Dangerous Liaisons - UW Libraries

June 2, 2017

Geospatial Literacy? Yes, Maps Need Critical Evaluation Too

Kian Flynn

In the aftermath of the 2016 election season, “information literacy” and “news literacy” were much buzzed about terms in the popular media, and librarians across the country were called upon to battle the scourge of “fake news” and deceptive information in our “post-truth” world. PBS, The Chicago Tribune, The Seattle Times, and many other news organizations all dedicated space on their platforms to talk about the efforts librarians are making on these fronts. These certainly were not new efforts on librarians’ parts, but the efforts were now imbued with a new sense of urgency and purpose.

Over the past several months, The UW Libraries has been at the forefront of advocating for greater information and news literacy education on campus. On April 1, the Libraries hosted a conference featuring Attorney General Bob Ferguson (video of his interview) on the importance of news literacy skills in a democracy. Participants engaged in interactive sessions on topics including finding and evaluating government information and being a savvy news consumer. An exhibit (Making Sense of the News), currently on display on the ground floor of the Suzzallo Library, and an accompanying LibGuide (Savvy News Consumers) curated by Communications Librarian Jessica Albano walk patrons through evaluative criteria for determining the reliability of a piece of information. The LibGuide has received over ten thousand views since going live earlier this spring!

And last week, the UW Libraries hosted the UW GIS Symposium, which featured a keynote address from Tableau Research Scientist (and UW Geography alum) Dr. Sarah Battersby on the symposium’s theme – the importance of geospatial literacy. As the below Google Trends graph shows, geospatial literacy has not received the same attention as its information, news, and media brethren.

No love for Geospatial Literacy

No love for Geospatial Literacy.

With interactive and GIS maps commonplace in today’s academic and media environments (in one article last year, the NYT offered up 50 handy maps of the U.S. cultural divide), Dr. Battersby’s excellent and informative talk underscored the importance of being able to critically evaluate maps because research has shown that individuals can sometimes view maps as “more real than experience.” As part of her talk, she spotlighted some examples of bad (or perhaps deceptive) maps and stressed that librarians and GIS professionals have the power to educate the world by encouraging critical geospatial thinking skills. I’m sharing these examples with her permission (as long as the reader critically evaluates the maps).

Exhibit A. Source: USGS

Exhibit A. Source: USGS


Exhibit B. Source: Google Maps (this is the current Google Maps)

Exhibit B. Source: Google Maps (this is the current Google Maps) — you may need to click on the image to spot the deception.



C. Source: Neue Zürcher Zeitung (German daily newspaper). A map showing the range of North Korean missiles.

Exhibit C. Source: Neue Zürcher Zeitung (German daily newspaper). A map showing the range of North Korean missiles.


What bad or deceptive practices are employed by these maps? Scroll down to the bottom to find out!

As hopefully many readers of this blog know, we now subscribe to Data-Planet — the world’s largest repository of standardized and structured statistical data. Data-Planet provides a one-stop shop for a wide variety of geospatial data from authoritative sources such as the Census Bureau and the United Nations. Data-Planet provides some very basic map-making capabilities within its interface, but any geospatial data in the database can also be exported as a shapefile into a program like ArcMap (available on UW Libraries Access+ computers), where more sophisticated data and geospatial analysis can be performed.

A few minutes after Dr. Battersby spoke on the importance of Geospatial Literacy, I delivered a lightning talk at the symposium on how Data-Planet can be used for geospatial analysis. I used data from Data-Planet that showed SNAP (Food Stamps) participation rates, poverty rates, election results, and population counts by U.S. county. I was interested in how closely SNAP participation rates and poverty rates mirror each other. There is certainly correlation and causation: SNAP eligibility is based on a number of qualifications, which include having a net household income that is 100% of the poverty level. Nationwide, the poverty rate (15.5%) is about 1 percent higher than the SNAP participation rate (14.4%).

However, when we dig into the county-level data and display that data on a map of the United States we start to see a more nuanced pattern in action. There is a pronounced geography to what I’ll call the “social services gap”–the percentage difference between the poverty rate and SNAP participation rate in a given county. Counties in rural “middle America” are more likely to have poverty rates greater than SNAP participation rates. We can see that take shape in these maps.


The geography of the social services gap

The geography of the social services gap. On the left, black represents counties where the SNAP participation rate is greater than the poverty rate.

