Election Maps: Representing Area and Population

Maps of election results are a staple in the US during campaign season.  They appear in countless newspaper articles, TV news stories, and blog posts.  Given the state-by-state nature of our electoral presidential election system, these maps are very valuable in understanding the state of an election race.

They have a problem, however, that is common to many maps.  While maps represent area quite well, they can be deceptive when the data being visualized corresponds to something that doesn’t correlate well with area.  For election maps, that “not correlated with area” variable is voter population.  This is the reason that  traditional maps, the one with red and blue color-coded states, end up exaggerating the votes of rural areas.  See this map from the website 270towin.com showing the final 2017 election results.

traditional_map

Those big states like Montana and Wyoming contribute a lot of red to the map.  But they don’t contribute much to the 270 electoral votes required to win the election because of the small population of those states.  To overcome this limitation, many alternative charts have been developed which distort the geography in some way.  Here is one common approach, from a Business Insider article, in which the map’s state adjacency relationships are preserved but the areas are distorted dramatically.  It’s often called a cartogram.

but-of-course-the-us-population-is-not-evenly-distributed-among-the-states-this-map-distorts-the-sizes-of-the-states-to-reflect-their-varying-populations

You can see that in this map, Montana and Wyoming are much smaller than New Jersey because of their respective populations.  However, irregular areas are difficult for us to compare.  An alternative space-warping approach is to use a square or hexagonal grid to approximate the map.  This was done, for example, on the website FiveThirtyEight.  Here is their final “forecast” map (which, you might notice, is not an exact match for the final results of the election).

538_map

The last two maps are more proportional to how the population voted, but they no longer represent the states in their geographic locations.  The maps so far have either shown geographic area (map 1) or the magnitude of the vote (maps 2 and 3).

But can we do both?  The following map from the Financial Times is both simple in design, and effective at conveying both population and geography.  This version of the map shows 2012 results. Notice how you quickly see how Montana and Wyoming are part of a red but sparse section of the country.

ft_dot_map

I recently adopted a similar design to visualize the popular vote results from the 2016 election.  I use the same underlying map and dot metaphor, but this time with two categories for each state.  Each dot represents approximately 100,000 voters in the 2016 election.  You can see how close most states were, the areas of strength for both candidates, as well as the differences in population density across the United States.  Note that this map illustrates the 2016 popular vote, which Hillary Clinton won by nearly 3 million votes.  It does not show electoral votes, which pushed Donald Trump to victory despite the fact that he had fewer voters supporting him.

2016_votes_by_state