A choropleth map is a type of thematic map in which predefined geographic areas are filled with colors, shades, or patterns according to a statistical value associated with each area. Common mapping units include countries, provinces, counties, municipalities, and census tracts. Choropleth maps are widely used to display population density, percentages, rates, and other comparable regional measures. Each area receives a visual symbol representing its aggregate value; the map does not normally show how that value varies within the area.
Principle and suitable data
Choropleth mapping combines geographic boundaries with an attribute table. Each geographic unit is matched to a numerical observation, and its fill is determined by that observation. In cartography, this technique is especially useful for showing the regional distribution of a variable whose values are meaningfully compared across areas.
Suitable variables commonly include:
- Population density, such as residents per square kilometre.
- Percentages, such as the proportion of households with internet access.
- Rates, such as events per thousand residents during a specified period.
- Averages or medians, such as median household income.
- Indices calculated consistently for all mapped units.
The distinction between extensive and intensive quantities is important. Extensive quantities, such as total population or total production, generally increase with the size of the population or territory being measured. Intensive quantities express a relative magnitude, such as density, a proportion, or an average.
Mapping raw totals is technically possible, but often produces a misleading comparison: a region may have a larger count simply because it contains more people or covers more land. A proportional symbol map is frequently more appropriate for totals because symbol size can represent magnitude without assigning that magnitude to the entire area.
Normalization
Normalization converts counts into quantities that support the intended comparison. For area , a rate may be expressed as
where is the count of interest, is the relevant denominator, and is a scaling factor. For population density, the denominator is land area; for a percentage, it is the relevant total population or number of observations, with .
The denominator must match the question. A count divided by total population answers a different question from the same count divided by households, workers, or land area. Normalization does not automatically make data comparable: definitions, observation periods, population coverage, and collection methods must also be consistent.
Historical development
An influential early choropleth map was published in 1826 by Charles Dupin. It used graduated shading to depict differences in schooling across departments of France. During the nineteenth century, the technique spread alongside official statistical reporting and the development of population censuses.
The term choropleth was introduced by the American geographer John Kirtland Wright in 1938. It draws on Greek roots referring to area and quantity.
Computer mapping and geographic information systems later made choropleth production routine. Modern software can join statistical tables to geographic boundaries, calculate rates, assign class breaks, and produce interactive maps. These capabilities simplify production but do not resolve the statistical and perceptual choices involved.
Classification methods
Most choropleth maps group numerical values into a limited number of classes, each represented by a distinct fill. Class breaks are the thresholds separating those groups. Their selection can substantially change the apparent pattern even when the underlying data remain unchanged.
| Method | Principle | Main strengths and limitations |
|---|---|---|
| Equal interval | Divides the numerical range into intervals of equal width. | Easy to explain, but skewed data may leave some classes empty or place most areas in one class. |
| Quantile | Places approximately equal numbers of geographic units in each class. | Distributes units across the palette, but may separate very similar values or group substantially different values together. Tied values complicate equal counts. |
| Natural breaks | Places boundaries between clusters in the observed values. | Can reveal groupings, but produces dataset-specific breaks that may hinder comparison between maps. |
| Standard deviation | Defines classes by distance from the mean, often in standard-deviation units. | Highlights departures from the mean, but may be less intuitive and may fit strongly skewed distributions poorly. |
| Fixed or meaningful thresholds | Uses externally defined values or substantively important cutoffs. | Supports interpretation and comparison, provided the thresholds are relevant and documented. |
A common natural-breaks method is Jenks optimization, which seeks class boundaries that minimize variation within classes for a chosen number of classes. Its numerical objective does not establish that the resulting classes are the most meaningful for every purpose.
The number of classes creates a trade-off. Too few classes conceal variation; too many make colors difficult to distinguish. The appropriate number depends on the distribution, palette, display size, and intended audience.
Unclassed choropleth maps
An unclassed choropleth map assigns colors through a continuous numerical scale rather than a small set of discrete intervals. This avoids abrupt changes at selected class breaks, but readers may find precise values difficult to infer from small color differences. A continuous legend and access to numerical values are especially useful.
For maps comparing multiple years or regions, a shared classification and color scale usually make comparison more defensible. Recalculating breaks separately can cause the same color to represent different values in different panels.
Color and legend design
Color should communicate the structure of the variable rather than merely distinguish adjacent regions.
