Choropleth Maps on the CIESIN Site

by 

Dr. James R. Carter, Geography-Geology Department

Illinois State University, Normal, IL, USA

CIESIN provides a nice set of choropleth mapping programs on their web site.  They have given permission for my students to make maps from these sites and post them to the web as long as they acknowledge that the maps were generated on the CIESIN site and give a link to that site.  As of June 2003 there were three different mapping programs on the CIESIN site.  The newest package is more powerful and flexible than the earlier versions and that newest package is the one being discussed here.

Start by going to the CIESIN site at  http://www.ciesin.org and drop down to Data and Information: Online Tools and Applications. 
Or, go directly to:  http://plue.sedac.ciesin.columbia.edu/plue/ddviewer/  This takes you to Online Tools and Applications, then to United States Demographic Data Viewer.  From this page select the Java Edition v3.0.  This is the version that is being used in this discussion.  

The base map that comes up is shown below.  It is a base map of the U.S. with Alaska and Hawaii fit in on the south edge of the conterminous 48 states.  Alaska and Hawaii are not at the same scale as the conterminous U.S.  The map projection employed is the equirectangular projection, a simple projection of latitude and longitude values onto a Cartesian grid.  On this grid there is no convergence of meridians towards the north.  This is not an equal area projection.  All maps on this site are on this projection, including maps of the individual states.

Below the map base are buttons giving many options.  Note the Submit Job button.  After you have made selections, you need to toggle this button to create a new map.

First, you want to select a variable to map, assuming you are going to make a map of the U.S.  Hit the button Select Vars and you will get a page as below.  Note first that this is 1990 Census data.  We can anticipate that soon 2000 Census data will be incorporated into the site, but as of summer 2003 it is still 1990 data.  Look at the broad subject areas that are available:  Population, Income, Education, Employment, Housing and more.  For this exercise I selected percent of the total population that is age 65 years or older.  This is not the number of persons 65 and older, but that variable is available.  Having clicked on this variable it is shown in the smaller box on the right.  You can select more than one variable at a time, but start with just one.  Click Close to return to the map.

After selecting that variable and hitting Submit Job, the following map was created.  This is the default map.  The legend is in the upper left corner.  It shows data broken into four classes and the classes are labeled 1 Low, 2, 3, and 4 High.  The color scheme is yellow to red, in a fairly good graded series.  In this case the data are classified based on Quartiles, such that 1/4th of the states are in each category.  The legend does not show actual values in each class.

Note the Customize Map button above.  Clicking that button I was able to put the data into 7 categories broken into equal intervals. I also repositioned the legend to put it in the lower right corner of the image.  I added a title and positioned it below the map.  I could have included a sub-title.  I took the default color scheme in this example, but I did change the color of the text to a yellow.  In general, this is a good gradation of colors, but it is difficult to distinguish easily the top three shades of red.

On the Customize Map site, there are three pages.  On Ranging Method page below you can select from 2 to 10 classes of Quantiles or Equal Intervals.  On this page you select the colors for the ranges of data.  Double click on the yellow box labeled First Range and pick a new color.  Likewise, double click on Last Range and pick another color.  The program then fills in the intermediate colors.  Everyone should play with these color options to get an appreciation for the problems of getting a good graded color scheme. 

The color scheme below was created by selecting 10 intervals and then picking a green for the First Range and a cream for the Last Range.  It was easy.  Is it effective?  It is hard to create 10 colors in a sequence that can be distinguished one from another.  For this reason, you are advised to use no more than six or seven classes in most cases.

There is a Map Colors page on which you can choose the color for titles, boundaries, background and missing values.  It works like the examples above, in that you click on a color and pick a new color from a palette.

The Legends/Title page permits you to turn legends and titles on and off, to give titles and sub-titles, to assign colors to these and position them on the map.  A portion of that page is shown below.  To see the Title, you must click to have a title.  Then type in the text for the title and for a sub-title is you want one.  Click on the Color box and pick a new color.  You can choose the size of the text and the font, as well as whether you want the text in plain, bold, italic or bold italic.

To position the title on the map, click on Set Title.  Then click the X option and move the slider to move the title left to right.  Click on the Y option and move the slider to position the text up or down.  You will probably want to move the Legends/Title page so that you can see where the legend is.  This is not an intuitive interface, but it works to position supplemental information on the map.

Using the processes discussed above, the map was refined to better represent the data.  First, an examination was made of the statistical values.  Alaska has the lowest percentage of persons 65 and older, being only 4.0%.  The next lowest state is Utah, with 8.6%.  So, it seems that Alaska should be in its own class--much lower than any other state.  At the other end, Florida has the largest percent of persons 65 and older, being 18.3%.  The next highest state is Pennsylvania, being only 15.4%.  To find these individual values, I turned on the Label switch below the map and moved the cursor over each state.  In this example, the cursor is over Pennsylvania and we see the state name and the data value. 

Given this information, I chose to set up a custom Ranging Method.  I set the lowest class to include only Alaska.  The next higher class includes only Utah and shows that the value is greater than 8.6% but less than 10.0%.  For some reason the program added many digits to the right of the decimal point in the lowest two classes.  I was not able to fix this problem. 

The range of the other classes is in even steps of 2.0%, thus having breaks at 10, 12, 14, 16 and 18 percent.  I argue that users can better understand classes broken into whole numbers.  But note that I set the top class to go only to 18.3%.  Florida of course is the only state in that class.  But, note also that no state falls into the next to highest class.  So, does this map give a better representation of the distribution of older persons in the U.S.?  I think it does, but it may take some effort to see these distinctions. 

Below is still another variation of this map.  In this case I picked a unique color for each of the seven data classes.  I did not let the program fill in colors between the two extremes.  In this color sequence each color class is distinguishable, at least for my eyes.  We must realize that not everyone is able to distinguish all of these color differences, so that this map may not be effective for everyone.

And below is still another variation of the same map.  In this case I selected colors in the pink to purple portion of the spectrum.  Again, I selected each color and did not let the program interpolate colors between the two extremes.  In this example I changed the size of the legend text as well as the background color.

We have now seen five variations of the choropleth map showing the Percent of the Population that is 65 years of age or older, as reported in the 1990 Census of Population.  These five maps demonstrate the effects of changing the parameters set to make a map.  As one exercises the options in this program one gets a different map.  No one map is correct or right, but some maps are better than others.  

The person making a map has the responsibility to exercise the options to make a good map.  The map should be appropriate to the task at hand and the ability of the audience to read and interpret the map.  The map should be appropriate to the data.  As such it is incumbent on the map maker to look at the numbers and put some effort into deciding how many class intervals to employ and where to put those intervals.

Of course you can always take the default values, such as in the map below.  This is the same default map with values broken into quartiles, but in this case the colors range between a cream and purple.  This is a much different map from those where data were broken into seven classes based on equal intervals or my own classification scheme.  But, this map has value and it is a valid map.
The CIESIN Java 3.0 mapping applet is a powerful tool to let users create choropleth maps and vary the appearances of those maps. The program is not perfect and it has its idiosyncrasies, but it is fairly easy to use and one can produce some attractive and meaningful maps.  Thanks to CIESIN for providing this site with the good databases. 

This page was put together in support of the students of Geo204 - Geographic Techniques at Illinois State University.  A similar page was put together examining the choropleth mapping capabilities of the American FactFinder web site, from the U.S. Bureau of the Census. 

Return to the master page on Choropleth maps

     Return to the master pages of James R. Carter, the author of this page