Lying with Maps

Government is distrusted. The media is distrusted. Experts of all kinds are distrusted. Basic facts are distrusted. It's good to be skeptical but when dogma replaces reason it is a recipe for disaster. Unfortunately this is our reality.

It's tempting to give up and hope that the problem will somehow go away. But that would mean giving in to the charlatans and ideological zealots; a course of action that has never worked out well in the past. In the United States we like to believe that our system and Constitution immunize us against the self-destructive impulses that have caused much suffering in the world. Not so. Greediness and a willingness to blame "others" for our problems is deeply embedded in the human psyche and Americans are no different. I'm holding on to hope that the majority of people will not be fooled. Which requires making the effort to counter blatant lies and misrepresentations even though I know that the zealots won't care.

Since maps are my thing I'm going to start with the maps that claim to represent voter sentiment in the United States following the 2016 presidential election. Maps representing the distribution of votes on a county-by-county basis have been widely distributed. And voting data represented in this way results in a map where the United States is mostly red with small scattered islands of blue. An example of this can be seen here on the Breitbart website where the authors take the deception to new heights by claiming that Donald Trump "won" the popular vote by a landslide. The only qualification being that they only count votes from counties that Donald Trump won. That's just outright, blatant, lying.

The more common source of misrepresentation emerges from two basic statistical errors. When maps are used to convey statistical information (voting totals are summary statistics) the map maker must follow the same basic rules that apply when representing data in a graph. Specifically, a consistent scale must apply to the areas used to represent the values. Since counties have no inherent meaning (or scale) as a basis for representing votes in a presidential election maps that show votes on a county-by-county basis inherently misrepresent the information. This is known as the the modifiable aerial unit problem; a well known basis for lying with maps.

The second issue really stems from a deliberate desire to misrepresent voter sentiment. Representing a county where 47.7% of voters selected one candidate and 47.4% selected the other candidate as all "red" or all "blue" is just dishonest. When those red and blue counties are put together on a national map the intent is to deceive.

Here's my map showing the presidential election results on a map of the contiguous United States. I colored each county based on the outcome of the national popular vote. Since Hillary Clinton won the national popular vote, all counties are shaded blue. This map is absolutely accurate. And it misrepresents the data in the same way that the "sea of red" maps do.

A very good explanation of how maps can misrepresent election results.