In response to my recent blog post about Tableau 8, largely a critique of packed bubble charts and word clouds, Chad Skelton of The Vancouver Sun wrote a rejoinder. In it he essentially endorsed my critique, but also argued that it is sometimes appropriate to use packed bubbles to present data to the general public, such as in a news publication, because “bar charts are kind of boring.”
Skelton’s assertion suggests that there is a hierarchy of interest among graphs, perhaps based on shapes and colors that are used to encode data. Can we place value-encoding objects such as rectangles (as in bars), lines, individual data points (such as dots), and circles (as in bubbles) on a continuum from boring at the low end of interest to eye-popping at the high end? Back in the 1980’s, William Cleveland and Robert McGill proposed a hierarchy of graphical methods based on empirical research, but theirs was a hierarchy of perceptibility—our ability to perceive the values represented by graphical objects easily and accurately. The utility of their hierarchy was obvious, because our ability to understand the information contained in a graph is directly tied to our ability to clearly and accurately perceive the value-encoding attributes (positions, lengths, areas, angles, slopes, color intensities, etc.). So, back to Skelton’s suggestion, is there a hierarchy of interest among graphical value-encoding methods, and, if so, does it trump perceptibility?
What is the opposite of boring that Skelton advocates? He provides a clue:
A lot of people who create data visualizations—whether reporters, non-profits or governments—are fighting tooth and nail to get people to pay attention to the data they’re presenting in an online world crowded with endless distractions. And when you’re trying to make someone take notice—especially if the subject is census data or transit figures—a little eye candy goes a long way.
Data visualizations aren’t just a way to present data. They’re often also the flashing billboard you need to get people to pay attention to the data in the first place.
Apparently, this quality of visual interest has nothing to do with the information that’s contained in a chart. Instead, in this case interest is a measure of someone’s willingness to look at a chart. The more eye-catching a chart is, the more interesting it is. It is meaningless to catch someone’s eye, however, if you fail to reveal something worth seeing. I have no problem with the fact that packed bubble charts are eye-catching; they trouble me because once they catch your eye they have little to say for themselves. A packed bubble chart is like a child that keeps screaming, “Look at me, look at me,” but just stands there with a silly grin on his face once you do. Unless you dearly love that child, the experience is just plain annoying.
The argument that a chart must exhibit eye-candy to catch the reader’s attention even when that is accomplished by displaying data in ineffective ways suffers from two fundamental problems:
- In a world of noise, screaming louder and louder is not an effective means of cutting through the noise and being heard. Screaming louder just creates more noise.
- It assumes that you cannot draw someone’s attention to a data display without using an inferior chart, one that is visually eye-catching but information-impoverished.
Regarding the first error, when people get tired of looking at lots of pretty-colored circles randomly arranged on a screen, what will we be forced to do next—make the bubbles constantly move around and spin? Regarding the second error, a graph that gets attention by displaying data in a manner that ineffectively informs is an unnecessary failure of design. Graphs can be designed to catch the eye and inform without compromise. Doing this, however, requires skill.
To make his case that packed bubble charts such as those introduced in Tableau 8 are useful to reporters such as him, Skelton shared the following example of a packed bubble chart that he published last year in The Vancouver Sun:
This chart addresses a potentially interesting topic to Vancouver’s residents, but it reveals very little. Only a few of the agency names are recognizable and only six of the bubbles include values. Had bars been used, the values wouldn’t need to be included, but there’s no way to decode values from the sizes of these bubbles. The size legend (0 through 10) on the bottom left cannot be used to interpret the values that the bubbles represent because it is one dimensional, based on length, which applies to the diameter of the bubbles, but the values are encoded by their areas, not their diameters. This chart is embarrassingly impoverished. If a reporter expressed information this miserably in words his job would be in jeopardy, but we give graphics a pass, treating them like decoration rather than content.
I wrote to Skelton and offered to demonstrate that a bar graph of this content need not be boring if he would kindly send me the information on which his packed bubbles chart was based. I requested all of the information that was available (apparently this information about public servant salaries resides in a publicly available database) about these agencies, salaries, etc., both current and past. Skelton graciously responded by sending the data, but only what he used to produce his own chart: the list of agencies and number of employees in each who earned the top 100 salaries. This information alone is of limited interest. As a Vancouver resident, I would want to know how this year was different from the past and not just the numbers of people, but also the amounts that they were paid. As Edward Tufte wrote long ago, “If the statistics are boring, then you’ve got the wrong numbers.” A chart of any type that contains this information alone will generate relatively little interest. Perhaps Skelton displayed it as packed bubbles rather than a bar graph to camouflage the fact that the information is rather limp. Dressing up a chart in glitter and spangles to generate interest that doesn’t exist in the information itself treats readers disrespectfully. In The Elements of Style, Strunk and White wrote: “No one can write decently who is distrustful of the reader’s intelligence, or whose attitude is patronizing.” I agree.
A chart that displays a rich set of interesting data will not be boring, even if it is a humble bar graph. To illustrate this fact, I began with the data that Skelton provided, and then added to it the kind of information that a journalist could include to engage the reader’s interest. (Please note that I fabricated much of the data in the display below to illustrate my point.)
This enriched set of information, displayed in a way that is easy to read, lends itself to more than a glance. It invites the reader to examine the story in depth.
Skelton is concerned with a significant challenge that journalists currently face:
For news organizations, this constant tension—between excitement and accuracy—is second nature. The most accurate way to portray a council meeting might be a photo of a bunch of bored looking seniors waiting in line to speak. But most newspapers would run the photo of the one, animated speaker waving their finger and shouting.
When did it become the job of journalism to manufacture interest and excitement that doesn’t exist and isn’t warranted? I know that news organizations are struggling to survive. I share Skelton’s concern that it is getting harder and harder to be heard in a world of increasing noise. Not every message is worth hearing, however, and when the message is worthwhile, cutting through the din in ways that rob the message of clarity and accuracy only adds to the noise.
How can reporters get people to read their stories and examine their graphs? I believe they can only do this in a lasting way by consistently providing content that is interesting, accurate, clear, and useful. If they do this often enough, they will become a trusted source. The New York Times is my primary source of news. I subscribe to this publication to show my support for expert journalism and to do what I can to keep it alive. I browse the articles on my iPad and read a few of them daily. How do I pick the articles that I read? I browse the titles, which clue me into the content. I don’t pick articles because they’re eye-catching. I pick those that are mind-catching.