit of the animated uncertainty display.

check

Compare the planned speed against a faster version

Failure Sign: Speeding up the HOP stops improving decisions or lowers confidence on ambiguous samples. Quick Check: Show the same ambiguous samples at the planned rate and at a much faster rate; if the faster version removes the advantage, playback is too fast. Stronger Test: Run repeated forced-choice judgments and check whether the evidence needed for correct decisions rises as the per-sample duration is shortened.

fix

Lengthen per-sample dwell time

  • Increase per-sample dwell time until viewers can still perceive individual observations.
  • Use a regular-rate HOP around 400 ms per sample as the starting point for this task, not 100 ms-like rapid playback.
  • Re-test the speed on ambiguous samples after any change.

References

Kale, A., Nguyen, F., Kay, M., & Hullman, J. (2019). Hypothetical Outcome Plots Help Untrained Observers Judge Trends in Ambiguous Data. IEEE Transactions on Visualization and Computer Graphics, 25(1), 892–902. https://doi.org/10.1109/TVCG.2018.2864909