Stage 3 Talent

The Makings of a Bad Visualization: Election Edition

Data literacy is a key topic for us here at Stage 3 Talent. We explain how to read data presentations in our Data Literacy course. However, sometimes, it’s tricky. Especially, when you can manipulate your audience or capture their attention at just the right or wrong time. I found this image while talking with some friends about the current US elections:

Early Voting in Battleground States - reported by MSNBC, focusing on ages 18-29 in Florida, North Carolina, and Michigan.  Florida had 44,107 early voters ages 18-29 in 2016 and 257,720 early votes ages 18-29 in 2020 as of 10/23/2020.  North Carolina had 25,150 early votes ages 18-29 in 2016 and 204,986 early voters ages 18-29 as of 10/23/2020.  Michigan had 7,572 early votes ages 18-29 in 2016 and 145,201 early votes ages 18-29 as of 10/23/2020.

Let’s talk about some of the problems with this visualization.

Column Order

When presenting timely data, it makes the most sense to present it chronologically from left-to-right, past-to-present. However, this visualization breaks that and displays it from present to past. By presenting the columns in a non-natural order, it can lead to incorrect assumptions.

Did the 2016 election really have that great of a turnout and 2020 has a low turnout? Wait… look at those columns again. The 2020 election has a significantly larger turnout compared to the 2016 election.

(Note: As was pointed out by a reader, this tweet came from someone who may have right-to-left as a setting for their environment or as part of the news site’s culture. Knowing the intended audience may have helped us notice this earlier. This is why it is important to question our sources.)

Color Choice

The columns for 2016 and 2020 are blue and red, respectively.

The problem with using red and blue for these columns at election time is that the colors already have the meanings of Republican (red) and Democrat (blue) parties. At a quick glance, people may see these colors and jump to conclusions of Republican votes vs Democrat votes.

Realistically, these statistics are not indicating anything party-wise. However, the seasonality of this data makes these color choices harder to process at a first glance.


This specifically applies to early voters – not the complete voter turnout. Many folks who first saw this visualization missed the fact that this is filtered specifically to early voters.

To make things even more complicated, it is filtered further to look at ages 18-29. However, with the size of the title and other data, the “AGES 18-29” blends in. There could be better font-size or font-color differentiation to make it more clear that this filter is there.


The presentation of visualizations is key. One misstep can lead to wrong assumptions, heated feelings, and other confusing experiences. The color scheme is important to consider, especially if those colors are associated with other things in the current context. Timely data order presentation matters due to a natural expectancy of its presentation. When data is filtered, those filters should be clearly stated – in font size, color, positioning. These set the stage for people to properly interpret the visualization.

If you want to learn more about reading data visualizations or working with data to create data visualizations, reach out to us at for more details!