This is what I call a process post—and a long process post at that. It’s a post about how I prepared my “Short Story” talk for the February Tapestry Conference in Annapolis, MD (the video is below). Five of us were each given 15 minutes to speak. According to Nancy Duarte, this would require at least 7.5 hours to prepare my slides and 7.5 hours to practice. Um, yeah, I’m actually a little embarrassed to say that I probably exceeded both of those numbers.
First, for the motivation. I’ve always had a nagging problem with conveying and communicating large numbers. Social Security, the federal budget, the number of people who receive food stamps…these are all really large numbers that are hard for anyone to conceive or relate to. So when Amanda Cox from the New York Times used the term “Goofy Comparisons” in her talk at OpenVisConf last May, I immediately got in touch with her to discuss these types of comparisons (and to ask if I could steal the phrase). Amanda and I converse fairly frequently about these types of comparisons, share both good and bad examples, and how we might each do a better job making these numbers more familiar to our readers.
Now, to develop the story. When I started, I wanted to convey some ideas or thoughts or hypotheses about how to tackle this challenge. I thought I would start with something like, “Here’s a challenge. How do we get people to understand the large numbers we talk about? I don’t know the answer, but here are some thoughts.” My wife pointed out that perhaps it’s not the best strategy to start off by saying you don’t have an answer to the question you’re asking. I’m not sure that’s 100% true, but in this case she was definitely right.
So instead of saying “I don’t know,” I decided that I wanted to start with a story that would illustrate my frustration with making these sorts of comparisons. That led me to think about my first real data visualization, a graph where I thought strategically about what would best convey the information to the reader. It turned out that my first data visualization was a redesign of a dense table in a Congressional Budget Office report on policy options for Social Security. Now, it was one thing just to launch into that example, but it turns out that I could pair the example with a story about how I became interested in data visualization (it turns out I was working on that project at the time I attended an Edward Tufte course; and yes, dear reader, I can hear you groaning). Perfect—a personal story as a hook that dovetails directly into content. The story led to showing dollar amounts for two years of Social Security spending—really, really big numbers—and how we generally show that type of large number by converting them to some other amount, such as a percentage of gross domestic product or as per capita dollar amounts.
I now had my beginning. I then wanted to give some concrete guidelines for people to consider. After much brainstorming and sketching—I kept a notebook with notes, sketches, pictures, and copied passages from books—I narrowed it down to the following:
- Mix Modalities: switch the comparison to something with which people are more familiar.
- Big Enough, but not too big: Make the comparisons big—because these are big numbers—but make them just big enough so that the reader can understand or relate to them.
- Have a Soul: Make these comparisons something people can feel and relate to; something they care about.
Okay, but this post isn’t really about content, it’s about the process (if you want more content, watch the video at the end of this post). So with these three ideas, I now had a five-part talk: Introduction (with a personal story blending into content); Idea #1; Idea #2; Idea #3; Conclusion.
Ah, the conclusion. I actually had an idea for this early on. I wanted to circle the whole talk back to two things: the example of Social Security spending I used in the beginning, and something personal to mirror the story of how I became interested in data visualization in the first place. For the first, I constructed an example of how to apply the three ideas in the middle section to the Social Security numbers I showed in the beginning. Instead of using a bar or line chart, I showed a map that compared how far one could drive 50 years ago and today, originating from the conference location in Annapolis.
For the second, I wanted to return to the large numbers I was talking about, such as government spending, disasters, and numbers about our environment. This allowed me to drill down from numbers that affect our planet to those that affect our country to those that affect where we live to those that affect our families. That was easy to convert to slides: image of Earth; map of the U.S.; picture of town sign (I took a picture of the sign where I live in McLean, VA); and a picture of my family (I used a recent picture of me with my two kids).
As for the rest of the content, I aimed to combine images of different comparisons with images and quotations from various books and source documents. Images of different good and bad comparisons was pretty easy, having kept a decent library and over the past few months, as well as the ongoing conversation with Amanda. I had taken pretty good notes of each of the books and articles I read, so it was then a matter of piecing them together:
- How we are more likely to remember visual stimuli than just text, from John Medina’s fabulous book, Brain Rules.
- How context and familiarity helps us to remember and recall information (so-called “chunking”), from Joshua Foer’s book Moonwalking with Einstein. I pulled in more information and experiments from the source documents including research by Adriaan DeGroot (1965), William Chase and Herbert Simon (1973), and others.
- How we are more likely to identify with a specific person than simply through numbers or statistics from Jonah Lehrer’s book How We Decide (other issues of his credibility aside) and then again from the source documents (specifically, experiments by Small, Lowenstein, and Slovic, 2007).
- Other experiments and articles—for example, that by Chevalier, Vuillemot, and Gali (2013) and Slovic, Zionts, Woods, Goodman, and Jinks (2011)—were great and relevant, but ended up on the cutting-room floor.
- Side note here: I’m a big believer in reading the source references. Sometimes that’s easy—especially when the authors are good writers or the article is something of a meta-analysis—but other times it’s more difficult. I’ve read many of the papers cited in the Vision Chapter of John Medina’s book, but I’ve had trouble finding the exact source for the numbers in this often-quoted passage:
- Text and oral presentations are not just less efficient than pictures for retaining certain types of information’ they are way less efficient. If information is presented orally, people remember about 10 percent, tested 72 hours after exposure. That figure goes up to 65 percent if you add a picture.
- If anyone knows the exact spot, please let me know.
- Another side note, this time about using quotations: The first draft of the slides included the text of the various quotes I wanted to use (from Medina, Foer, and Lehrer; also from President Reagan and Mother Teresa, which didn’t end up making it). I wanted more imagery, however, so I added a picture from each author’s book. It then dawned on me: Why make people read the quote at all, especially when I wanted to say it aloud? I can just say the quote and let the book cover serve as the image. Admittedly, this makes the slides less useful as a standalone product and it might be harder for people to record or tweet the quote, but my goal was to present to an audience I hoped was more interested in the content than in tweeting it or later reading my slides online.
Now that the slides were built, it was time to practice. I probably practiced at home 6 or 7 times—recording each one on my iPhone’s Voicenotes—before trying it in front of my wife. (Of her many, many excellent qualities, she is my go-to when it comes to my writing and presentation preparation.) That 15 minutes was probably the most important part of the entire process: She saw (as did I) the holes, the gaps, and the places where things didn’t hang together. Back to the drawing board—another couple of weeks of sketching, rewriting, moving slides around, and more practice. By the time she saw it again two weeks later, it was more or less ready to go. (I would also usually practice in front of my co-workers and friends, but it was a busy time and people weren’t available. Fortunately, I was able to do a dry run with my friend Ann Emery and some others.)
I won’t lie—there were nights when I went to bed thinking about this talk, different images and slide transitions, different things I wanted to be sure to say, and research to cite. I’d get up, walk to my office and jot a note down in my notebook or on a 3×5 card (if you didn’t already know, I’m a big fan of 3×5 cards), and try to go back to sleep. It was a long process and I learned a lot about different fields, but also a lot about my own presentation preparation and design method. I’m sure most people don’t go through such lengths to prepare their talks—especially ones that are 15 minutes long—but based on the very high quality of talks at Tapestry last year, I felt it important to give the best talk I possibly could. Not only was I pleased with the result, but the entire process gave me more to think about when it comes to conveying large—and sometimes abstract—numbers to an audience.
Here is the video of my Tapestry talk, and be sure to check out the other videos on the conference YouTube page.