Thematic maps display one subject — income, rainfall, migration, disease — on a geographic frame. Each technique fits particular data: choropleths shade areas by rates, proportional symbols size circles by amounts, dot density scatters points for counts, isoline maps draw contours of continuous fields, flow maps trace movement. Choosing the wrong type is the most reliable way to build a misleading map with correct data.
The area-based family
Choropleth maps shade enumeration units by value — the workhorse of statistical mapping, with one iron rule: map rates and ratios, never raw counts, or large empty districts shout while dense small ones vanish. Classification scheme and class count shape the story; defensible breaks belong in the legend.
Cartograms trade geographic fidelity for value-proportional area — powerful for population-weighted stories, disorienting if readers need real geography.
Points, lines and surfaces
Proportional and graduated symbols map absolute quantities cleanly and survive the count-versus-rate trap. Dot density conveys distribution and intensity intuitively (one dot = n units) but places dots randomly within units — a subtlety readers rarely suspect. Isoline maps (contours, isotherms) express continuous surfaces; heat maps show event density and inherit every caveat about population baselines.
Flow maps encode movement — trade, commuting, migration — where width carries magnitude and geometry must be generalised or the map becomes spaghetti.
Matching technique to data and question
The decision tree is short: Is the phenomenon continuous (isolines/surfaces) or discrete? Attached to areas (choropleth for rates, symbols for amounts) or points (dots, symbols)? About movement (flows)? About density of events (kernel maps with baselines)? Multi-variable stories may justify bivariate schemes — used sparingly, they reward; used casually, they bewilder.
And every choice interacts with the zonation: the modifiable areal unit problem stalks all area-based mapping, so test robustness across unit schemes when stakes are high.
Frequently asked questions
Why should choropleth maps not show raw counts?
Because area and population distort perception: a vast sparsely peopled district with 100 cases looks graver than a small dense one with 1,000. Normalise to rates (per capita, per km²) so shading compares like with like.
When is a heat map appropriate?
For visualising event density where the baseline is roughly uniform — or where you explicitly normalise by population or opportunity. Raw event heat maps over inhabited landscapes mostly rediscover where people live.
Related topics
Need this done professionally?
GLOBEIR delivers this work for government and enterprise clients.