A toolkit to help your reader

Scientific visualization

Seeing clearly, showing honestly, telling better stories

How to build low-friction paths from questions, through evidence, to understandable interpretations.

The role of visualization

Visualization bridges an expertise gap.

  • You know the data because you analysed it.
  • Your audience meets abstractions (marks, colours, labels, and scales).
  • A good visualization guides them from the question, through the relevant evidence, to a warranted interpretation.

Start with the reader's task

Weak start

"What chart should I use?"

Better start

"How can I clearly convey a well-supported narrative?"
"What comparisons do I want the reader to make?"

Your purpose leads to a question, the question defines the reader task, and the reader task determines the visual construction.

Your role is important

A visualization is a guided argument, not a decoration after the analysis.

Storytelling does not mean exaggeration. It means controlling the order in which the evidence becomes understandable. You provide a path through the evidence.

  1. Question What are we asking?
  2. Comparison What must be compared?
  3. Evidence What pattern supports it?
  4. Interpretation What may we conclude?
  5. Caveat What uncertainties matter?

Explore, explain, or exemplify?

Explore Open the data for investigation, many patterns may emerge.
Explain Guide attention toward a specific finding. This is your interpretation.
Exemplify Make an abstract quantity concrete or memorable. Build intuition.

Think about what your reader needs. Exploration invites investigation; explanation guides interpretation; exemplification builds intuition.

Part 1

The grammar of visualization

A chart is assembled, not selected

A chart type is not a primitive object. It is a combination of marks, channels, coordinates, scales, and context.

Marks carry data through channels

A visualization maps variables onto properties of marks: position, length, hue, lightness, shape, area, slope, and sometimes motion.

Not all channels are equally precise

Position on a common scale is usually easiest for quantitative comparison. Area, angle, colour intensity, and volume need more care.

Coordinate systems change the easy comparison

Cartesian, polar, and geographic coordinates organise space differently. Changing coordinates changes what the reader can compare quickly.

Scales translate values into distances

Linear, percentage, relative-growth, logarithmic, temporal, categorical, and diverging scales answer different questions. Ask: what does equal visual distance mean?

Make uncertainty part of the result

Intervals, dots, ensembles, and fans communicate different uncertainty concepts. Choose the form that matches what the uncertainty means.

Context completes the visualization

Titles, subtitles, labels, units, annotations, sources, sample sizes, definitions, and methods are active parts of the argument.

Part 2

How perception organises the chart

Gestalt principles are design operations

Proximity, similarity, enclosure, connection, continuity, alignment, and whitespace determine what viewers treat as belonging together.

Hierarchy: colour, salience, and first read

Colour and salience guide attention before conscious reasoning. Use them to guide the first comparison, then test whether the message survives colour-vision differences and grayscale.

Reduce working-memory tax

Legends, distant labels, and unexplained symbols force the audience to decode while you want them to reason.

Based on Franconeri et al.: direct labeling and annotation often reduce memory load.

Axes and scales are ethical design decisions

Truncated axes, log scales, area encodings, and baseline choices can be valid, but only when the reader can see what changed.

Part 3

Chart types as combinations of choices

Same data, different perspective

Visualize your data from multiple angles

A table facilitates easy lookup; a scatterplot supports pattern finding. Switch between modes of visualization to see different aspects of the data..

Reader task becomes chart construction

Chart types are recurring solutions: ranking uses a common scale, time uses temporal position and connection, relationships use two quantitative positions, and flows use connection, direction, and width.

Time series: choose the tradeoff

Use the visual form that matches the task: follow one trend, explain a jump, or untangle many individual patterns.

Categories: rank, compare, compose, or show spread?

Category charts are not interchangeable. The right form depends on whether the reader needs order, paired comparison, composition, or spread.

Maps: use geography only when geography matters

A map is excellent when spatial pattern explains the result. If the task is ranking, a chart often beats geography.

Source: By Carter et al. in prep 2026, figure made by Casper van Elteren using Ultraplot.

Flows: show direction and conservation

Flow diagrams work when the reader can follow direction and check what is conserved. They fail when every branch gets equal attention.

Part 4

Assemble the visual argument

Same summary, different shape

Similar averages and correlations can come from very different distributions. Think about your metrics and, for example, show the x-y shape before trusting a single number.

Inspired by Anscombe-style datasets and Franconeri et al. on global statistics, comparison, and working-memory limits.

Case study — information graphic

One visual, a whole campaign

Charles Joseph Minard’s 1869 map makes the changing scale of Napoleon’s 1812–13 Russian campaign visible at a glance. It is a classic information graphic because every encoding serves one question: what happened to the army on this journey?

Carte figurative, Charles Joseph Minard (1869). The chart's band width is proportional to the number of troops.

Read the encodings

Width
Estimated troops present: the band begins at 422,000 and ends near 10,000.
Colour + route
Tan traces the advance; black traces the retreat through the mapped places.
Time + temperature
The lower strip locates dated winter temperatures on the return journey.

Read critically: the graphic pairs losses with cold; it does not, by itself, prove what caused each loss.

Orient, guide, explain, qualify

OrientWhat is this about?
GuideWhere should I look?
ExplainWhat comparison matters?
QualifyWhat should I not conclude?

Titles, direct labels, annotations, highlighting, and captions implement these stages. They guide interpretation without hiding the science.

Before publishing: check the visual argument

  • Is the total or baseline clear? Percent of what, and relative to which year or group?
  • Are units, scales, transformations, and methods explicit?
  • Can the audience see uncertainty, sample size, or scope limits?
  • Would another honest chart change the apparent conclusion?
  • Does the annotation guide attention without concealing counterevidence?

Activity - critique, redesign, and sketch

Diagnose the visual argument, then plan an infographic

In groups: mark the question, claim, comparison, encoding, attention path, and caveat. Then make a rough first design.

  1. What should the chart make easy to see?
  2. What became difficult or disappeared?
  3. Which claims are supported by your infographic?
  4. What caveat does it require?

The most important part: Be explicit about why we need your work. What question does it answer, and what are the consequences?

Take-home rules

  1. Storytelling is the order in which evidence becomes understandable.
  2. Build charts from marks, channels, coordinates, scales, and context.
  3. Choose encodings by the reader task, not by chart-menu habit.
  4. Use perception to guide attention, not to hide evidence.
  5. Optimize the story, but never edit away the science.

Sources and further reading

  • Franconeri, S. L., Padilla, L. M., Shah, P., Zacks, J. M., & Hullman, J. (2021). "The Science of Visual Data Communication: What Works." PDF
  • Datawrapper. "Data vis do's and don'ts" and chart-type guidance. datawrapper.de
  • Rodrigo Zamith. "Data-Driven Storytelling." Working parts of a chart
  • Charles Joseph Minard (1869). Carte figurative des pertes successives en hommes de l’armée française dans la campagne de Russie 1812–1813. Original public-domain record: Wikimedia Commons.
Charles Joseph Minard's 1869 flow map of Napoleon's Russian campaign: a wide tan band advances toward Moscow, a narrow black band retreats, and a temperature record runs beneath the map.