Catalog

Guideline Catalog

Browse visualization guideline records with sections, labels, and references.

781 records

Page 8 of 33

  1. Directly encode pairwise differences when relations are the task

    For relation judgments over paired quantitative values, use direct delta encoding on comparison charts to improve fidelity and mitigate slow, error-prone extraction of increases and decreases for viewers scanning or summarizing many pairs.

    • purpose:refine
    • basis:empirical
    • task:relate
    • data:quantitative
    • quality:fidelity:use
    • +3

  2. Directly label categories when colors are similar or repeated

    For dense multi-category charts, use direct labels on same-colored or similarly colored marks to improve readability and mitigate reliance on a separate color key for readers scanning the chart.

    • purpose:refine
    • basis:heuristic
    • lever:text-annotation
    • component:label:use
    • component:legend:avoid
    • +3

  3. Directly label colored series instead of relying on a color key

    For color-coded series identification, prefer direct labels on chart marks to improve readability and mitigate color-key decoding for readers with color-vision deficiency.

    • purpose:refine
    • basis:heuristic
    • quality:accessibility
    • lever:text-annotation
    • component:label:use
    • +3

  4. Disclose the size of the hypothesis search on exploratory result displays

    For interpreting exploratory findings from many tested relationships, use text annotation on result displays to improve trust and mitigate false certainty from hidden multiplicity for audiences judging hypothesis-generating claims.

    • purpose:refine
    • basis:empirical
    • quality:trust:use
    • lever:text-annotation
    • communication:credibility
    • +1

  5. Do not add redundant class encodings just to improve mean-comparison accuracy

    For grouped comparison tasks with multiclass scatterplots, avoid adding redundant class encodings on point marks to prevent unnecessary encoding changes and mitigate unsupported attempts to improve fidelity for viewers comparing which group is on average higher.

    • purpose:refine
    • basis:empirical
    • task:compare
    • chart:scatter
    • lever:encoding
    • +1

  6. Do not make 3D navigation necessary when the view must stand on its own

    For presentation and review contexts with static frames or shared control, avoid navigation-dependent 3D views to prevent comprehension failure and address interaction bottlenecks for viewers who cannot steer the scene.

    • purpose:refine
    • basis:empirical
    • quality:readability
    • lever:interaction-access
    • communication:workflow

  7. Do not present relative risk reduction as the only treatment-effect summary

    For explaining a single intervention effect, avoid relative-only effect wording on quantitative labels and captions to improve trust and mitigate inflated perceived benefit for health professionals and consumers.

    • purpose:refine
    • basis:empirical
    • scope:single-result
    • data:quantitative
    • quality:trust:use
    • +3

  8. Do not rely on axis-break indicators to neutralize truncation

    For overview judgments of change, avoid relying on axis-break or continuation cues on truncated bar charts to improve fidelity and address the mistaken belief that visual warnings remove exaggeration for viewers reading severity from shape.

    • purpose:refine
    • basis:empirical
    • chart:bar
    • data:quantitative
    • quality:fidelity:use
    • +3

  9. Do not rely on boundary softness alone to correct bounded uncertainty displays

    For intuitive impact judgments across ordered future time, avoid using boundary softness alone on bounded geospatial forecast maps to prevent false confidence in improved comprehension and address persistent size-and-distance misreadings for viewers with low domain knowledge.

    • purpose:refine
    • basis:empirical
    • time:ordered-time
    • data:geospatial
    • quality:fidelity
    • +3

  10. Do not rely on caption text alone to correct a confidence-interval chart

    For treatment-effect judgments from confidence-interval summaries, avoid caption-only annotation on mean-with-interval charts to prevent persistent overestimation of treatment effectiveness and address misreading of inferential uncertainty for lay readers.

    • purpose:refine
    • basis:empirical
    • task:compare
    • quality:fidelity:use
    • lever:text-annotation
    • +3

  11. Do not treat truncation cues as a fix for a non-zero y-axis

    For truncated quantitative bar charts where readers should judge effect size without added emphasis, avoid using broken-axis or gradient continuation cues as the fix for a non-zero y-axis to prevent exaggerated perceived effect size and address false confidence that disclosure alone removes the distortion for readers comparing small differences.

    • purpose:refine
    • basis:empirical
    • chart:bar
    • lever:scale-order
    • component:axis
    • +1

  12. Do not tune perceptually uniform multi-hue palettes to chase retrieval performance

    For retrieve tasks, avoid switching among perceptually uniform multi-hue sequential color scales on quantitative color encodings to prevent over-interpreting negligible performance differences and address unnecessary palette tuning for viewers comparing values against a legend.

