Content & messages

USA elections 2020: a “best of” gallery of DataViz.

No matter how your electoral system works, you can learn a lot from the USA 2020 vote data visualizations. Good DataViz can drive journalism.

In January 2020, Fast Company made the headline “Data visualization has gone mainstream”.

Starting from February 2020, with the pandemic, data visualization – which crosses information design, data science and storytelling – has definitively proved, if needed, how important it is to use visualizations to explore, understand and above all communicate data.

Data visualizations can save lives, I wrote something about that in March 2020. I am still convinced of it, even if the battle against the COVID-19 infodemics and cognitive dissonances is a tough one. You can have the best data designers in the world, the most motivated to make themselves understood.

But remember: most population in the world has still very low data literacy.

Today, in November 2020, the American elections have been the field on which the best newsrooms, not only in the United States, offered not only data updated in real-time but tools to know how to read and interact with those data with an extent that you never experienced before.

For us living in Europe, the US elections are as complex as understanding viruses. But this is our world: complex and full of data. Our duty as communicators is not to pretend that it is simple (´cause it is not), but to break it down and make it intelligible.

The information designers gave their best this time. Also, when necessary, elegantly dismantling some graph formats to reveal their limits.

This post is a roundup of graphics and charts of the 2020 American elections, a “Best Of” gallery with some notes and remarks.

There is no better way to open this gallery with this one, a graph showing how choropleth maps are misleading and a better job to display votes across a country can be done.

Acres do not vote, people do.

This week, a series of GIF maps created in 2019 by Karim Douieb, co-founder of Jetpack.AI, a data science company based in Brussels, went viral for showing different ways in which you might display election data in map form. His designs, first released in 2019 as a response to a map displayed by Donald Trump, visualise election returns in ways that give more texture to where and how Americans are voting. 

The 2019 Trump´s map framed thousands of miles of empty land as voting for Trump instead of representing the few people living in it. The problem with choropleth maps is well known among information designers: these cartographic depictions reduce every nuance of politics down to a dichotomy of blue and red — no texture permitted.

Instead of a choropleth map, Douieb used symbols, with a size representing the actual number of electors in each territory, not just painting the territory in red or blue. Updated with 2020 data and transformed into an animated GIF, the graph became the most popular across the Internet, for a good reason. It needs no explanation. The transition from one way of thought to another is so striking that it’s almost uncanny. 

To know more about this, you can read: “Try to impeach this” by Karim Douieb at Jetpack. and A Belgian man made the most viral election map of 2020 in 2019: the story of this election map, told by Fast Company.

Maps to make sense of the election.

You can find many alternatives to traditional maps all over the Internet. Some best have been published by the New York Times electoral coverage, with different versions, each highlighting a different aspect of the electorate. The week right before the elections, the New York Times also published a detailed piece about information design mistakes when dealing with electoral maps and how to avoid them.

In the 2020 coverage by the New York Times, the traditional State-by-state, winner-takes-it-all map is complemented by additional maps that display: 

  • Where do the electoral votes come from? 
  • Where the candidates lead, county by county and the amount each county’s leading candidate is ahead. 
  • The shifts from 2016: how much did each county vote shift from one party to the other? 

One of the most common alternatives to traditional maps is the Hex map: a form of a cartogram, meaning that the geography is adjusted such that regions of equal importance–e.g. by population or electoral significance–take up the same area on the map. Hex comes from “hexagon”, but the NYT version uses simple squares to get the same result. 


The FT electoral map merged the territory and the symbol formats: the solution was elegant and of immediate understanding.

The Guardian offered readers maps to drill-down so to explore votes at a State and County level.

The closer look: demographics, social data, correlations.

Diving deeper in data is not the easiest thing to do in live, real-time coverage of elections. The one thing to do is to prepare dataset in advance and the template charts. That is what the New York Times did with its by-the-County demographics charts, like this one below, and Bloomberg with some background data about ethnicity, household income, and unemployment rate.

The snapshot: tables still work (if used with care).

Big tables with many numbers are never good. Small tables with few numbers, instead, can provide a good quick overview. When introduced by a short, well-done explainer text, they work at their best. See these tables on the crucial States at stake provided by the Guardian.

Other tables and cards:

Election needles on the New York Times.

In business, we love needles, and we use them often in our dashboards. Maybe too often. But they work: easy-to-read indicators, a mix of quantitative and qualitative analysis. These graphs by the NYT show two things, relevant for the most critical States: 

  • as the vote counts proceeds, a needle shows the estimated margin of Trump vs Biden.  
  • Below the first indicator, a “probability” needle shows who is expected to win in each State and its probability degree.  

Interactive graphs: the “Paths to win” scenarios simulation on the New York Times.

A less sophisticated version of scenario mapping was on the Neue Zürcher Zeitung.

Exit Polls: who voted whom.

Exit polls are per se less and less relevant, not only in the USA. Here, the newsworthy stuff is a real-time demographic analysis of voters. It says something, confirmed later by other analysis: many Latinos love Trump, not only the Cuban-American living in Florida.

The matrix heat map.

Not so common and not so easy-to-read, as it relies on a careful choice of colours. Yet, this is a powerful example used to show the political evolution of each state in the last thirty years, to one or the other party.

Bar charts: boring, but useful.

Bar charts: boring, but still useful. This graph comes from the FT and shows which candidate was leading where and which States were still uncertain.

Tracking the vote count on the New York Times.

Excellent charts: a vote tracker, updated as the vote count proceeds and showing the votes as % and as an absolute difference between the two candidates.

Motion graphics on TV.

If you watched some live coverage by the big American networks, it was like the fireworks on New Year. Visuals at their best. Where I live, in Germany, a special mention goes to WELT TV (Axel Springer). Interactive infographics in a virtual studio were the hosts could immerse themselves and virtually interact, move and touch data.

How to visualize data: Datawrapper and Flourish created template collections for public use.


No matter how your electoral system works, there is plenty we can learn from the 2020 USA election vote coverage on how to help readers and viewers to understand things as they are happening. But it takes time to plan and think about the best format. The rest is simpler. None of the above graphs is new or complex. Few of them require coding. They explain, engage and provide value. In a world full of distorted interpretations of data, good data visualizations matter and drive journalism. You need not be the NYT or the FT to make good visualization work. Tools are there to facilitate your development work. The thought behind the DataViz: that is the investment you need to do in your newsroom. 

But remember: data never speak by themselves. They need text and explanations, especially about the applied method. Check out this piece provided by Bloomberg. As mentioned at the beginning of this post, on the NYT you can read one of the best explainers about how misleading electoral maps can be, and why good information design is a value for democracy. Not all maps are made equal. “Charts can lie”, so Alberto Cairo wrote in 2019: his book is the best suggestion to close this post.

This post is part of the newsletter: “Little Data Stories”. You can subscribe here: