This entry is part of the Chart Index, the reference library for the Chart Design Field Guide.

A cartogram redraws a map so that geographic area represents a quantitative value rather than land. A populous state becomes larger; a sparsely populated state becomes smaller. The geography is deliberately distorted so the visual weight of each region follows the subject of the analysis.

That trade is the form's purpose. A conventional map answers where and how much land? A cartogram answers where and how much of the thing being measured?

What it is

Every region begins with a geographic location and a value. The layout then scales each region's area to the value while preserving enough neighbourhood and relative position for the map to remain recognisable.

The hero chart above is a Dorling cartogram of the House of Representatives: each state and territory is a circle whose area is its number of seats, anchored near its geographic centre and nudged apart until none overlap. The country's shape survives — Western Australia is still west, Tasmania still hangs below Victoria — but the picture is now of votes rather than land. New South Wales and Victoria hold more than half the House between them, and the map finally looks like it. Randomise the seat counts and the circles resize in place, jostling their neighbours as they grow.

When to use it

  • Geographic area would give large, lightly populated regions too much visual weight.
  • The reader needs to compare regional totals rather than rates.
  • Broad location and neighbourhood still matter, but exact boundaries do not.
  • The distortion itself helps explain an imbalance, concentration, or mismatch.
  • The audience can be given enough labels and context to recognise the transformed geography.

When not to use it

  • Exact boundaries, routes, distances, or locations are part of the question.
  • The geography is unfamiliar and distortion would make regions impossible to identify.
  • Precise value comparison is required. A sorted bar chart will be faster and more accurate.
  • There are too many small regions to label or distinguish at the intended size.
  • The map is being used only because the data contains place names. Geography must contribute to the reading task.

Design principles

Let area carry the total, not the land

The strongest case for a cartogram is a total mapped onto a country whose regions differ wildly in size. Shaded onto real boundaries, the total is read through the land: a big empty region looks important, a small crowded one looks trivial. The pair below is the hero's data both ways.

House seats as a choropleth
Western Australia's sixteen seats cover a third of the map; the ACT's three are invisible and the Territory's two fill a block the size of France. The colour is right and the impression is wrong.
The same seats as a Dorling cartogram
Now area is seats. New South Wales and Victoria dominate as they dominate the chamber; Western Australia drops to a modest circle in the west; the two territories are small but present and labelled. The geography is approximate and the quantity is exact.

Choose the cartogram family deliberately

Contiguous cartograms distort region shapes while preserving shared boundaries. They retain adjacency but can produce unfamiliar outlines.

Non-contiguous cartograms scale regions independently around fixed centres. Shapes remain recognisable, but gaps open between neighbours.

Dorling cartograms replace regions with circles. They sacrifice boundaries and shape for clean area comparison, reliable packing, and compact labels.

The Dorling form is often the strongest choice for a business audience because its contract is visible: circle size is data, circle position is geography.

Encode value with area, not radius

If one region represents four times the value of another, its mark needs four times the area, not four times the radius. Scaling radius directly exaggerates differences because circle area grows with the square of the radius. The second figure below squares every seat count before scaling, which is exactly what a radius-scaled cartogram does to the reader's eye.

Area proportional to seats
New South Wales has three times Western Australia's seats and its circle has three times the area. The eye reads it as roughly three to one — which is the truth.
Radius proportional to seats
Values squared before scaling, so the radius grows with the seat count instead of the area. New South Wales now looks nine times Western Australia rather than three, and every small region has shrunk toward a dot. The layout is the same; the ratio is a lie.

Preserve the geographic skeleton

Keep regions near their true positions and preserve important neighbourhoods where possible. A cartogram should still read as a transformed map. The layout balances two forces — collision pushes circles apart, geographic attraction pulls them home — and if the attraction is too weak the circles wander until the map is gone. Once the marks are freely rearranged by rank, the result is a bubble chart.

Provide orientation

Use direct region labels, a familiar national outline, or a small conventional inset map. Distortion removes geographic cues, so the design needs to restore the minimum context required for recognition. On the figures in this entry the state codes inside each circle do the work; for a less familiar geography, a faint outline of the real map behind the circles is worth its ink.

Use colour for a second job only

Area already encodes the primary measure. A single restrained hue is often sufficient. Add colour only for a meaningful second variable, such as region type, change direction, or uncertainty, and provide a clear legend.

Colour for a second variable: seats gained or lost at the redistribution
Area is still seats. Colour now says whether the state gained a seat, lost one, or held steady in the last redistribution — Western Australia up in blue, New South Wales and Victoria down in orange, the rest in neutral grey. Two variables, two channels, and neither is doing the other's work.

Make small regions selectable

Small values create small marks. Give them generous interaction targets, concise tooltips, and labels that remain legible without covering neighbouring regions. Do not let the smallest places disappear merely because their measured value is low. The renderer here enforces a minimum radius, so the Territory's two seats still draw a circle the reader can hover.

State what has been distorted

Name the measure encoded by area, its unit, the period, and whether the values are totals or rates. Readers should never have to infer why the map no longer resembles the geography they know.

Anatomy

The composition of a Dorling cartogram
An anatomical guide

A cartogram combines five elements:

  • Geographic anchor — the original centre or location of each region.
  • Area-scaled mark — a transformed polygon or circle whose area represents the value.
  • Displacement — movement required to prevent overlap while retaining broad position.
  • Region identifier — a label or interaction target that makes the transformed geography readable.
  • Area legend — reference sizes when readers need more than relative comparison.

In a Dorling cartogram, the layout balances two competing forces: collision pushes circles apart, while geographic attraction pulls each circle towards its original location. The result is not a conventional map and should not be read as one; position is approximate, while area is analytical.

  • Choropleth — preserves geographic shape and encodes a normalised measure with colour.
  • Value-by-alpha map — keeps boundaries but reduces the visual weight of regions with little supporting data.
  • Hexbin and H3 map — replaces administrative regions with uniform spatial cells for point density.
  • Tilegram — allocates equal, countable tiles to regions while preserving broad geography.
  • Bubble chart — compares area-scaled circles without a geographic constraint.
  • Geographic base map — preserves location and boundaries when geometry itself matters.

Reading list

  • Dorling, D. (1996). Area Cartograms: Their Use and Creation.
  • Gastner, M. T. and Newman, M. E. J. (2004). Diffusion-based method for producing density-equalizing maps.
  • Nusrat, S. and Kobourov, S. (2016). The State of the Art in Cartograms.
  • Tobler, W. (2004). Thirty Five Years of Computer Cartograms.