As the 2012 US presidential race heats up, we will undoubtedly see lots of politically motivated visualizations. This is great, but we want to caution everyone to think carefully about what you see. Like with any medium for communication, it is possible for data to be misrepresented. The visualizations are often biased because they are aiming to promote one side while discrediting or at least diminishing the other. In addition, the people who make these things are often new to data visualization. Sometimes they use the wrong visualization type for the data, or just don’t understand how a visualization works.
Both sides of the presidential race are guilty of some questionable visualization practices. Here’s one from each side.
Venn and Euler diagrams are one of the simplest types of visualization to understand. They represent sets of items with specific qualities, and the overlap shows the items that have qualities in common. Unfortunately in this case, the Venn diagram has been completely misused. Here’s an attempt at putting the diagram in plain english: President Obama’s promise to lower health insurance premiums by $2,500 is in group A. The “result” (Annual health insurance premiums have increased by $2,393) is in group B. The promise and the result have in common a $4,893 price increase over Obama’s promise.
Obviously this is the wrong visualization to represent this data. The relationship between the promise and the result is not two sets with a subset of things in common. The relationship is clearly quantitative between two numbers. A bar chart showing what Obama promised vs. the result would be a much better choice. Using the wrong visualization isn’t the only issue here, though. The Affordable Health Care Act passed in March of 2010, with some of its provisions not scheduled to come into effect until 2019. The data for the survey came out in September 2011, when many of the provisions designed to cut costs had not been enacted yet.
As a visualization viewer, think critically about every step of the process, including the data sourcing and the recency or relevancy of the data.
This map shows locations to which Romney outsourced jobs. The problem is that a choropleth map is the wrong kind of map to use. At the scale of the world, the actual locations are the size of a pinpoint — a few cities scattered around the globe. Using a choropleth map shows the size of the land that these cities are in, and has nothing to do with the actual quantity of jobs that were outsourced.
If it had to be a map, pinpoints showing the cities that the jobs moved to would be a better choice. Ideally, though, the best visualization for this data would be a bar chart showing how many American jobs were shipped to different locations. On another bar chart, it would also be relevant to show how many people Romney’s companies employ in America vs. other countries.
There are several issues causing these problems. First, political campaigns are new to data visualization. They don’t necessarily know how to use it correctly, and they don’t recognize that incorrect usage looks like lying, whether it was intentional or not. Second, there are definite motives behind producing these, with a clear goal to exaggerate differences between each candidate. This motivation encourages exaggeration and misuse of all communication mediums, including visualizations. When you see a political visualization, remember to always think critically about the visualization, the data, and the source.