Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Visualization

Abstract

Outlines the foundational principles for creating compelling and accurate data visualizations. These principles are grounded in cartographic theory, information design, and empirical research in visual perception.

Purpose & Audience

Define your message before you design. Every effective map or chart begins with a clear understanding of:

  • What is the story? - What insight or finding do you want to communicate?

  • Who is the audience? - What is their background and knowledge level?

  • What action or decision will this visualization support?

According to cartographic design theory, a map or chart that cannot stand alone or fails to communicate a specific message to its intended audience has failed in its primary purpose—regardless of its aesthetic beauty or technical sophistication Bertin, 1967Tufte & Graves-Morris, 1983.

Data Integrity

Represent data truthfully and avoid distortion. It is easy to lie with maps, although all maps distort the truth somewhat, you just need to decide which truth you’re displaying. Distorted visualizations mislead audiences and undermine decision-making. This is not just an ethical issue—it destroys trust and credibility.

The most important rule in visualization is that your visualization must not lie. This means:

  • Accurate scaling - Visual dimensions must be proportional to data values

  • Proper context - Show data in appropriate context; don’t cherry-pick ranges

  • No misleading effects - Avoid visual tricks that exaggerate or minimize differences

  • Document your methods - Clearly explain data sources, processing, and any transformations

Common distortions to avoid include

  • Truncated axes that exaggerate small differences

  • 3D effects that distort comparisons

  • Area/volume charts that don’t scale properly with data

  • Cherry-picked time ranges or data subsets without context

According to Tufte & Graves-Morris (1983) the Lie Factor is defined as:

Size of effect shown in graphic / Size of effect in data

A value of 1.0 indicates accurate data representation.

Visual Encoding

The way you encode data visually determines how effectively viewers can extract information. Different visual variables work better for different types of data. Research in graphical perception shows that certain visual properties are processed pre-attentively (instantly, without conscious effort). Bertin’s foundational work established that visual encoding must match the logical structure of the data being represented. This principle remains central to modern cartography and information design Bertin, 1967. Using the wrong encoding forces viewers to work harder to understand your data. These visual properties encode different types of information at different effectiveness levels Table 1.

Table 1:Visual variables from Bertin, 1967.

Visual VariableBest ForEffectiveness
PositionPrecise comparisons, continuous dataHighest
SizeQuantities, hierarchiesHigh
Color (Hue)Categorical differences, qualitative dataHigh
Value (lightness)Ordered sequences, quantitative rangesHigh
Texture/PatternCategorical distinctionsMedium
OrientationDirectional data, flowMedium
ShapeCategory identificationLower (hard to compare)

A key principle is to match the esncoding to your data type. These include:

  • Categorical data → Use hue (distinct colors), shape, or texture

  • Ordinal data (ranked) → Use value/lightness, size, or position

  • Quantitative data (continuous) → Use position, size, or value (lightness)

  • Temporal data → Use position (typically left to right) or animation

Simplicity

Minimizing clutter and too many items in a map maximizes clarity and focuses the audience on your key messsages. Viewers have limited cognitive capacity. Every element competing for attention reduces the clarity of your message. Research in visual perception consistently shows that simplified, focused visualizations communicate more effectively than complex, decorated ones.Your visualization should communicate its message as efficiently as possible. Every element should earn its place.

Edward Tufte’s foundational concept of the data-ink ratio advocates for maximizing the amount of ink devoted to representing data in any data visualization Tufte & Graves-Morris, 1983. Tufte argues that “non-data-ink” or decoration should be eliminated or minimized.

Some examples of items to eliminate:

  • Chartjunk - Decorative elements that don’t convey data (excessive text, north arrows, scale bars, 3D effects, background images, unnecessary grid lines)

  • Redundant encoding - Showing the same data in multiple ways

  • Competing visual elements - Elements that distract from the main message

  • Excessive labeling - Text that clutters rather than clarifies

For Maps:

  • Remove unnecessary boundary lines

  • Use basemap details that provide context (not distraction)

  • Simplify your color scheme

  • Zoom to your area of interest

  • Clear, minimal labeling

For Charts:

  • Remove decorative gridlines (or make them subtle)

  • Use only the necessary axes

  • Choose bar charts or dot plots over pie charts (when comparing values)

  • Consistent, limited color palettes

  • One insight per chart

Visual Hierarchy

Guide viewer attention and integrate all elements intentionally. Don’t add items such as north arrows and scale bars just because everyone else does. A well-designed visualization controls the order and emphasis of information, guiding viewers from the most important insights to supporting details. Modern visualization experts including Few (2013) emphasize the importance of visual hierarchy in directing attention and improving comprehension.

Use these techniques to emphasize importance:

  • Size - Important elements larger than supporting elements

  • Color - Highlight key data with contrasting colors; use muted tones for context

  • Position - Place primary insights prominently (top-left for readers of left-to-right languages)

  • Contrast - Use white space and contrast to separate important information

  • Saturation - More saturated colors draw attention; desaturated colors recede

All elements of a map should work together:

  • Title - Clear, specific (not “US Population” but “Population Growth in the US, 2010-2020”)

  • Legend - Essential but not intrusive; only show what’s necessary. Sometimes the legend can be part of the explanatory text.

  • Data Sources - Always cite your data sources

  • Annotations - Use strategically to highlight key insights without overwhelming

  • Color scheme - Consider colorblind-friendly palettes

A reminder of some basic design principles

  • Consistency - Use the same colors and symbols throughout a series of maps/charts

  • Balance - Distribute visual weight; avoid one-sided emphasis

  • Rhythm - Repetitive patterns (when intentional) can emphasize patterns in data

  • Gestalt Principles - Humans group similar elements together; use this intentionally

Viewers naturally scan visualizations with a specific eye pattern. By understanding this, you can guide them to your key insights first, then let them explore details. This makes your visualization more persuasive and memorable.

References
  1. Bertin, J. (1967). Sémiologie graphique: Les diagrammes les réseaux, les cartes. (No Title).
  2. Tufte, E. R., & Graves-Morris, P. R. (1983). The visual display of quantitative information (Vol. 2). Graphics press Cheshire, CT.
  3. Few, S. (2013). Information dashboard design (2nd ed.). Analytics Press.