A chart can make a pattern obvious in seconds, or it can hide the truth behind decoration and distortion. As data becomes part of every job, the ability to read and make honest charts is a basic skill. This article explains how people perceive charts, how to choose the right type and how to avoid the most common distortions.
How people read charts
In a classic study, Cleveland and McGill tested how accurately people judge quantities shown in different ways. Judgements were most accurate for positions along a common scale, as in bar charts and dot plots, and less accurate for angles, areas and colour shading, as in pie charts and bubble charts. The lesson for chart-makers is to encode the most important comparison as position or length whenever possible.
Choosing the right chart
- Comparing categories: a bar chart, sorted by value.
- Change over time: a line chart with time on the horizontal axis.
- Relationship between two measures: a scatter plot.
- Parts of a whole with few parts: a stacked bar, or a pie only when there are two or three slices.
- Distribution of values: a histogram or box plot.
Common ways charts mislead
A bar chart whose value axis does not start at zero exaggerates differences, because bar length is read as quantity. Three-dimensional effects distort sizes. Selecting a convenient time range can turn a long decline into an apparent rise. Using area to show a one-dimensional quantity, such as drawing a coin twice as wide to show double the money, makes the difference look four times as large. Tufte called the ratio of graphic effect to data effect the 'lie factor', and urged that it stay close to one.
وَلَا تَقْفُ مَا لَيْسَ لَكَ بِهِ عِلْمٌ
“And do not pursue that of which you have no knowledge.”
Honesty is part of the craft
A chart is a claim about the world. Presenting data in a way that leads viewers to a conclusion the data does not support is a form of misrepresentation, even if every number is technically correct. Muslim analysts carry an added responsibility: the Qur'an forbids asserting what one does not know and commands justice in testimony. Clear labelling of sources, units and time periods is part of that honesty.
Less decoration, more meaning
Tufte argued for maximising the 'data-ink ratio': the share of a graphic's ink that represents data rather than decoration. Heavy gridlines, background images, shadows and unnecessary legends compete with the data for attention. Removing them, labelling lines directly instead of using a separate legend, and highlighting the one series that matters most usually make a chart clearer.
Colour should carry meaning, not decoration: one strong colour for the main point and neutral tones for context. Many readers have some form of colour-vision deficiency, so important distinctions should not depend on red versus green alone.
A checklist for honest charts
- The title states the finding, and the axes are labelled with units.
- Bar charts start at zero.
- The time range is the relevant one, not the convenient one.
- No 3D effects or decoration that changes how sizes are read.
- The source of the data is stated.
Sources and further reading
- Cleveland, W. S., & McGill, R. (1984). Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association, 79(387), 531–554.
- Tufte, E. R. (2001). The Visual Display of Quantitative Information (2nd ed.). Cheshire, CT: Graphics Press.
- Qurʾān, al-Isrāʾ 17:36.