Guidelines
Suggest edit

Avoid slope encodings for pairwise relation judgments

For compare tasks on grouped paired quantitative displays, avoid slope encoding on paired-value charts to improve fidelity and mitigate slow and inaccurate relation judgments for viewers.

  • purpose:refine
  • basis:empirical
  • task:compare
  • scope:grouped-result
  • data:quantitative
  • quality:fidelity:use
  • lever:encoding
  • channel:orientation:avoid

advice

Replace slope as the relation carrier

Avoid slope when the display must support relation-direction judgments across many paired values. For example, remap the same paired relation from slope to position-based marks or bar-length marks when readers must search for one opposite relation or decide which relation direction is more prevalent.

reason

Why avoiding slope helps these judgments

In these tasks, slope made the relation harder to judge than the other tested channels. Readers were slower or less accurate when the paired relation was carried by slope.

Mechanism: When the same paired relation is mapped to position or length instead of slope, the relation is easier to distinguish across many pairs.

Evidence: In the collated record, slope-based variants ranked below position-based and length-based variants for target-relation search time and for prevalence judgments; the original study reports significantly worse performance for slope encodings than for position and length on these relation-direction tasks (Zeng & Battle, 2023; Nothelfer & Franconeri, 2020).

Notes: The source limits this finding to the studied displays and tasks.

context

Use when slope is carrying the pairwise comparison

  • User Goal: Judge which paired relation is present or dominant.
  • Task: Search for a target relation or decide which relation direction is more prevalent.
  • Data: Paired quantitative values shown repeatedly across many categories or items.
  • Chart Setting: The display currently uses slope to carry the pairwise relation and can be redrawn with the same relation mapped to position or length.
  • Success Criterion: Faster relation search or higher accuracy for direction judgments.

exceptions

Do not generalize beyond the studied task setting

Break it when: The display is not being used for the studied relation-direction judgments across many paired quantitative marks. Why: This evidence only establishes the slope penalty for the tested search and prevalence tasks in this display setting.

costs

Costs of avoiding slope in these views

Sacrifice: You give up using slope as the main carrier of the pairwise relation. Risk: Keeping slope can slow target-relation search and reduce accuracy for majority-direction judgments. Mitigation: Remap the same signed relation to position or length while keeping the pairing structure unchanged.

mistakes

Common failure mode with slope encodings

Mistake: Keeping paired slope marks for displays whose main question is which relation direction appears or dominates. Why it fails: The studied tasks were slower or less accurate with slope than with position or length.

check

Check whether slope is hurting relation judgments

Failure Sign: Relation-direction questions are slow or error-prone in a slope-based paired display. Quick Check: Redraw the same display with the relation mapped to position or length and compare one target-relation judgment and one majority-direction judgment. Stronger Test: Time both versions or compare accuracy; keep the non-slope version if it performs better.

fix

Fix the encoding channel

  • Remap the paired relation from slope to position.
  • Or remap the same paired relation to bar length.
  • Keep the baseline and category positions constant so the encoding channel is the main change under review.
  • Reserve slope only for cases outside these studied pairwise relation-judgment tasks.

References

Nothelfer, C., & Franconeri, S. (2020). Measures of the Benefit of Direct Encoding of Data Deltas for Data Pair Relation Perception. IEEE Transactions on Visualization and Computer Graphics, 26(1), 311–320. https://doi.org/10.1109/TVCG.2019.2934801
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