Guidelines
Suggest edit

Facet categories instead of coloring a single scatterplot when summary tasks face many dense groups

For grouped summary tasks with many categories or dense overlap, prefer small-multiple faceting over a single-view category-colored scatterplot to improve fidelity and mitigate overplotting errors for readers comparing groups.

  • purpose:select
  • basis:empirical
  • structure:small-multiples:use
  • structure:single-view:avoid
  • scope:grouped-result
  • density:dense
  • group-cardinality:many
  • quality:fidelity

advice

Replace one crowded color view with faceting

Facet categories into separate panels instead of keeping all categories together in one color-coded scatterplot when users must compare groups under high category count or heavy overlap. For example, use row faceting by category for maximum-finding or average-comparison tasks rather than a single scatterplot that distinguishes groups only by color.

reason

Why faceting beats the crowded color view here

When many points and many categories share one panel, category colors begin to occlude each other and summary judgments break down. Faceting removes that overlap cost, even though readers spend longer comparing across panels.

Mechanism: Separating categories into panels reduces congestion and occlusion that otherwise distort group-level judgments in a single shared scatterplot.

Evidence: For summary tasks, colored scatterplots degraded as cardinality and congestion increased, while faceted charts held accuracy better and regained a better rank under high-cardinality conditions despite slower completion times (Kim & Heer, 2018).

context

Use when summary accuracy matters more than speed

  • User Goal: Compare categories, identify the category containing the maximum, or compare category averages.
  • Task: Group-level summary judgment.
  • Data: Many categories, many records per category, or clustered points that produce visible overlap.
  • Chart Setting: A choice between a single-view category-colored scatterplot and a faceted point plot.
  • Success Criterion: Fewer summary-task errors under congestion.

exceptions

Do not use when speed or space dominates the requirement

Break it when: Fast completion or limited vertical space is more important than summary accuracy. Why: Faceted charts stayed accurate but took longer and often required more scrolling as the number of categories increased.

costs

What this costs

Sacrifice: You give up some speed and screen economy. Risk: Readers must compare across multiple panels instead of one shared panel. Mitigation: Make this swap when overlap in the single-view plot is already causing group-comparison errors.

mistakes

Common failure around this move

Mistake: Keep adding categories and points to one color-coded scatterplot for a summary task. Why it fails: Overplotting and occlusion increase errors on group-level judgments.

check

How to test the choice

Failure Sign: Category-colored points overlap heavily and readers cannot isolate groups reliably. Quick Check: A/B compare the current single-view colored scatterplot against a faceted version on one maximum-finding or average-comparison question. Stronger Test: If the faceted version reduces summary-task error enough to justify a slower read, prefer the faceted version.

fix

What to change

  • Replace category color with category faceting.
  • Keep shared quantitative axes across panels so groups remain comparable.
  • Use the faceted version when category count or point overlap is high enough to hurt summary accuracy.

References

Kim, Y., & Heer, J. (2018). Assessing Effects of Task and Data Distribution on the Effectiveness of Visual Encodings. Computer Graphics Forum, 37(3), 157–167. https://doi.org/10.1111/cgf.13409