Guidelines
Suggest edit

Use box plots instead of line graphs for interval spread comparisons

For spread comparisons across ordered-time intervals, use box plots instead of line graphs to maximize fidelity and address inferred variation from raw traces for exact interval judgments.

  • purpose:select
  • basis:empirical
  • task:distribute
  • time:ordered-time
  • chart:box-violin:use
  • chart:line:avoid
  • quality:fidelity
  • lever:chart-family

advice

Choose box plots for spread

Choose a box plot when readers must compare how spread out values are within each time interval. For example, use one box plot per month instead of a line graph when the question is which interval is most spread out from its average.

reason

Why box plots beat lines here

Spread comparison depends on seeing interval-level variation, not tracing a raw sequence of points. Box plots expose summary distribution structure directly, while line graphs force readers to infer spread from many individual values.

Mechanism: The box, whiskers, and central summary reduce the interval to a directly comparable distribution cue, so readers do less visual estimation of variability.

Evidence: In the time-series experiment, the box plot was the top-performing design for the spread task and outperformed the plain line graph, and the review summarizes the same paper as showing that different chart types support different summary tasks differently. (Albers et al., 2014; Zeng & Battle, 2023)

context

Use when spread is the real question

  • User Goal: Compare dispersion across repeated time intervals.
  • Task: Interval spread comparison.
  • Data: Ordered time series partitioned into repeated intervals such as months.
  • Chart Setting: Readers do not need to inspect every raw point while making the spread decision.
  • Success Criterion: Higher answer accuracy on interval spread questions.

exceptions

Do not use when raw points or anomaly counts must stay visible

Break it when: Readers must inspect raw point values or count outliers within each interval. Why: the box plot replaces the raw series, and the study did not use it for the outlier-count task.

costs

What you give up

Sacrifice: Direct visibility of every raw value in the series. Risk: Tasks that depend on local features or anomaly counts can become harder or unsupported. Mitigation: Use a different view when the task depends on raw points instead of spread summaries.

mistakes

Common selection mistake

Mistake: Keep a line graph for a spread comparison because it preserves every point. Why it fails: Readers must infer dispersion from many points, and accuracy is lower than with a box plot.

check

How to check the chart choice

Failure Sign: Reviewers trace the whole line to estimate which interval varies most. Quick Check: Show the same spread question once with a line graph and once with interval box plots. Stronger Test: Keep the box plot if it yields more accurate answers to a forced-choice question about which interval is most spread out.

fix

How to fix it

  • Replace the line graph with one box plot per target interval.
  • Keep the boxes in the same time order as the original series.
  • If the real question is range rather than spread, switch to explicit interval minima and maxima instead of a line graph.

References

Albers, D., Correll, M., & Gleicher, M. (2014). Task-driven evaluation of aggregation in time series visualization. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 551–560. https://doi.org/10.1145/2556288.2557200
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