I spent most of my five minute talk promoting Data-Planet and demonstrating how it could be used for data and geospatial analysis. After Dr. Battersby’s talk, however, I wished I had devoted more time to speaking about the critical analytic choices mapmakers are forced to make when they visualize their data and how the maps I made for the presentation could be construed as being misleading.

There are a few exceptions I (and others) could take with my own maps if I were to avoid addressing these issues:

  1. The national poverty rate is 1.1% higher than the national SNAP participation rate. A county that has a poverty rate that’s 1% higher than its SNAP participation rate would be displayed in black on the map to the right even though it has a social services gap below the national average. If I move the cutoff point to 1.1% instead of 0%, my map might look very different. Why did I choose the cutoff point I did? I wanted to show which counties were potentially being under-served by social services, even if they didn’t stand out from the national average.
  2. Whitman County in Washington has a 28% poverty rate and only a 8% SNAP participation rate for a 20 point social services gap. Ferry County in Washington has a 22% poverty rate and a 21% SNAP participation rate for a single point gap. Both counties look identical on the map to the left, even though they both tell a very different story about the social services gap. Why didn’t I show color or shading gradations? I liked the stark contrast of the above maps and thought they were easier for the eye to understand, even if they didn’t account for some of the nuance that a map with shading gradations would have supplied.
  3. And, finally, like many of the electoral maps we’re used to seeing, the maps exaggerate vast, lightly population counties and diminish small, densely populated counties. Los Angeles County, which has a social services gap of 7%, is hardly discernible on the map despite the 10 million people that live there. Cherry County in Nebraska, which has a social services gap of 6%, can almost be pointed out from space despite only having a population a tad above 30 thousand people.

No map is perfect, but maps can still tell powerful stories and help make complicated datasets more digestible for a broader audience. Gene Balk, the Seattle Times’ FYI Guy, recounts such a case in this article from the fall of 2015 that describes how maps Gene Balk put together of the foreign-born population in King County using Census data changed how a local nonprofit conducted their community needs assessments.

My hope is that with tools like Data-Planet at our community’s disposal, more students can make sense of data in a way that is informative and revelatory. As librarians, it’s our responsibility to make sure that students have the information and geospatial literacy skills to make those stories accurate and honest.

-Kian Flynn


A few post-scripts:

  • More maps from the lightning talk!Election Results by County, 2016

On the left, we see counties carried by Hillary Clinton in the 2016 presidential election. Dark blue represents counties with a SNAP participation rate greater than their poverty rates. On the right, counties carried by the Republican candidate. Dark red, likewise, represents counties with a greater SNAP participation rate.

  • More stats from the lightning talk that help supplement the above maps (sometimes a few numbers are easier to process than a map with over 3000 pieces of information):

1. 80% of counties have SNAP participation and Poverty rates within 5% of each other

2. The median population size of counties without a social services gap is 36,523. The median population size of counties with a social services gap is 20,871. This reflects a potential urban/rural divide.

3. Counties where the SNAP participation rate is more than 5 percent greater than the poverty rate favored Hillary Clinton by nearly 13 points (56.2%-43.8%) in the 2016 election (this analysis omitted third-party candidates and abstentions). In all other counties, 50.9% voted for Clinton and 49.1% voted for Trump.

4. 50.9% of the counties won by Clinton in 2016 had a social welfare gap; 65.7% of the counties won by Trump had a social welfare gap.

5. On average, a 4% increase in the social services gap correlated with a 3% decrease in the Democratic vote for president in 2016.


  • Why are the “bad maps” bad?

Exhibit A. So many reasons.

  1. Eight percent of males are red-green colorblind. This map would be inaccessible to them.
  2. Besides that, the choices for the rest of the color scheme don’t follow any comprehensible logic.
  3. This is basically a population map. States with lots of people use more water than states with fewer people. Surprise!

Exhibit B. You have to zoom into the bottom right and find the scale bar to figure this one out. Because that Google map is a Mercator projection, the scale bar is going to be terribly inaccurate most everywhere on that map. Try to measure the distance that the scale bar says is 1000 miles up near Greenland and you’ll find that that distance actually represents about 300 miles closer to the poles.

Exhibit C. This is a fun one! One, the center of the bulls-eye is not actually on North Korea. Two, more troubles with the Mercator projection. A more accurate and updated version of the map was eventually published in the paper:

Corrected map

Corrected map