Lab: Color and Mood, with Proportional Symbols

One of two projects with statistical techniques, this one involves using proportional or graduated symbols to represent data.  This method is especially useful to avoid the pitfalls of choropleth mapping, particularly when your mapping units are all different sizes.

For this map, we’ll be looking at refugee populations, either by country of origin or country of destination.  Mapping this for the whole world is challenging, so I encourage you to choose a region, such as the eastern Mediterranean, or sub-Saharan Africa.  You may want to look at the data before deciding.

Getting the Data

You’ll need to download countries from Natural Earth and the refugee data from the World Bank.  For the refugee data, this link takes you to data by country of origin, but you can switch to country of asylum in the panel to the right of the graph.  You may use either, but be sure you know which one you selected!  Download the data as an Excel file.

Data Preparation

  1. Add the countries to your map.
    1. Dissolve on SOV_A3, so that we have one record for each country, instead of one for each tiny island of a country.
    2. Keep the dissolved version, and remove the other one from your map.
  2. Open the Excel file and examine the data. You should have two tabs of metadata and one of data.  On the Data tab, we need to format it so ArcMap can read it.
    1. Locate the header row (has labels that identify each column).
    2. Delete every row above the header row so that the header row is now row 1.
    3. Save As to save the file with a meaningful name in your Working folder.
    4. Add the new file to your map. Note that Excel files can’t simply be dragged & dropped – use the Add Data button for best results.
  3. Join the Excel table to the shapefile. Explore both the shapefile and the Excel table to know which fields to base your join on. You may find it useful to uncheck Keep All Target Features in this case, as we only want to see countries with data.
    1. After you join, make sure to export to a new shapefile to make the join permanent.
    2. Go back and remove the join from the original layer so you can use it as a basemap.

Just symbolizing the data as proportional/graduated symbols at this stage will place the symbol, whatever it is, at the center of a particular country (and also do strange things to the polygons). To better control the symbols and their locations, we’ll use a tool to create points from the countries.

  1. Feature to Point tool: Use your dissolved countries layer as the input feature and leave all other options as the defaults. This will produce a point feature with the same attribute information as the polygon.

When you set up your map, your selected region should (obviously) be at the center, but this leaves us with a problem: Some large countries may automatically have their symbols placed out of view once you zoom in. Instead of expanding your view extent, just move the points to the part of the country that is visible (First, make sure to create a bookmark so you can quickly return to your current view extent):

  1. Edit the points (as needed): Switch to the Edit ribbon and use the Select tool to select a point you want to move. Zoom out until you can see all the center points. Click on the points you will need to move and move them to where you will be able to see them at your bookmarked view extent.  (Don’t forget to turn off selectability on the other layers!) Use the Move tool to move a feature, then click the green checkmark when you’re done moving it.  Save your edits.  You can do as many iterations of this process as you need to get the points in the correct locations.

Pro Tip: You may not know what region you want until after you symbolize the points.  Editing can be hard on large circles, because you need to find the exact center.  Set the points back to single symbol and they will be much easier to move.

Symbolizing the data

Once your points are ready, use proportional or graduated symbols to map a particular year’s refugee count. Consider the differences between the two methods and decide which better represents your message.  Make sure the legend is displaying the appropriate units. Do you think you should apply Flannery’s compensation? Ask yourself these questions, and experiment before you make your final, educated decision.

Choose a size range for your symbols that gives you some cohesion in dense areas, but doesn’t totally obscure everything.  Remember that proportional symbols have a size for each value, while graduated symbols use classes, just like a choropleth.  Decide which best tells the story of your data.

  • You may find it helpful to use Vary symbology by attribute to make the larger circles more transparent, so they don’t blot everything out.

Label the countries so people know where these things are happening, and if terrain is relevant to the story you’re telling, try the shaded relief from Natural Earth or Shaded Relief Archive – just don’t let it overwhelm your data.  Once you’ve selected a final region, don’t forget to set an appropriate map projection.

Tips for Success

  • Pick your map extent after you look at the data.  You might be telling the story of the Syrian Refugee Crisis, or the Rwandan Genocide, but how much area does your map need to show?  It’s often best to determine this after the first basic symbolization, when you can see the area of impact of an event.
  • Make the symbols bigger than you think.  The default setting for proportional and graduated has a top size of something like 18 pt, which is not nearly large enough.  In most cases, you are talking about the mass migration of millions of people, and the map should demonstrate the magnitude of that.  It’s okay if they overlap – this topic should hit you in the feels, and tiny circles won’t do that.
  • Label with commonly recognized names.  The data contains the full formal name for all the countries, but consider your audience.  The shorter, more common names are probably a better choice, both for readability and aesthetics.
Advertisement