    • purpose:refine
    • basis:empirical
    • task:retrieve
    • data:quantitative
    • quality:fidelity
    • +3

  13. Do not tune single-hue sequential palettes to chase retrieval performance

    For retrieve tasks, avoid switching among single-hue sequential color scales on quantitative color encodings to prevent over-interpreting negligible performance differences and address unnecessary palette tuning for viewers comparing values against a legend.

    • purpose:refine
    • basis:empirical
    • task:retrieve
    • data:quantitative
    • quality:fidelity
    • +3

  14. Double-encode any ordered variable shown with shades

    For quick-reading charts that use shades for an underlying order or second measure, prefer color encoding on values already visible in position, order, size, or an obvious companion key to improve readability and mitigate hidden second-variable encodings for readers interpreting color.

    • purpose:refine
    • basis:heuristic
    • quality:readability
    • lever:encoding
    • channel:color-lightness:use
    • +1

  15. Draw outliers separately from the density context

    For outlier inspection in uncertain multivariate plots, use a separate outlier layer on density plots to improve insight and address the loss of unusual values in PDF summaries for viewers scanning focus and context.

    • purpose:refine
    • basis:empirical
    • lever:encoding
    • operator:uncertainty
    • shape:outlier-rich
    • +1

  16. Embed essential context directly in the chart

    For distribution outside the original context, use text annotation on charts to prevent misinterpretation and address missing surrounding explanation for readers encountering shared graphics on their own.

    • purpose:refine
    • basis:rhetorical
    • task:distribute
    • quality:fidelity
    • lever:text-annotation
    • +2

  17. Embed essential context directly in the visualization

    For message interpretation when a visualization may be read without enough surrounding explanation, use direct context annotation on the chart to improve trust and mitigate guessing or assumption-based interpretation for viewers who would otherwise have to infer the topic.

    • purpose:refine
    • basis:rhetorical
    • quality:trust
    • lever:text-annotation
    • communication:context
    • +1

  18. Emphasize distribution means in proportion to certainty

    For small multivariate uncertainty views, use certainty-scaled mean emphasis on density plots to improve readability and address loss of individual sample locations for viewers inspecting manageable numbers of samples.

    • purpose:refine
    • basis:empirical
    • lever:encoding
    • operator:uncertainty
    • density:sparse
    • +1

  19. Emphasize key marks to guide attention

    For static explanatory viewing, use visual emphasis on key marks in charts where multiple elements compete for attention to improve readability and mitigate the loss of key insights for your audience.

    • purpose:refine
    • basis:rhetorical
    • quality:readability
    • lever:encoding
    • polish:focus

  20. Encode a single risk proportion as a countable unit array

    For exact reading of a single risk at one time point, use repeated unit marks on a risk proportion display to improve readability and mitigate confusion about affected versus unaffected cases for patient decision-makers with low numeracy.

    • purpose:refine
    • basis:empirical
    • time:timepoint
    • quality:readability
    • lever:encoding
    • +3

  21. Encode compared groups with color instead of shape

    For grouped comparison tasks with multiclass scatterplots, prefer color encoding on class membership to improve fidelity and mitigate mean-position judgment errors for viewers comparing which group is on average higher.

    • purpose:refine
    • basis:empirical
    • task:compare
    • chart:scatter
    • lever:encoding
    • +3

  22. Encode elapsed time with segment-length ticks on simple 2D trajectories

    For exact time lookup on ordered-time trajectories, prefer segment-length ticks on simple 2D path views to improve fidelity and mitigate misreading of elapsed time from color shading for readers locating time at a point on the path.

    • purpose:refine
    • basis:empirical
    • time:ordered-time
    • quality:fidelity:use
    • lever:encoding
    • +3

  23. Encode pairwise differences as visual objects when comparison is the task

    For comparison across many paired values, use direct difference encoding on grouped charts to improve insight and mitigate slow one-by-one comparison for readers scanning multiple pairs.

    • purpose:refine
    • basis:empirical
    • task:compare
    • quality:insight:use
    • lever:encoding
    • +3

  24. Encode quantitative values with position when reading accuracy matters

    For exact quantitative lookup or comparison, use position encoding instead of area encoding on quantitative displays to improve fidelity and mitigate magnitude misjudgment for readers making precise value readings.

    • purpose:refine
    • basis:empirical
    • task:retrieve
    • quality:fidelity:use
    • lever:encoding
    • +3