Split opposing dual-axis time series into separate views

Split measures into separate aligned views when one chart would otherwise require opposing y-axes. For example, replace a single dual-axis time-series chart with two side-by-side panels that share time, and encode a negative quantity with bars connected to the zero baseline instead of reading it from a reversed second axis.

Separate views reduce scale conflict. Readers can decode each measure against its own axis instead of first deciding which of two incompatible y-axes applies.

Mechanism: One axis per measure simplifies exact lookup and makes the direction of each quantity easier to read.

Evidence: Across the study, redesigned charts improved knowledge-level value retrieval; in the markets case, the redesign replaced one chart with opposing positive and negative y-axes by two side-by-side charts and a zero-baseline bar for the negative series, while higher-level summaries and arguments for that chart stayed largely similar across versions (Burns et al., 2020).

Notes: The supported gain in this case is mainly exact value retrieval, not a broad change in all higher-level interpretations.

• User Goal: Locate and report exact values from two measures over time.
• Task: Exact retrieval from each series.
• Data: Two quantitative measures that would otherwise need separate scales, including cases where one measure is negative-valued.
• Chart Setting: A static time-series display with a shared time range.
• Audience: General readers.
• Success Criterion: Readers retrieve values correctly without decoding conflicting axes.

Break it when: The chart’s main success criterion is the overall takeaway, trend summary, or argument rather than exact value lookup. Why: In the markets case, the split-view redesign did not materially change the kinds of summaries, trend descriptions, or policy arguments readers produced.

Sacrifice: You give up some compactness from a single combined plot.
Risk: Readers must compare across panels instead of within one plotting area.
Mitigation: Keep the panels side by side and aligned on the same time range.

Mistake: Keep both measures in one plot with opposing y-axes when readers need exact values. Why it fails: Readers must first map each mark to a different scale, which increases readout difficulty.

Failure Sign: Readers miss simple point-lookup questions on one or both measures.
Quick Check: Compare a dual-axis draft and a split-view draft on one lookup question per measure; keep the version with fewer readout errors.
Stronger Test: Run several value-retrieval questions on both versions and compare accuracy.

• Move each measure into its own panel.
• Align the panels on the same time span.
• Replace a reversed axis for a negative quantity with bars connected to the zero baseline.
• Keep the combined single view only if exact lookup is not the main success criterion.