Skip to content
Module 04 of 1040 min readMixed

Tufte's principles, distilled

Data-ink ratio, small multiples, sparklines, chartjunk. The rules from The Visual Display of Quantitative Information that still hold in 2026.

40%

Listen along

Read “Tufte's principles, distilled” aloud

Plays in your browser using on-device text-to-speech — nothing leaves the page.

Learning objectives

By the end of this module, you should be able to:

  • 01Compute the data-ink ratio of any chart and improve it
  • 02Use small multiples to compare patterns across groups
  • 03Recognise and remove chartjunk in your own work

Edward Tufte's The Visual Display of Quantitative Information (1983) remains the most influential book on data visualisation ever written. Most of its principles are forty years old, and most are violated daily in analyst work. The principles are about cognition — how the human eye and brain process visual information — and cognition has not changed since 1983.

Data-ink ratio

Tufte's central idea: of all the ink on a chart, what fraction is conveying data? The data-ink ratio should be maximised. Everything else — gridlines, redundant axes, 3D effects, drop shadows, background colours — competes with the data for the reader's attention.

The minimalism test

For each visual element on your chart, ask: if I removed this, would the reader lose any information? If no, remove it. Apply this iteratively. The first pass removes the obvious chartjunk. The third pass produces a publication-grade chart.

Small multiples

Instead of one chart with five overlapping lines, draw five small charts arranged in a grid — each showing one line, with the others shown in light grey for context. The eye can compare patterns across grid cells far more easily than untangle a single jumbled chart. Tufte called these 'small multiples'; ggplot calls them facets; the FT calls them 'a small-multiple series'. Same thing.

text
BEFORE (single chart, 5 lines, illegible)
──────────────────────
/ ────────────
/───/────────
/───/
────/
AFTER (small multiples, one panel per series)
[Asia] [Europe] [Africa] [Americas] [Middle East]
───── ── ──── ───── ──
Each axis labelled, same scale, comparison instant.
Small multiples turn 'figure out the mess' into 'compare panels'. Cognitive load drops by an order of magnitude.

Chartjunk — Tufte's list, updated for 2026

  • 3D effects on bar and pie charts (distort perception, no information gained).
  • Gridlines heavier than 0.5pt grey (compete with the data line).
  • Frames around the entire plot area (waste of pixels).
  • Legends when an inline annotation would do (eye-jumps cost cognitive energy).
  • Drop shadows, gradients, and 'professional' Excel styling (signals lack of taste).
  • Background colours other than white (only justified for editorial brand identity).

The sparkline

Tufte's other gift: the sparkline. A small, axis-less line that lives inline with text or in a table cell, showing the shape of a series. Bloomberg terminal table views are dense with sparklines for a reason — they pack temporal context next to a number in a way that pure numbers cannot.

Exercise

Pull a chart from your last presentation. List every visual element on it. For each, decide: is this conveying data? If not, what would the chart look like with it removed? Redraw it minimally. How does the redrawn version compare?

Key takeaways

  • Data-ink ratio: maximise the ink that conveys data; eliminate ink that doesn't.
  • Small multiples beat one cluttered chart almost every time.
  • Tufte's rules from 1983 still apply because they're about cognition, not aesthetics.
Loading progress…
LeadAfrikPublic Economics Hub