"It's dreadful, right?"

People periodically tag me in social media or send me emails about potentially problematic data visualizations they encounter in the wild. The accompanying commentary varies in specifics, but often shares similar underlying sentiment:

“@storywithdata would hate this!”
“This is awful! What were they thinking?”
“Whoever made this needs to read @storytellingwithdata books!”

I ignore most of these. I try not to jump on any bandwagon to bash other people’s work, in particular in cases where the context of purpose, constraints, and other factors are unknown or have been ignored.

That said, there are some occasions where I can’t not weigh in. Today, I share with you such an instance. I returned to my inbox after being out of the office the past two weeks and among my emails encountered one where the subject line read “Please agree how bad this is…” and included the following image.  

It’s an interactive graphic from The New York Times, published as part of an article entitled “Who’s Running for President in 2024?” (updated April 25, 2023). I received it from my friend JD Schramm (with whom I shared a draft of this post and who volunteered to be mentioned by name). He wrote, “I saw this today in The New York Times. It’s dreadful right?” JD went on to pose some questions about it (“What does the angle mean? What are all the dots? Where is the story here? Am I missing something?”) and closed with the appeal, “I’ve spent ten minutes here and I just don’t get it. Any ideas to help me?”

I replied:

“It might surprise you to hear that I don't hate it. The angle seems to be determined by a combo of campaign start date (x-axis) and length of campaign (y-axis). The circles represent when the given candidate withdrew. I agree that it takes a little time/patience to understand how to read it, but efficiency isn't always the sole goal when visualizing data.”

I’d like to expand upon that here and share three reasons why I like this graph.

The novel view shows something more nuanced than a traditional graph. When discussing graphs used frequently to communicate in our workshops, we often field questions about less common varieties. From my perspective, the best use case for a novel view of data in a business setting is when that less familiar view allows people to see something important that they otherwise would not (or that would be difficult). Could you communicate everything the NYT example does in a basic bar chart? Not likely. 

(Related: I emphasize “in a business setting” because there are a host of scenarios outside of that when novel for the sake of novel is perfectly fine. Examples: data art that is designed for beauty rather than to inform, or data viz designed for public forums to wow by illustrating someone’s creativity or prowess in their chosen tool.)

It expands the audience’s data literacy. One challenge using a graph that is less familiar is that you first have to teach your audience how to read the graph. This can be a hurdle, because you need to both keep their attention long enough and have sufficient patience on their part in order to help them understand things—and all of this likely has to happen before any insight can be gained from the data. On the benefits side, when you teach your audience something new in the context of a graph, you can increase their understanding of the various ways in which data can be encoded (both in the given instance, and extending beyond). 

This won’t be appropriate in all settings. But it will be entirely apt in some. In the NYT example, the graph was not meant to communicate on its own—there’s an entire article that goes with it! One might argue that this is the perfect setting in which to try something new, particularly when paired with the other reasons detailed here.

The interactivity invites people to investigate. This graph demands that you engage with it, beginning with the symbols and annotations under the headline that tell you how to interpret the lines. If your curiosity is piqued (if it’s not, as mentioned, you have the article to read), you’ll study those, then jump down to the data to try to understand it. For me, this took some bouncing back and forth between the legend and the graph to decipher what I was looking at and pick out some specific instances in the data to confirm to myself I was interpreting the specifics correctly. 

You can’t passively observe this graph—to get anything out of it, you have to take an active role. In doing so, you’re more committed to understanding what you’re looking at than you might otherwise be. That sounds like a win on the part of the graph designer. Additionally, there’s a further level of interactivity when you actually hover over lines in the non-static version and more details are revealed. You can play with this graph. Might you be more likely to remember it and talk about it with others as a result of this? I am—as evidenced by this post!

I encourage you to consider when and how you might show data in a different way to achieve one or more of the objectives I’ve outlined above. Also, the next time your instincts tell you a graph is terrible: take a step back, give the creator some grace, and reflect on what context might reveal that the visual is actually much better than it seemed at first glance.