Map values to a shared spatial scale.
Aligned position usually supports precise quantitative comparison.
Side 121
Data Visualization
Data transformed into visual structure so patterns, comparisons and uncertainty can be perceived without letting graphic choices quietly rewrite the evidence.
data→encoding→perception→comparison→judgment
04lenses
16working concepts
V0content
SS-1.0standard
Every chart maps variables onto visual channels.
Position, length, angle, area, color and shape differ in how precisely people can compare them.
Use extent to represent magnitude.
Bar length is easy to compare when the baseline and scale are explicit.
Magnitude judgments become less precise.
Circles and bubbles can emphasize size while making exact comparison harder.
Use hue and intensity for category or ordered value.
Color is powerful but should not carry distinctions that disappear for common forms of color-vision difference.
Chart choice should follow the comparison task.
The same dataset can support different questions, and no one chart is universally best.
Show shape, spread and unusual values.
Histograms, density displays and box-type summaries answer different distribution questions.
Show how variables vary together.
Scatterplots reveal association while still requiring caution about causation.
Show part-to-whole structure.
Stacking is useful for totals but can make internal segment comparison difficult.
Show movement through time or sequence.
Temporal charts should distinguish trend from irregular intervals and missing observations.
Uncertainty should be visible when it matters to interpretation.
Point estimates alone often create more certainty than the analysis warrants.
Show plausible ranges around estimates.
Intervals need clear interpretation because confidence, credible and prediction intervals answer different questions.
Show many possible trajectories.
Small multiples or simulated paths can reveal distributional structure hidden by one band.
Translate probabilities into repeated cases.
Icon arrays can make risk magnitude easier to understand for some audiences.
Show how conclusions change under assumptions.
Visualizing alternative specifications can reveal robustness more directly than a single preferred result.
Graphical choices can distort even accurate numbers.
Axes, aggregation and visual emphasis affect the story the viewer perceives.
Axis range changes apparent magnitude.
Truncation can be appropriate but must not hide the baseline needed for the comparison being made.
Grouping can create or erase patterns.
Time bins and category definitions are analytical decisions, not cosmetic choices.
Direct attention to the intended inference.
Annotations should clarify evidence rather than instruct the viewer to see unsupported meaning.
Filtering changes the visible evidence set.
Interactive views need stable context so exploration does not become silent cherry-picking.
A visualization is an argument made with position, scale and emphasis. Good design increases perceptual access to the data while minimizing opportunities for the graphic to imply relationships the evidence does not contain.