«
»

Choropleth Limitations


*This post was updated on 1/27/2016.

County level choropleth map of planted soybeans by area, created by PetersonGIS via some crazy USDA data wrangling efforts. This data and the associated geo-analytics (gathered for most of the country) were used as part of the inputs to a Colorado State University climate model.

Choropleth mapping, which shows the distribution of a variable within an enumeration unit, is a staple in every cartographer and GIS professional’s arsenal. However, it is always good to keep in mind the limitations of our tools, and the choropleth map does have some. These limitations are:

  • The distribution of the variable is static at the scale of the map. No inference can be made about the variable at a more detailed scale. For example, the distribution of noxious weeds can be shown by state within the U.S. However, the distribution of noxious weeds within states (by county, perhaps) cannot be determined from a state map.
  • It can look like the variable is constant across the unit, possibly resulting in map-reading failure. (Similar to above.) For example, if a state is colored red to denote a high level of noxious weeds, this may make the map reader think the entire state is over-run with noxious weeds when perhaps it is only one area of the state.
  • The change between the enumeration units can appear to be quite crisp when in reality the variable is probably much smoother across those boundary lines. One state might be green for noxious weeds and a neighboring state could be red even though the counties on either side of the state line are more alike than different.
  • Usually, you don’t show the actual value for the variable, you show the area-weighted or population-weighted value. This could be a limitation if a future map-reader wants to use your values for a different analysis or visualization because the raw data counts aren’t available, only the ratio values. We aren’t generally concerned about this limitation as cartographers, but if you feel like your data will be needed as raw data values in the future, the raw data could be supplied along with the map in a database or table.
  • Make certain that your choropleth shows area-weighted or population-weighted data when needed (hint: almost always). For example, a map of TV watchers will probably resemble a population distribution map and should therefore be normalized by population such as the number of TV watchers in each county divided by the number of people in each county.
  • The enumeration units may not have much to do with the variable. Noxious weeds probably have much more to do with land use and soils, for example, than with the state administrative boundaries.
  • If your study units are of widely different sizes it can be difficult to show the color shading adequately in the smaller units. For example, in a map of U.S. states the smallest units – D.C., Rhode Island, Delaware – can be difficult to see. Some modern maps alleviate this difficulty by using hexagons as the enumeration units, with each hexagon signifying a state, for example. However, these are only useful for well-known geographic units like U.S. states or Admin 0 level data (countries).

NPR Hex Map

  • Related to the above point, misinterpretation can occur if a very large unit is colored brightly. If a large state has the same noxious weed / area ratio as a smaller state, and they are both colored bright-red, the larger state may appear to have a bigger problem even though the ratio is the same.

You may also be interested in these posts:

Choropleth vs. Heat Map

Choropleth Legend Format

Your donation in support of this blog is very appreciated! It’s simple to make a donation with PayPal using my link.

 


This entry was posted on February 12, 2011, 7:07 pm and is filed under Best Practices. You can follow any responses to this entry through RSS 2.0. Both comments and pings are currently closed.

Comments are closed.