Choropleth Maps and Their Limitations
Choropleth maps are one of the most widely used geospatial visualization types — they use color intensity or hue to encode a quantitative variable across geographic regions (countries, states, counties, census tracts). They are effective at showing spatial patterns and regional variation at a glance, and familiar enough to most audiences that they require minimal interpretation overhead. However, they have significant methodological challenges that every geospatial visualization practitioner must understand.
The biggest choropleth map problem is the size confounding issue: large geographic regions visually dominate the map regardless of their relative importance in the data. A choropleth map of U.S. county population density shows sparsely populated counties (which are geographically large) visually overwhelming the map, while densely populated urban counties (which are small) are barely visible. This makes the map a visualization of land area more than of the variable of interest. Solutions include: using a cartogram (a distorted map where region area is proportional to the data variable rather than geography), adding size-proportional symbols over the regions (combining a choropleth with a bubble map), or explicitly noting the size discrepancy in the chart caption.
Classification method is the most important methodological decision in choropleth map design. Quantile classification divides the data into equal-sized groups by value rank — ensuring each color class contains the same number of regions, but potentially masking meaningful differences when values are bunched. Natural breaks (Jenks optimization) identifies class boundaries where the data has natural gaps, preserving the distribution shape. Equal interval divides the data range into equally sized bins, which is intuitive but can produce poorly populated classes if the data is skewed. Standard deviation classification places class boundaries at standard deviation intervals around the mean — appropriate when the distribution matters for interpretation.
Normalization is critical for choropleth accuracy. A choropleth of raw crime counts will make large cities look like crime hotspots compared to rural areas — simply because large cities have more people. Normalizing by population (crimes per 100,000 residents) produces a crime rate map that shows genuine geographic variation in relative risk. The choice of denominator in normalization is a significant analytical decision: normalizing crime by population produces a different map than normalizing by land area, or by number of businesses, or by number of potential victims. The appropriate denominator depends on the analytical question.