Catalog

Guideline Catalog

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

781 records

Page 27 of 33

  1. Use absolute risk reduction instead of number needed to treat when understanding is the goal

    For explaining a single intervention effect, prefer absolute-risk wording on quantitative effect labels and captions to improve fidelity and mitigate misinterpretation of treatment benefit for readers interpreting treatment effects.

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

  2. Use adjacent rather than stacked bar pairs for correlation comparison

    For comparison of similarity between two quantitative series in paired bar charts when using conventional small multiples, prefer adjacent placement over stacked placement to improve fidelity and mitigate cross-panel correspondence errors for viewers judging which pair is more correlated.

    • purpose:refine
    • basis:empirical
    • task:compare
    • chart:bar
    • structure:small-multiples
    • +3

  3. Use aligned bar charts instead of labeled scatter plots for item-level lookup

    For item-level lookup in multi-measure record lists, prefer aligned bar charts over scatter plots to improve lookup fidelity and mitigate label-search problems for readers of static graphics.

    • purpose:select
    • basis:empirical
    • task:retrieve
    • scope:record-list
    • chart:bar:use
    • +4

  4. Use an aligned bar chart instead of a pie chart for proportion comparison

    For compare tasks, use bar charts on a common scale rather than pie charts to improve fidelity and address errors from judging slice angles for readers making exact value judgments.

    • purpose:select
    • basis:empirical
    • task:compare
    • chart:bar:use
    • chart:pie-donut:avoid
    • +4

  5. Use an area chart when time intervals are uneven

    For ordered-time part-to-whole reading with irregular date spacing, use an area chart instead of stacked columns to improve temporal fidelity and mitigate incorrect equal-spacing cues for readers interpreting time intervals.

    • purpose:select
    • basis:heuristic
    • time:ordered-time
    • chart:area:use
    • chart:bar:avoid
    • +3

  6. Use an area chart when time intervals between dates are irregular

    For showing shares over ordered time, use an area chart instead of a stacked column chart on irregularly spaced dates to improve fidelity and mitigate incorrect spacing of time points for readers interpreting elapsed time.

    • purpose:select
    • basis:heuristic
    • time:ordered-time
    • chart:area:use
    • chart:bar:avoid
    • +3

  7. Use an integrated visualization instead of a crowded dashboard for multifaceted analysis

    For complex analytical tasks on multifaceted data, prefer an integrated single-space structure over a dashboard of separate simple views to improve insight and mitigate mental recombination across panels for analysts exploring multiple facets simultaneously.

    • purpose:select
    • basis:empirical
    • structure:single-view:use
    • structure:dashboard:avoid
    • density:dense
    • +3

  8. Use an ordered color gradient for continuous data

    For comparing low-to-high values, use an ordered color gradient on quantitative color encodings to improve fidelity and mitigate false category cues for readers interpreting continuous data.

    • purpose:refine
    • basis:heuristic
    • data:quantitative
    • quality:fidelity
    • lever:encoding
    • +1

  9. Use an ordered lightness scale for values with inherent order

    For comparison of ordered values, use a quantitative color scale on chart encodings to improve readability and address unordered hue assignments for readers scanning low-to-high meaning.

    • purpose:refine
    • basis:heuristic
    • data:quantitative
    • data:ordinal
    • quality:readability
    • +2

  10. Use an overlaid function graph to verify interval solutions

    For compare tasks on non-temporal quantitative functions, use a line chart instead of text-only checking on paired function displays to improve fidelity and mitigate missed extra-intersection mistakes for learners verifying inequality intervals.

    • purpose:select
    • basis:empirical
    • task:compare
    • chart:line:use
    • chart:text:avoid
    • +4

  11. Use an unclassed color scale to show continuous patterns

    For overview reading of continuous values, use an unclassed color scale on quantitative choropleth maps to improve fidelity and mitigate oversimplified regional differences for readers scanning general spatial patterns.

    • purpose:refine
    • basis:heuristic
    • chart:choropleth
    • data:quantitative
    • quality:fidelity
    • +2

  12. Use animated probabilistic sampling when full density computation is too expensive

    For overview of very large uncertain multivariate data, use animated probabilistic sampling on scatter and parallel plots to improve insight and address hidden outliers for viewers scanning dense distributions.

