Guidelines
Suggest edit

Use bar charts instead of line charts for cluster detection

For cluster detection in small static two-dimensional displays, use a bar chart instead of a line chart on tabular data to improve fidelity and speed and mitigate missed or ambiguous group structure for readers scanning for similar values.

  • purpose:select
  • basis:empirical
  • chart:bar:use
  • chart:line:avoid
  • data:tabular
  • quality:fidelity:use
  • lever:chart-family
  • density:sparse

advice

Choose bar over line

Use a bar chart when the main question is how many groups of similar values appear in the display. For example, replace a line chart with a bar chart when readers must count clusters rather than follow a trend.

reason

Why bar works better here

A bar chart makes separate groups easier to scan as distinct magnitudes, while a line chart emphasizes connection across points. That helps readers detect and count groups more accurately and with less time.

Mechanism: Bar marks support scanning for separated groups of similar values, while connecting those values with a line can pull attention toward continuity instead of grouping.

Evidence: In the experiment, bar charts ranked above line charts for cluster tasks in accuracy, time, and user preference, and the later collation summarizes this study as recommending bar charts for cluster tasks. (Saket et al., 2019; Zeng & Battle, 2023)

Notes: The tested displays were static and used 5-34 marks.

context

Use when the task is group finding

  • User Goal: Count or detect groups of similar values.
  • Data: Tabular data shown as a small static display.
  • Chart Setting: Two-dimensional chart with 5-34 marks.
  • Success Criterion: Better accuracy and faster cluster finding, with stronger user preference.

exceptions

Do not use when the task changes to correlation

Break it when: the task changes from finding groups to judging whether two attributes are correlated. Why: line charts were among the strongest options for correlation, while bar charts were not the top choice there.

costs

Costs of switching from line to bar

Sacrifice: You give up the connected trend emphasis of the line chart. Risk: Using a bar chart by default can hide that the reader’s real question is about relationship rather than grouping. Mitigation: Switch only when the primary question is cluster detection.

mistakes

Common failure mode

Mistake: Use a line chart for cluster detection because the categories are ordered. Why it fails: the connecting line can foreground continuity even when the task is to count separate groups.

check

Check the task against the chart choice

Failure Sign: Reviewers must count groups or spot clusters from a line chart. Quick Check: Render the same values as both a bar chart and a line chart and see which version makes group counting easier on a sample question. Stronger Test: Time one cluster question on both versions and compare both answer errors and response time.

fix

Fix the chart choice

  • Replace the line chart with a bar chart when the question is how many groups are present.
  • Keep the same summarized values and redraw them as separate bars instead of connected points.
  • If the line chart is needed for another task, present a separate bar chart for the clustering question.

References

Saket, B., Endert, A., & Demiralp, Ç. (2019). Task-Based Effectiveness of Basic Visualizations. IEEE Transactions on Visualization and Computer Graphics, 25(7), 2505–2512. https://doi.org/10.1109/TVCG.2018.2829750
Zeng, Z., & Battle, L. (2023). A Review and Collation of Graphical Perception Knowledge for Visualization Recommendation. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1–16. https://doi.org/10.1145/3544548.3581349