Cartogram Construction
- Introduced by
- Cartogram tradition (diffusion method by Gastner & Newman; circular method by Dorling)
- Year
- 2004
- Type
- Map transformation that rescales region area to represent a variable
- Family
- Cartographic transformation / anamorphic mapping
What does it do?
A cartogram is a map in which the area of each region is rescaled so that it is proportional to some variable — population, votes, GDP — rather than to its true geographic size.
The aim is to correct the visual bias of ordinary maps, where large but sparsely populated regions dominate the eye while small, populous ones nearly vanish, by making each region as big as the quantity it represents. Cartogram construction is the family of techniques that produce these value-by-area maps, ranging from contiguous density-equalizing diffusion to non-contiguous circle and rectangle methods, each balancing the accuracy of areas against the recognizability of shapes.
The intuition behind the method
On a normal map of a country, a vast empty desert province shouts while a tiny crowded capital whispers, even though far more people live in the capital. A cartogram fixes this by inflating and shrinking regions until each one's size matches the thing you care about — so a state with ten times the population becomes ten times the area, and the map shows where the people, the votes, or the money actually are. The catch is that you cannot resize regions arbitrarily while keeping the map both connected and recognizable: something has to give. Contiguous cartograms keep neighbours touching but warp shapes; non-contiguous ones keep shapes crisp but pull regions apart. The art is choosing how much distortion of familiar geography you will tolerate to make the variable legible.
History
The idea of resizing regions by value goes back at least to the early twentieth century, and value-by-area cartograms were popularized in the 1930s and refined manually for decades, but automated construction long resisted a satisfactory solution. Waldo Tobler pioneered computer cartograms in the 1960s and 1970s, and Daniel Dorling introduced his circular non-contiguous cartogram in the 1990s as a robust, legible alternative. The decisive advance came in 2004 when Michael Gastner and Mark Newman published a diffusion-based method in PNAS that produced smooth, contiguous, density-equalizing cartograms efficiently, and its election maps of the United States made the technique widely known. Subsequent work, including faster flow-based algorithms by Gastner, Seguy and More, has made high-quality cartograms routine in modern mapping software.
When to use it
Use a cartogram when the geographic size of regions would mislead readers about a variable that is really about people or activity — most famously for election maps, where vast rural districts visually overwhelm dense urban ones on a conventional map. They suit population, votes, economic output, disease burden, and similar count-like variables where you want the eye to weigh regions by quantity rather than land area. Contiguous diffusion cartograms are best when keeping the map connected and roughly recognizable matters; Dorling and rectangular cartograms when easy value comparison outweighs geographic fidelity. They are poor choices when precise location, true distances, or unfamiliar regions are essential, or when the audience cannot invest effort in reading a deliberately distorted map; in those cases a normalized choropleth or proportional-symbol map is safer.
Limitations and considerations
- Distorts familiar geography, so unfamiliar or oddly shaped regions can become unrecognizable.
- There is an unavoidable trade-off between area accuracy, shape fidelity, and topology.
- Requires positive values and careful handling of zeros, negatives, and very large outliers.
- Readers untrained on cartograms may misread the distorted map or distrust it.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Removes the area-size bias of ordinary maps, sizing regions by the quantity that matters.
- Makes count-like variables such as population and votes directly comparable by area.
- The Gastner-Newman diffusion method preserves contiguity and topology while equalizing density.
- Non-contiguous and Dorling forms give crisp, easily compared shapes for clear value reading.
Applications
- Election maps that size districts or states by population or electoral votes rather than land area.
- Visualizing world or national population so dense regions are given visual weight.
- Mapping economic indicators such as GDP or trade volume by country.
- Showing public-health burdens like case or death counts proportionally across regions.
- Communicating any count-based statistic where land-area bias would distort the message.
How it works
How it works
Construction begins by setting, for every region, a target area proportional to the value being mapped, so that a region with twice the population should end up with twice the area on the cartogram. A common scaling constant converts the variable into map units while preserving the total mapped area. This target defines the goal of the transformation; the methods that follow are different ways of deforming or replacing the geometry to hit those targets as closely as possible.
Formula
A_i \propto v_i
The Gastner-Newman method produces a contiguous cartogram by treating the mapped variable as a density and letting it flow until it is uniform everywhere, exactly as a physical substance diffuses to equalize concentration. The density field is evolved under the diffusion equation, and the regions' boundaries are carried along by the resulting flow, so areas of high density expand and areas of low density contract until every region's density is equal — meaning its area is now proportional to its value. Because the boundaries move smoothly with the flow, the regions stay connected and topology is preserved, though shapes are deformed.
Formula
\frac{\partial \rho}{\partial t} = \nabla^2 \rhoAn alternative abandons the true region boundaries entirely and replaces each region with a simple shape sized by its value — a circle in the Dorling cartogram, a rectangle in others — with the circle radius proportional to the square root of the value so that area is proportional to the value itself. The shapes are then nudged apart by a repulsion rule until they no longer overlap, ideally staying near their true locations and preserving rough adjacency. This sacrifices recognizable outlines and contiguity but makes the encoded values extremely easy to read and compare.
Formula
r_i \propto \sqrt{v_i};\quad \text{repel circles to remove overlap}Every cartogram is a compromise, so the final step is to judge whether the chosen method strikes the right balance for the audience and message. The analyst weighs area accuracy — how closely each region's size matches its target value — against shape and topology recognizability and the preservation of relative position, and may tune algorithm parameters or switch methods accordingly. A cartogram that is perfectly accurate in area but unrecognizable, or recognizable but inaccurate, fails its purpose; good construction finds a usable middle ground.
Formula
\text{trade off } |A_i - k v_i|\ \text{against shape/topology recognizability}Common pitfalls
- Assuming distance, direction, or precise location remain meaningful on a deliberately warped map.
- Using a cartogram for a variable that is a rate or ratio, where a choropleth conveys it better.
- Choosing a contiguous method when the regions become unrecognizable, frustrating the reader.
- Forgetting that extreme outliers can swamp the layout, shrinking everything else to invisibility.
Frequently asked questions
What is the difference between a contiguous and a non-contiguous cartogram?
A contiguous cartogram, such as the Gastner-Newman diffusion type, keeps every region connected to its true neighbours while deforming their shapes so that each area becomes proportional to its value; the map stays a continuous, if warped, sheet. A non-contiguous cartogram, including the Dorling circular form, instead resizes regions independently and lets them separate or be replaced by simple shapes, sacrificing connectivity and outline in exchange for accurate areas and clear, easily compared symbols.
How does the Gastner-Newman diffusion method work?
It treats the mapped variable as a density spread over the map and then lets that density flow according to the physics of diffusion — the same equation that describes how heat or a dissolved substance spreads to even out concentration. As the density equalizes, the region boundaries are carried along by the flow, so dense regions swell and sparse ones shrink until every region has the same density, which means its area is now proportional to its value. Because the boundaries move continuously, the regions stay connected and their topology is preserved.
When should I use a cartogram instead of a choropleth map?
Use a cartogram when the quantity you are mapping is a count or total — population, votes, cases, dollars — and you want regions to be visually weighted by that amount, correcting the bias that makes large empty areas dominate a normal map. Use a choropleth when you are mapping a rate, ratio, or density and want to keep true geography intact, shading each region by its value. The two are often most powerful together: a cartogram for magnitude and a choropleth for intensity.
Related methods
Sources
Gastner, M. T., & Newman, M. E. J. (2004). Diffusion-based method for producing density-equalizing maps. Proceedings of the National Academy of Sciences, 101(20), 7499–7504.
Open DOI referenceCopied.