Tradeoffs in information graphics

The visual display of quantitative information (to use Edward Tufte’s wonderful term) is a diverse field or set of fields, and its practitioners have different goals. The goals of software designers, applied statisticians, biologists, graphic designers, and journalists (to list just a few of the important creators of data graphics) often overlap—but not completely. One of our aims in writing our article [on Infovis and Statistical Graphics] was to emphasize the diversity of graphical goals, as it seems to us that even experts tend to consider one aspect of a graph and not others.

Our main practical suggestion was that, in the internet age, we should not have to choose between attractive graphs and informational graphs: it should be possible to display both, via interactive displays. But to follow this suggestion, one must first accept that not every beautiful graph is informative, and not every informative graph is beautiful. . . .

Yes, it can sometimes be possible for a graph to be both beautiful and informative, as in Minard’s famous Napoleon-in-Russia map, or more recently the Baby Name Wizard, which we featured in our article. But such synergy is not always possible, and we believe that an approach to data graphics that focuses on celebrating such wonderful examples can mislead people by obscuring the tradeoffs between the goals of visual appeal to outsiders and statistical communication to experts. . . .

The full article (with Antony Unwin) is here. We are responding to discussions by Robert Kosara, Stephen Few, Hadley Wickham, and Paul Murrell.

I’m hoping that, by framing graphics in terms of tradeoffs, we can move the discussion forward. In our earlier discussions of statistical graphics and data visualization, we were slammed by statisticians for being too nice to infovis, and slammed by infovis people for being too mean. Here we’re emphasizing that:

1. You can’t expect to satisfy all goals with a single display, and thus,
2. Multiple graphs of a single page, or on multiple pages, are typically the way to go.

9 thoughts on “Tradeoffs in information graphics

  1. Maybe I’m missing something in the article, But I was really interested in reading the discussions by “Robert Kosara, Stephen Few, Hadley Wickham, and Paul Murrell”. Do you have any links to those discussions?

    Thanks in advance,
    Manoel

  2. I’m very glad that you have taken on the topic of bridging the gap between statistical graphics and info-vis. Many statisticians that I speak with about visualizations are very dismissive of the design aspect info-vis. It is important to recognize that statistical arguments are about communication, and if you don’t hold the attention of your audience, you’ve already lost.

    We (as a profession) need to learn what info-vis does right, so we can help shape the area, rather than be left in its wake, screaming that they need more p-values.

  3. Pingback: My talk on statistical graphics at Mit this Thurs aft « Statistical Modeling, Causal Inference, and Social Science

  4. Andrew and Antony,

    Thanks so much for replying thoughtfully to my critique of your article. I’m writing to respond to two of the points that you made in your reply.

    I’d be willing to bet that you didn’t actually read the study by Hullman, et al titled “Benefitting InfoVis with Visual Difficulties.” I’m confident that, if you had, you would have realized that it is nonsense. I read it carefully and critiqued it in an article, which you can find at http://www.PerceptualEdge.com. What I found was a deeply flawed misapplication of a few psychology studies. Have you ever encountered bad studies in your own field and then cringed when others granted them credibility by citing them? If you have, you’ll understand how disturbing I find it to have you cite this study as evidence of your position.

    The psychology studies that Hullman, et al, cited propose the usefulness of visual difficulties when applied to text in a specific way under specific conditions for a specific purpose. If readers approach text casually, comprehension can be improved when a visual difficulty such as light text or a barely-legible typeface forces them to shift from unreflective System 1 thinking into reflective System 2 thinking (see Daniel Kahneman, “Thinking, Fast and Slow”). This has not been tested in relation to data visualization. A flashy, barely comprehensible infographic would not motivate viewers to proceed to a more informative statistical display, as you propose. People who find impoverished infographics appealing will be satisfied with them alone, rarely moving on to a better graphic. Those who actually care about the data rather than an eye-catching graphic, don’t need visual fluff to attract their interest. They will proceed to the better graphic immediately and would prefer to skip the impoverished infographic altogether.

    By “better graphic,” I am referring to one that is not only statistically accurate, but one that is also clear and aesthetically pleasing in a way that draws viewers into the data. While I appreciate your comment that I am “selling my expertise short” when I argue that statisticians can easily learn to design quantitative graphs in visually pleasing ways, my assertion is based on empirical evidence. Everyone who attends my course in graph design (of whom there have now been thousands, including many statisticians) leave the class understanding the basic principles and practices that make a graph aesthetically pleasing. These principles and practices, when properly taught, are very easy to understand and remember. The only difficulties that people encounter when applying these best practices are the bad formatting defaults of many charting software, which they must learn to work around.

    • Stephen:

      I agree that visual difficulty is not generally a plus, when considering the statistical goal of presenting and understanding data. But, for the goal of grabbing the interest of outsiders and pulling them in to think about the data, I think visual difficulty can be very useful. For a recent example, see here and also Jessica Hullman’s comment.

  5. Pingback: An epithet I can live with « Statistical Modeling, Causal Inference, and Social Science

Comments are closed.