Guidelines
Suggest edit

Use a dual-axis line chart when correlation is the main message

For communicating the relationship in ordered-time paired series, prefer a dual-axis line chart over a connected scatterplot to improve insight and address missed positive or negative correlation cues for readers unfamiliar with connected scatterplots.

  • purpose:select
  • basis:empirical
  • task:relate
  • time:ordered-time
  • chart:line:use
  • chart:scatter:avoid
  • quality:insight
  • lever:chart-family
  • operator:association
  • literacy:novice

advice

Choose the chart family for correlation cues

Choose a dual-axis line chart when readers need to come away with a clear sense of positive or negative correlation. For example, use the line-chart form when parallel lines or an X-shape are the main takeaway, and avoid a connected scatterplot if the same data would make readers infer correlation from a single diagonal path.

reason

Why the dual-axis line chart works better for correlation

Readers already have learned visual associations between familiar line-chart patterns and correlation. Those learned cues do not transfer reliably to the diagonal segments of a connected scatterplot.

Mechanism: Parallel or crossing line patterns in a dual-axis line chart directly cue positive or negative relationships, while connected-scatterplot shapes draw attention to turns, loops, and path geometry instead.

Evidence: In the qualitative study, participants used correlational language far more often for dual-axis line charts than for connected scatterplots, and the paper concludes that a traditional dual-axis line chart may make positive or negative correlation much more salient when that is the intended takeaway (Haroz et al., 2016).

context

Use when correlation is the takeaway

  • User Goal: Explain whether two measures move together or in opposite directions.
  • Task: Help readers identify a positive or negative relationship.
  • Data: Two simultaneous time series with matched sampling.
  • Chart Setting: A static presentation graphic for paired time series.
  • Audience: Readers who have much more experience with line charts than with connected scatterplots.
  • Success Criterion: Readers explicitly describe a direct or inverse relationship.

exceptions

Do not use when another message matters more

Break it when: The main goal is to attract initial attention or to highlight distinctive loops, crossings, or L-shapes in the paired series. Why: The paper found that readers prioritized connected scatterplots for viewing and repeatedly noticed their unusual shapes.

costs

Costs of switching to a dual-axis line chart

Sacrifice: You give up the distinctive loop and L-shape patterns that connected scatterplots make visually salient.
Risk: Intersections in a dual-axis line chart are only meaningful when the units and scales of the two vertical axes are the same.
Mitigation: Do not rely on line intersections as evidence of a relationship unless the two vertical scales are directly comparable.

mistakes

Common chart-choice failure

Mistake: Use a connected scatterplot for a correlation takeaway without teaching readers how its diagonal segments encode association. Why it fails: Readers tend to describe the shape and motion of the path instead of stating a direct or inverse relationship.

check

Check which chart makes correlation more explicit

Failure Sign: Readers talk about loops, turns, or erratic motion but do not name a positive or negative relationship.
Quick Check: Show naive readers both the connected scatterplot and the dual-axis line chart and ask for the main relationship message.
Stronger Test: Keep the version that yields more explicit direct or inverse relationship language.

fix

Fix a missed-correlation presentation

  • Replace the connected scatterplot with a dual-axis line chart when the main message is correlation.
  • Keep the same paired series and time span so the chart-family change is the only difference.
  • If you keep the connected scatterplot, annotate the positive or negative relationship directly.

References

Haroz, S., Kosara, R., & Franconeri, S. L. (2016). The Connected Scatterplot for Presenting Paired Time Series. IEEE Transactions on Visualization and Computer Graphics, 22(9), 2174–2186. https://doi.org/10.1109/TVCG.2015.2502587