Visual and Data Communication

A table of numbers can hide the very pattern it contains. A good chart makes that pattern visible in a glance — a trend, a gap, an outlier — doing in one image what paragraphs of prose would only approximate. Showing beats telling when the point is a shape in the data.

But the same power that reveals can deceive. A truncated axis, a misleading area, or a screen of decoration can push a reader toward a conclusion the numbers do not support. Communicating visually is therefore an ethical skill as much as a design one: you are shaping what someone believes is true.

Show, don't tell

Reach for a visual when the message is a relationship — how something changes over time, how groups compare, how two variables move together. If the point is a single number, a sentence is clearer than a chart. Match the form to the message: lines for trends, bars for comparisons, scatter plots for correlation. The question is never "what looks impressive" but "what will a reader understand fastest."

Graphical integrity

Edward Tufte, in The Visual Display of Quantitative Information (1983), argued that a graphic should tell the truth about the data. His lie factor measures this: the size of the effect shown in the graphic divided by the size of the effect in the numbers. A value near one is honest; far from one, the picture exaggerates or understates. The classic offenders are bar charts whose vertical axis does not start at zero, and images scaled by width and height so that doubling a value quadruples the visible area.

Clarity over decoration

Tufte named two more ideas worth carrying. The data-ink ratio is the share of a chart's ink that actually encodes data; the higher, the better. Chartjunk is everything that does not — heavy gridlines, 3-D effects, gratuitous images, needless color. Every removable ornament is a small tax on the reader's attention. Good visual communication is closer to editing than to illustration: you keep cutting until only the meaning is left.

A team wants to show that customer complaints spiked after a policy change. In a table, the jump is buried among twelve monthly rows. Plotted as a simple line with the change marked, the spike is unmistakable in a second, the axis starts at zero so the rise is honestly scaled, and there are no gridlines or shadows competing for attention. One chart, one message, nothing wasted.

The same data becomes a 3-D bar chart with a vibrant gradient, a background image of a call center, and a vertical axis starting at 40 instead of 0. Now a modest rise looks like a cliff, the tilted bars are hard to read against each other, and the decoration draws the eye away from the numbers. It looks more designed and communicates less — and its lie factor is well above one, so it is not merely ugly but dishonest.

During the Crimean War, Florence Nightingale found that most British soldiers at the Scutari barracks hospital were dying not from battle wounds but from preventable diseases spread by filthy conditions. To persuade officials who could not read a statistical table, she designed what is now called a polar-area or coxcomb diagram — her 1858 "Diagram of the Causes of Mortality in the Army in the East," covering April 1854 to March 1856. Each month was a wedge; the area showed deaths, with a large blue region for preventable disease dwarfing the red for wounds.

The picture made the argument that numbers alone had not: the enemy was sanitation. Nightingale, a serious statistician who in 1858 became the first woman elected a Fellow of the Royal Statistical Society, used data visualization to drive reform. Her example and Tufte's principles point the same way — a visual earns its place when it makes a true pattern impossible to ignore.

• "A more elaborate chart is a better chart." Added dimensions and decoration usually lower the data-ink ratio and bury the point.
• "Charts are objective, so they cannot lie." Axis choices, scaling, and framing can all mislead while every number stays technically correct.
• "Any data deserves a chart." A single figure or a small comparison is often clearer as a sentence.
• "Starting an axis wherever fits looks best is fine." For bar charts especially, a non-zero baseline distorts the size of the effect.

Find a chart you made or saw recently. Remove every element that is not data — gridlines, 3-D, background images, redundant labels — one at a time. Then check the vertical axis: does it start at zero, and if not, is the effect still honestly scaled?

Think Like a Maester: Before adding anything to a chart, ask whether it helps the reader see the truth faster or just makes the chart look busier.

1. When is showing a pattern with a chart clearly better than describing it in words?
2. What does Tufte's lie factor measure, and what value marks an honest graphic?
3. Why can a bar chart whose axis does not start at zero mislead?
4. What is chartjunk, and why does it weaken a visual?
5. How did Nightingale's rose diagram make an argument that a table of numbers had not?

Visual communication turns numbers into something a reader can grasp at a glance — but only when it is both clear and honest. Show rather than tell when the message is a pattern, match the chart to that message, and protect graphical integrity so the size of the effect on the page matches the size in the data. Cut chartjunk until clarity carries the point. Done well, as Nightingale showed, a single image can make a truth impossible to ignore.

• Visualising Data
• Communicating with Impact