Use separate displays instead of co-plotting two series for average-position judgments

Show each data series in its own display instead of co-plotting two series when readers must judge each series’ average vertical position after brief viewing. For example, separate two lines, two bar series, or a line-plus-bar pair into different displays rather than placing both series in one shared frame when viewers must reproduce one target series’ average height after a short delay.

A second series in the same display shifts the remembered average of the target series toward the non-target series. Splitting the series removes that specific cross-series pull, which otherwise exaggerates or reduces the baseline bias depending on whether the non-target series sits above or below the target.

Mechanism: Co-plotted series act like anchors on each other, so the irrelevant series changes where viewers remember the target series’ average position to be.

Evidence: In compound displays with two lines, two bar series, or a line-plus-bar pair, target average-position estimates were pulled toward the irrelevant series, and the authors advise avoiding plotting two series in the same display when trying to avoid this effect (Xiong et al., 2020).

Notes: Changing one of the series from a line to bars does not remove the pull; the effect generalized across same-type and mixed-type pairs.

• User Goal: Read each of two series separately and reproduce each series’ average vertical position accurately.
• Task: View a target series briefly, then report its average level from memory.
• Data: Two quantitative series.
• Chart Setting: A design choice between one shared display and separate displays for the two series.
• Audience: Readers making memory-based judgments of one cued target series at a time.
• Success Criterion: Smaller cross-series distortion in reproduced average positions.

Break it when: The display contains only one series, or the task is not a delayed average-position report for one target series. Why: The paper establishes perceptual pull only for two-series displays in the tested average-position task.

Sacrifice: Readers no longer see both series in one shared frame. Risk: Splitting the series does not remove the baseline bias of the chosen chart family, because single lines were still underestimated and single bars were still overestimated. Mitigation: After separating the series, test each display for the remaining line or bar bias.

Mistake: Keep both series in one display but change one series to a different mark type and expect the interference to disappear. Why it fails: Line-plus-bar displays still showed perceptual pull, not a clean escape from it.

Failure Sign: Estimates for a target series shift toward the other series when both appear in the same display. Quick Check: Compare a shared two-series display against a version where each series is shown in its own display, using brief viewing, a short delay, and average-position reproduction. Stronger Test: Run the comparison for each pairing you might use, including two lines, two bar series, and a line-plus-bar pair.

• Remove the non-target series from the display used for the average-position readout.
• Put each of the two series in its own display instead of co-plotting both in one frame.
• Do not use a line-plus-bar combination as a workaround for a two-line or two-bar shared display.