Maps that mislead: choropleths, population and why per-capita matters

TUREFF• •0

Maps are some of the most shared visualisations online, and some of the most easily misread. A map feels authoritative, but the way regions are shaded, sized and grouped can create an impression the data does not support. Here is how to read maps critically, and how to make better ones.

What a choropleth is

A choropleth map shades regions, such as countries, states or council areas, according to a value. Election maps and maps of income or disease rates are typical examples. They are popular because they are simple to make and attractive to look at. They also have built-in weaknesses.

Problem one: big areas look important

On a choropleth, the visual weight of a region depends on its land area, not on how many people live there.

A large, sparsely populated rural region takes up far more of the map than a dense city with many times more residents.
Election maps are the classic example. A country can look overwhelmingly one colour when the vote was close, because the winning party's support was spread across large rural areas.
Better options: a cartogram, which resizes regions by population; a hexagon or square map giving each unit equal size; or proportional symbols, such as circles sized by the number of people, placed on the map.

Problem two: maps that are really population maps

Many maps of totals end up showing where people live, rather than anything about the topic.

A map of total crimes, total cases, total shops or total complaints by region will usually look much like a population map. More people, more of almost everything.
That makes it look as if the biggest cities have the biggest problem, when per person they may not.
The fix is to normalise. Divide by population to get a rate per person, per 1,000 or per 100,000. Or divide by the relevant base, such as cases per tests carried out or accidents per vehicle kilometre.
Totals are not always wrong. If you need to know where resources should go, raw numbers matter. The question is which one answers the question being asked.

Problem three: the class breaks

To shade a map, continuous values are usually grouped into a handful of bins. The cut-off points change the story.

Equal intervals can leave almost every region in one colour if the data is skewed.
Quantiles put the same number of regions in each bin, which can make tiny differences look dramatic.
Custom breaks can be chosen to emphasise, or hide, a pattern.
What to check: read the legend carefully. Look at the actual ranges in each colour band, not just the colours.

Problem four: the boundaries themselves

Different-sized units are hard to compare. Rates in small areas with few people jump around randomly from year to year, so the most extreme values often come from the smallest populations.
Averages hide variation inside each region. A county-level average can hide a very rich town next to a very poor one.
Changing the boundaries changes the result. Grouping the same data into different areas can produce different patterns, a known issue in geography.

Problem five: projection and framing

Many web maps use the Mercator projection, which makes areas near the poles look much larger than they are. Greenland looks similar in size to Africa, but Africa is far larger.
Cropping the map or choosing which regions to show can also shape the message.

A quick checklist for reading maps

Is it showing totals or rates? If totals, is it just a population map?
What is the denominator for any rate?
What are the colour bands, and how were the breaks chosen?
Are large, empty areas dominating the picture?
How small are the smallest units, and could their values be unstable?
Where does the data come from, and when?

Making better maps

Normalise by the right base for your question.
State the units and the date clearly in the title or legend.
Use a sequential palette for ordered values and a diverging palette only when there is a meaningful middle.
Consider a chart instead. A ranked bar chart is often clearer than a map when geography is not the point.

Have you seen a map that changed meaning once you looked at the legend? Share it in the comments.