- Sequential schemes progress from low to high, usually through an ordered change in lightness. They suit quantities such as density or percentage.
- Diverging schemes extend in two directions from a meaningful midpoint, such as zero change or a specified reference value.
- Qualitative schemes distinguish categories without implying numerical order. They are appropriate for nominal categories, but not normally for an ordered choropleth.
A diverging palette can exaggerate a division if its midpoint has no substantive meaning. Conversely, an arbitrary sequence of hues may obscure ordering. Color-vision accessibility, contrast, printing conditions, and background colors all affect readability.
A legend should identify the mapped variable, units, class limits, and treatment of missing values. Interval notation must make clear where boundary values belong. Missing, suppressed, and unavailable observations should not silently receive the color assigned to zero.
Geographic representation
Choropleth maps generally use polygon boundaries stored as geospatial vector data. Statistical records are linked to polygons through geographic identifiers. Matching names alone can introduce errors when names are duplicated, spelled differently, or changed over time.
Boundary dates matter. A value collected for an earlier administrative geography cannot necessarily be attached directly to current boundaries. Comparisons may require reconstruction or allocation to a common geography, introducing additional uncertainty.
Area and visual prominence
Large polygons occupy more visual space than small polygons, regardless of their population or statistical importance. A sparsely populated rural region may therefore dominate a map while a densely populated city is barely visible. Insets, alternative geographic units, or complementary charts can help expose these differences.
Map projection also affects apparent area. Equal-area projections preserve relative area, although they do not eliminate the broader problem that geographic size differs from population size. The map scale and degree of cartographic generalization determine whether small units remain legible.
Interpretation and limitations
Aggregation and boundary effects
A polygon’s fill represents its aggregate value, not a measurement at every location inside it. A district with a population density of 200 people per square kilometre may contain both a dense town and extensive uninhabited land.
The modifiable areal unit problem describes how statistical results can change when observations are aggregated at different scales or grouped using different boundaries. Apparent clusters and contrasts may therefore depend partly on the selected geography rather than solely on the underlying phenomenon.
Sharp color changes at administrative borders do not necessarily indicate abrupt real-world transitions. Equally, uniform shading within a polygon does not demonstrate internal uniformity.
Ecological inference
The ecological fallacy occurs when relationships observed for groups are assumed to hold for individuals. For example, an association between two district-level averages does not establish that the corresponding characteristics are associated among individual residents. Choropleth maps show aggregate patterns and cannot by themselves establish individual behavior or causation.
Uncertainty and small denominators
Rates based on small populations can fluctuate greatly. Two areas with the same estimated percentage may have very different levels of uncertainty because their sample sizes or data quality differ.
Conventional choropleths usually give equally definite fills to estimates of unequal reliability. Supplementary uncertainty maps, confidence intervals in interactive displays, or clearly marked suppression can make these differences visible. Statistical smoothing can reduce instability but changes the displayed quantity and must be distinguished from the original observations.
Visual clusters also do not constitute a formal test of spatial autocorrelation. Class breaks, neighboring areas, geographic size, and the underlying data distribution all influence apparent clustering.
Variants and related techniques
A bivariate choropleth displays two variables simultaneously, often with a two-dimensional color legend. It can reveal combinations such as high values of both variables, but is harder to decode than a single-variable map and requires careful control of the number of categories.
Dasymetric mapping uses ancillary information, such as land cover or residential locations, to redistribute aggregate quantities into more plausible subareas. Unlike a conventional choropleth, it attempts to represent variation within the original reporting units. Its results depend on the allocation assumptions and quality of the ancillary data.
Dot-density maps and proportional symbol maps offer alternative ways to show counts. By contrast, contour maps and maps based on spatial interpolation generally represent continuously varying surfaces rather than values assigned to reporting areas. Interpolating aggregated regional values does not automatically recover the true distribution within those areas.
Applications
Choropleth maps are common in demography, economics, education, environmental reporting, and public administration. They support regional comparison, identification of broad patterns, and communication of geographically organized statistics.
In a GIS workflow, essential checks include verifying geographic joins, distinguishing missing observations from zeros, confirming appropriate denominators, inspecting the numerical distribution, and documenting classification choices. Interactive displays can provide exact values, denominators, dates, and uncertainty alongside the map, but remain subject to the same aggregation and visual-design limitations as printed choropleths.