    • purpose:refine
    • basis:empirical
    • lever:encoding
    • operator:uncertainty
    • density:dense
    • +2

  13. Use animated transitions to show the largest change in bar charts

    For comparison of two quantitative series in a biggest-mover task, use an animated single-view transition on bar charts to improve fidelity and mitigate missed largest-change judgments for viewers working from brief impressions.

    • purpose:select
    • basis:empirical
    • task:compare
    • chart:bar
    • structure:single-view:use
    • +3

  14. Use animation instead of small multiples for fast passive trend presentations

    For passive presentation of trends over ordered time, prefer a single animated view on multidimensional scatter plots to improve insight speed and mitigate slow panel scanning for viewers.

    • purpose:select
    • basis:empirical
    • task:trend
    • time:ordered-time
    • structure:single-view:use
    • +4

  15. Use annotations to explain and emphasize chart elements

    For explanatory charts and maps, use annotations on chart elements and notable data points to improve insight and address missed context for readers following the intended message.

    • purpose:refine
    • basis:heuristic
    • quality:insight
    • lever:text-annotation
    • component:annotation:use
    • +2

  16. Use area instead of color-saturation when the primary quantitative field cannot use position

    For multivariate point-based views where the primary quantitative field cannot use position, prefer area on the primary field to improve fidelity and mitigate higher decoding error from color-saturation for readers interpreting the main measure.

    • purpose:refine
    • basis:empirical
    • quality:fidelity:use
    • lever:encoding
    • measure:multi
    • +2

  17. Use area over hue for extreme-value judgments

    For extreme-value judgments, use area instead of hue on quantitative marks in ordered one-dimensional sequences to improve fidelity and mitigate missed smallest-or-largest values for viewers scanning static displays.

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

  18. Use bar charts instead of line charts for cluster detection

    For cluster detection in small static two-dimensional displays, use a bar chart instead of a line chart on tabular data to improve fidelity and speed and mitigate missed or ambiguous group structure for readers scanning for similar values.

    • purpose:select
    • basis:empirical
    • chart:bar:use
    • chart:line:avoid
    • data:tabular
    • +3

  19. Use bar charts to emphasize discrete value comparison

    For discrete comparison between separate quantitative observations, prefer a bar chart on simple comparison displays to improve readability and mitigate unintended trend interpretations for readers.

    • purpose:select
    • basis:empirical
    • task:compare
    • chart:bar:use
    • chart:line:avoid
    • +2

  20. Use bar charts to support main-effect inferences from familiar multivariate data

    For open-ended interpretation of grouped three-variable data, use a bar chart instead of a line chart on multivariate quantitative graphs to improve insight into main effects and mitigate overemphasis on interactions for viewers with higher graph skills reading familiar content.

    • purpose:select
    • basis:empirical
    • task:compare
    • chart:bar:use
    • chart:line:avoid
    • +3

  21. Use bar encoding instead of line encoding when aggregation is fixed

    For communicating non-temporal associations with the same aggregation level, prefer bar encoding over line encoding on grouped or paired-value displays to improve fidelity and mitigate stronger causal readings for viewers interpreting observational data.

    • purpose:select
    • basis:empirical
    • task:relate
    • time:non-temporal
    • chart:bar:use
    • +3

  22. Use bars instead of pie or donut slices for percentage comparisons

    For comparing shares across categories, prefer bar charts over pie or donut charts to improve readability and mitigate hard-to-see percentage differences for general audiences.

    • purpose:select
    • basis:heuristic
    • chart:bar:use
    • chart:pie-donut:avoid
    • operator:part-whole
    • +3

  23. Use box plots instead of line graphs for interval spread comparisons

    For spread comparisons across ordered-time intervals, use box plots instead of line graphs to maximize fidelity and address inferred variation from raw traces for exact interval judgments.

    • purpose:select
    • basis:empirical
    • task:distribute
    • time:ordered-time
    • chart:box-violin:use
    • +3

  24. Use category context to disambiguate semantic color assignment

    For category lookup with ambiguous or symbolic labels, use contextual semantic cues on categorical color encoding to improve fidelity and mitigate wrong-sense color assignments for viewers interpreting multi-meaning terms.

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