Artistically impressive doesn't equal informative

As Stephen Few and others have written about on various occasions, representing data in a visual way can serve a variety of purposes, one of which is to create art. As they also point out, though, it’s important to keep in mind that “data art” is a fundamentally different animal than the visualizations that most of us create as part of our work. One is designed to inspire an audience, or challenge them, or have some other impact on them that good art can have. The other is designed to communicate quantitative information in the most straightforward and informative way possible. Yes, it is possible for data art to also communicate quantitative information in a straightforward and informative way, but it often doesn’t because, ultimately, that’s not its main purpose.

This distinction came to mind when I came across the graphic below, which was recently published on the National Geographic website:

I discovered it because several people posted it on the few social media channels that I actually look at. One person said it was “beautiful and useful,” another, “gorgeous and informative.”

Beautiful? Gorgeous? I think so, too.

Useful? Informative? Not so much.

If the purpose of this visualization is to inspire us to think broadly about how the U.S. really is—and has always been—one big melting pot, then I think that it accomplishes that goal magnificently, and I suspect (hope?) that that’s the goal that the designers had in mind. If, however, its purpose is to enable us to notice patterns, make comparisons, and spot useful insights within this data, i.e., to be informative, I’d consider it to be a failure. For example, we can see that immigration from Asia increased over time, but we can’t really say much more than that about it. Has it been a steady increase? How many people immigrated from Asia in the 1990s? Was it more or less than the 1980s? How did Asia’s increase compare with that of Latin America? Even very basic questions such as how the total number of immigrants from all regions has changed over time are nearly impossible to answer with any kind of accuracy.

While we might be able to answer some of these questions eventually, it would require a lot of time and cognitive energy and our answers probably wouldn’t be very accurate. There are a variety of perceptual reasons why our visual systems have difficulty seeing patterns and making accurate comparisons based on a design such as this, but you don’t need to know anything about visual perception in order to see for yourself that it’s hard to do anything with it other than appreciate its beauty and make vague observations, such as “There’s been an increase in the number of people immigrating from Asia in recent decades.”

These limitations come into sharper focus if we see the same data as a boring old line chart (I dug up a data set that looks to be the same, or at least similar, to the one that was used by the National Geographic designers):

 (Click image to enlarge.)

I doubt many people would describe this graph as artistically impressive but, when I created it, a slew of meaningful insights jumped out at me almost immediately that I hadn’t noticed in the original graph at all:

  • Immigrants from all regions plummeted in the 1930s and 1940s, perhaps because of isolationist U.S. policies in effect at the time.
  • Until about the 1910s, virtually all immigration came from Europe.
  • The post-war spike from Europe in the 1950s is a lot smaller that I would have expected.
  • Immigration from Canada spiked following each major U.S. war, with the exception of the Vietnam War.
  • Immigration from Latin America and Asia started to climb at around the same time (in the 1950s). I could go on…

How many of these insights did you notice in the original graph? Even once you know that these insights exist, are you able to spot them easily?

As I said, I’m willing to give the benefit of the doubt to the designers at National Geographic and assume that their only goal was to inspire us to think generally about the melting pot nature of the U.S. population. If that’s the case, then these criticisms are moot. Indeed, I only felt it necessary to write this post because I saw people characterizing this graph as not just “beautiful,” but also “useful” and “informative.”

As I also mentioned, I’m not suggesting that it’s impossible for a graph to be both artistically impressive and informative. A graph can be both, although these tend to be rare because they require a combination of skills that few people possess (and I’m not one of those people since I don’t have a visual arts background). I am saying, though, that we shouldn’t make the mistake of assuming that, because a graph is artistically impressive, it must also be informative. That’s often not the case.