wunder beta

📘 How do you make charts that tell the truth?

At its core, a chart is a rule that turns numbers into geometry: each data record becomes a visual mark, and each variable is encoded by some property of that mark. Jacques Bertin, in Semiology of Graphics, was the first to systematize this

6
lessons
~30 min
to learn
Adults
level
Start the course →

What you’ll learn

  1. Why Visualization Works: Encoding Data as PicturesExplain how a chart maps data to visual marks and channels, and why this lets the eye decode quantities a table cannot.A data visualization is a systematic mapping from data values to visual marks (points, lines, areas) and the visual channels that vary across them (position, length, color, area). Because human perception is fast and parallel, a well-chosen mapping lets readers decode patterns, comparisons, and outliers that are invisible in a raw table. The whole discipline rests on choosing encodings that the eye reads accurately and that faithfully represent the underlying numbers.
  2. Graphical Perception: Choosing Channels the Eye Reads AccuratelyRank the common visual channels by perceptual accuracy and use that ranking to justify chart-type choices.Cleveland and McGill's research established that visual channels are not equally readable: people judge position along a common scale most accurately, then length, then angle and slope, with area, volume, and color shading judged least accurately. This empirically grounded ranking explains why bar and dot charts usually beat pie charts and bubble charts for precise comparison. Using the ranking, you put the variable that most needs accurate reading on the most accurate available channel.
  3. Decluttering: Data-Ink, Chartjunk, and Visual FocusApply Tufte's data-ink ratio and preattentive/Gestalt cues to remove clutter and direct attention to the data.Edward Tufte argued that a graphic's ink should overwhelmingly carry data, defining the data-ink ratio and the term 'chartjunk' for non-data decoration that distracts. Maximizing data-ink within reason, plus exploiting preattentive attributes and Gestalt grouping, lets a designer steer the reader's attention to the signal rather than the frame. The aim is not austerity for its own sake but clarity: every removed element should leave the data easier to read.
  4. Graphical Integrity: Avoiding Charts That MisleadDiagnose distorted charts using the lie factor, the principle of proportional ink, and correct scale and axis conventions.A chart has integrity when the visual magnitude of an effect matches the magnitude in the data. Tufte's lie factor quantifies distortion as the ratio of effect-shown to effect-in-data, and the principle of proportional ink (Bergstrom and West) requires shaded areas to be proportional to the values they represent, which is why filled bar charts must start at zero. Recognizing truncated axes, area-encoded one-dimensional data, dual axes, and distorting aspect ratios lets you both avoid and audit deceptive graphics.
  5. Color, Type, and Annotation in Service of the MessageSelect data-appropriate, accessible color palettes and use typography and annotation to make a chart self-explanatory.Color choice should follow the data type: sequential palettes for ordered magnitudes, diverging palettes for data with a meaningful midpoint, and qualitative palettes for unordered categories, as systematized in Cynthia Brewer's ColorBrewer. Accessible design accounts for color-vision deficiency by relying on luminance and redundant encoding rather than hue alone. Clear titles, labels, units, and well-placed annotations convert a chart from raw geometry into a self-contained argument the reader can trust and follow.
  6. Guided Project: Build and Defend a Visualization ArtifactApply the course's principles end-to-end to redesign a flawed chart into a defensible artifact and justify each choice in a case memo.In this capstone you choose a real dataset and a flawed source chart, diagnose its problems with the course's vocabulary, then redesign it and write a short case memo defending every decision. The memo ties each design choice to a named principle, the task it serves, and the integrity check it passes. The deliverable is a small artifact (the redesigned chart) plus the memo, demonstrating that you can both build and justify a principled visualization.

Questions this course answers

In Munzner's framework, what is the difference between a 'mark' and a 'channel'?

Marks are the geometric building blocks (points, lines, areas), while channels are the visual properties of those marks (position, length, area, hue, luminance, angle) that vary to carry data values.

The 'expressiveness' principle of visualization is violated when a designer:

Expressiveness requires encoding all of the data and only the data. Mapping an unordered category onto an ordered channel like position falsely implies a ranking, communicating something the data does not contain.

Why can a scatterplot reveal an outlier that a same-sized data table hides?

The visual system processes position pre-attentively across many marks at once, so a point in an extreme location pops out, while reading a table is a slow, serial, cell-by-cell process.

According to Cleveland and McGill's ranking of elementary perceptual tasks, which channel is decoded MOST accurately?

Their experiments found position along a common scale is decoded with the least error, which is why bar and dot charts on a shared axis support the most accurate comparisons.

Why does a dot plot generally support more accurate value comparison than a pie chart of the same data?

Pie slices rely on angle and area judgments, which Cleveland and McGill found are read less accurately than position along a common scale, the channel a dot plot uses.

A choropleth map encodes magnitude with color luminance even though luminance is low on the accuracy ladder. What is the design justification?

On a map, position is committed to location, so magnitude must use a remaining channel like luminance; this trades reading precision for the geographic context the map provides.

Grounded in trusted sources

  • Jacques Bertin, Semiology of Graphics: Diagrams, Networks, Maps (University of Wisconsin Press, English ed. 1983; reissued ESRI Press, 2010)
  • Tamara Munzner, Visualization Analysis and Design (CRC Press, 2014)
  • William S. Cleveland and Robert McGill, 'Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods,' Journal of the American Statistical Association, Vol. 79, No. 387 (1984), pp. 531-554
  • William S. Cleveland, The Elements of Graphing Data, revised ed. (Hobart Press, 1994)
  • Edward R. Tufte, The Visual Display of Quantitative Information, 2nd ed. (Graphics Press, 2001)
  • Colin Ware, Information Visualization: Perception for Design, 4th ed. (Morgan Kaufmann, 2021)

Every Wunder lesson is built from real, reputable sources — never invented.

Related courses

📘Why isn’t the chart the story?📘How does hypothesis testing decide evidence?📘How does SQL SELECT actually read?📘Joins: SQL Analytics📘How does pandas shape tabular data?📘How do you automate a data workflow?📘Why do features make the model?📘How do classifiers draw the line?

Wunder is a personalized learn-anything platform — tell it any topic and it builds a beautiful, fact-checked course in minutes, with narration, a knowledge check, and a college-style University track.

All topics · Home

© 2026 Wunder Learning LLC · Terms & Privacy