Listen to this article
As the presidential election results emerge, there is at least one certainty: television reports and websites will display a map of the US and progressively colour states Republican red or Democratic blue, like this one showing the result of Barack Obama’s re-election in 2012.
Although it has become the default representation of US elections, it is arguably a deeply flawed map.
The famous chart is a choropleth — a thematic map in which regions are coloured according to a statistical value. Although there are several well-established alternatives, they involve trade-offs that we wanted to avoid. We set out to find a third way that corrected the limitation of the standard choropleth while avoiding the weaknesses of existing alternatives.
The choropleth conundrum
In 2012, Barack Obama was elected with 51.1 per cent of the popular vote. This was enough to amass 332 electoral votes, or 61.7 per cent of the electoral college. Yet the states whose electors he secured account for only 37 per cent of the area of the United States — 44 per cent if you exclude Alaska.
It is a well-known problem: Because the US population is not distributed evenly, states with large areas will feature disproportionately on geographical maps, even though they are sparsely populated and have few electoral votes. Montana’s area is 15 times that of Vermont’s, but they have the same number of electoral votes. The effect is a political bias: Since large rural areas tend to vote Republican, the map shows a far greater proportion of red than the election results warrant.
The cartogram alternative
A more truthful representation of the data could be achieved using a cartogram, a type of map that adjusts the area of states for their number of electoral votes.
Making a cartogram inevitably means distorting either the states’ shape or their position relative to each other, so cartograms trade familiarity for precision.
At one extreme, contiguous cartograms preserve states’ borders, but distort their shapes strongly. Mark Newman, of the University of Michigan, produced cartograms of this sort for the 2012 election.
More common in news election reports is another cartogram type that represents each state as a circle or square, resized so the shapes’ area is proportional to the state’s clout in the electoral college.
The one shown herewas published by the Financial Times in 2012, and similar forms have been used this year by National Public Radio among others.
The tile grid cartogram shows each state as a number of shapes equal to their votes in the electoral college. This allows the states to approximate their shape and borders, and makes it easier to compare states’ sizes.
The FT used a tiled grid cartogram as early as 1972, but the format’s popularity has grown only recently. It is being used this year, with a variety of tile shapes by FiveThirtyEight, the Wall Street Journal and the Washington Post.
Having tested these options and debated at length, we arrived at a compromise map for our live results graphic that attempts to take the best from the other methods.
The white underlying geographic map places states in their familiar size, shape and location, allowing them to be identified quickly. Using a cluster of dots rather than a solid fill to represent the outcome ensures that the amount of red and blue on the map accurately reflects states’ weight in the election outcome, rather than the (irrelevant) surface area.
Like the tiled grid cartogram, the number of electoral votes in each state is easy to compare visually without counting or interpreting numbers printed on the map. Because each electoral vote is a discrete mark, it is possible to accurately represent the split electoral votes that are possible in Maine and Nebraska, or the possibility of a faithless elector.
Tom Pearson, the FT’s interactive developer, created a compromise map, which debuted in our electoral college explainer video and our election night guide. We later discovered it was similar to the solution devised for the 2013 Australian election by Gabriel Dance and Nick Evershed of the Guardian.
But even this representation has limitations. The densely populated north-east is covered in dots, making the region difficult to label. This also means the map works best on large screens — on small mobile screens, we have had to revert to the choropleth.
Useful for what?
So does the traditional map fall short? It depends entirely on which question you want to answer for your reader.
If the question is to discover the result in Utah, a conventional map is not a bad choice. For audiences familiar with US geography, the map is helpful for quickly identifying how a particular state voted. Since most presidential elections hinge on the outcome in a handful of swing states, this is an important consideration.
Cartograms may sacrifice ease of identifying states to better answer how important a victory in, say, Florida is, relative to neighbouring Georgia, or to get a more accurate sense of the outcome reflected in the shading of the adjusted surface areas.
We think our third way captures the best of both approaches.
The most important question for any election, though, is who has won. We have previously warned against overusing maps, and this is a case where no map or cartogram can beat the precision of the humble stacked bar chart — which is why that is what will appear on our homepage and above our results map.
A previous version of one of the maps included this article incorrectly indicated that Mitt Romney had won Florida in 2012. In fact, Barack Obama won the state.
Graphics by Chris Campbell and Tom Pearson; video by Tom Hannen, Barney Jopson and Russell Birkett
Get alerts on Data visualisation when a new story is published