Fix my chart: Presidential approval ratings

Welcome back to Fix My Chart, our blog series about data visualization redesigns. In this edition, we're happy to have Gregor Aisch share his process. Gregor is a Datawrapper co-founder and now Senior Visual Data Journalist at DIE ZEIT, where a colleague asked him for help with a line chart showing approval ratings of U.S. presidents.

On February 11, the polling institute Gallup announced that after almost 90 years, it will stop polling favorability ratings on sitting U.S. presidents. One of my colleagues at DIE ZEIT saw this as an opportunity for a story. They created this line chart that shows approval ratings of U.S. presidents since the 1940s, and asked me for feedback:

So what’s going on here?

Let's take a look at the chart above. There’s a yellow line showing the raw approval ratings over time, surveyed at irregular intervals. The blue line on top shows the average approval rating for each president. In the background, we see blue-ish and red-ish stripes that tell us which political party each president belongs to — Democratic or Republican. Finally, a few annotations give us a little more context, including a label for the blue line.

You would probably agree that this chart leaves space for improvements. Let’s dive into them one by one.

1. Let’s reduce the number of colors

There are too many colors: We have the semi-transparent blue and red background, plus two solid lines in yellow and blue. The blue in the background indicates Democrats, but the blue line in the foreground spans all presidents. The yellow has no particular meaning.

Here’s a little trick that helped me with quite a few charts in the past: whenever you have lines on a colored background, try removing the background and coloring the lines instead!

Now, on to the line showing the rating averages for each presidential term. In the second iteration, this line is purple, which makes a bit more sense as the color between blue and red. But there’s a better way. We show the raw approval rating lines in both red and blue, so let's do the same with the lines for the average rating. This has the added benefit of removing the “steps” aesthetic of the line:

2. Let’s make clear what readers should focus on

Still, the two sets of lines are fighting for attention a bit, so the readers can’t quite decide what to look at. In situations like these, we want to guide readers by making clear what we want them to focus on. So I removed the opacity of the raw approval ratings and increased the stroke width of the average rating lines:

3. Let’s embrace the wimmelbild

Now we have a clear, well-organized chart, but there’s one more step to turn it into a great one. Often, charts work best if they are clean and simple. But there are times when your data is telling many stories, and it makes sense to go “all in”. As in a wimmelbild, a picture full of details, characters, and mini scenarios that encourages viewers to explore and find connections on their own, we embrace the details and include them all in our chart.

First up, we label all the presidents! To fit all their names in, I had to go with a small font size, but that’s okay in a wimmelbild chart! After all, the names are not meant to jump out; they're for readers who want to dive in.

4. Let’s add annotations

Finally, we bring back the annotations that refer to some key historical events: two wars, the Watergate scandal, and the September 11 attacks.

And that's it! Our improved version of the presidential approval ratings chart is ready to be published. It has changed quite a bit, and I believe it allows readers to better see how average approval ratings compare among past U.S. presidents and in the context of history.

We hope you enjoyed this insight into Gregor's design process. Would you like to share your own organization's visualization redesign? Send a note to blog@datawrapper.de! We're looking forward to hearing from you.