Resizing regions by their data value fixes a choropleth's area bias but makes the map hard to recognize
Aliases: cartogram · value-by-area map · Dorling cartogram
What it is
A cartogram resizes each region so that its area on the map is proportional to its data value, removing the choropleth's "large region, large visual weight" bias at its source. The cost is just as direct: region outlines are stretched, compressed, or replaced entirely, breaking the familiar geographic shapes readers rely on. Two common forms exist: the contiguous cartogram keeps shared borders intact while allowing heavy distortion, and the Dorling cartogram replaces each region with a circle or square sized by its value, preserving adjacency only approximately.
Why it happens
The choropleth's area bias comes from a mismatch: color carries the data, but area still belongs to geography. A reader's visual weight is roughly the product of color intensity and region area, so a sparsely populated large region outweighs a dense small one. A cartogram turns area itself into the data channel, aligning visual weight with value and eliminating the "large-but-pale versus small-but-dark" conflict at its source. The loss of recognizability has a mechanism too: shape is the primary cue for locating a region on a map. Stretching breaks outline matching, and circles or squares remove outlines altogether, leaving readers to reconstruct geography from position and labels. Note also that human area perception is imprecise on its own; the cartogram fixes the weighting error, not the precision of reading exact values.
Where it stops holding
Whether a cartogram helps depends on the task and the reader. When the task needs geographic context—locating one's own region, judging real adjacency or distance—a distorted geometry is a net loss, and for readers unfamiliar with the underlying map a cartogram is close to unreadable without a reference. Comparisons become fairer on a cartogram, but absolute position and distance are gone. The deformation algorithms carry free parameters (iteration count, initial layout), so the same data can yield differently shaped results that should not be compared map-to-map.
Applying it
- Switch to a cartogram only when area bias is itself the message (for example, a population-weighted election map); otherwise keep the choropleth and state what area represents in the caption.
- On first use, place the geographic basemap side by side with the cartogram so readers can build the correspondence between the two representations.
- With a Dorling cartogram, keep circle centers at approximately true geographic positions with adjacency preserved, and label regions in the legend.
- Verify by asking readers unfamiliar with the map to find a named region on the cartogram; if they cannot, add the basemap or fall back to a choropleth.
Related
- Same group: U12.02.1 Large-area regions gain disproportionate visual weight in choropleth maps · U12.02.2 Choropleths should map rates rather than absolute counts · U12.02.3 Equal-interval, quantile, and natural-breaks classifications yield different map shapes · U12.02.4 The number of classes sets the grain of the visible spatial pattern
- Nearby: U12.01.3 The Mercator projection severely inflates areas at high latitudes · U12.02.1 Large-area regions gain disproportionate visual weight in choropleth maps
- Search terms:
cartogram·Dorling cartogram·contiguous cartogram
Cards in the same group
- U12.02.1A choropleth draws the eye to whichever region is biggest on the map, regardless of its value
- U12.02.2Coloring a map by raw count mostly draws a population map, not the pattern it claims to show
- U12.02.3Slicing the same values by equal interval, quantile, or natural breaks produces three different-looking maps
- U12.02.4Three color classes show only the roughest pattern; seven start revealing the finer structure