The field of information visualization is still relatively young. The milestones that mark its history can be listed and briefly described with the exhalation of a single deep breath. Only a few information visualization events today can be anticipated with guarantees of significance, including:

• IEEE’s annual InfoVis Conference, which always provides glimpses of a few new and promising innovations
• Edward Tufte’s seminar, if you’ve never been before, which will encourage you to strive for excellence
• A presentation by Hans Rosling, which will inspire you to use visualization to solve the world’s problems
• A new major release of software from Tableau or Spotfire, which will convince you that visualization can reach a broad audience when properly designed and commercially packaged

I hope that my workshops fall into this category as well, with the guarantee that you will walk away having learned many useful, simple, and practical skills.

At least one more event belongs on this list: the publication of a new book by Colin Ware. When Colin told me late last year via email that he had a new book on the way entitled Visual Thinking for Design, I began counting the days. As far as I know, Colin is the world’s leading authority on visual perception in terms of how our knowledge of its mechanics, strengths, and limitations can be applied to information visualization. Much of what I know about visual perception, I learned directly from his work. I’ve been relying on his book Information Visualization: Perception for Design for years, quoting it often in my own work. Based on its exceptional quality—not only its excellent content, but also Colin’s ability to express it clearly—I knew that the release of his new book would constitute a major information visualization event. I finally got my hands on a copy a couple of weeks ago. My high expectations were not disappointed. In fact, they were exceeded.

Visual Thinking for Design is intentionally less comprehensive than Information Visualization: Perception for Design, for it is more focused on visual perception as a process that can be tapped to help us think more effectively. This new book is also accessible to a broader audience of readers. Anyone interested in how visualizations should be designed—both the pictures and interactions with them—to help people discover and understand the meanings that live in data, will find this book comprehensible and a delight to read.

Colin begins by introducing the term “active vision.”

Active vision means that we should think about graphic designs as cognitive tools, enhancing and extending our brains. Although we can, to some extent, form mental images in our heads, we do much better when those images are out in the world, on paper or a computer screen.  (Preface, p. ix)

He goes on to explain why this way of understanding visual perception is important.

The active vision revolution is all about understanding perception as a dynamic process. Scientists used to think that we had rich images of the world in our heads built up from the information coming in through the eyes. Now we know that we only have the illusion of seeing the world in detail. In fact the brain grabs just those fragments that are needed to execute the current mental activity. The brain directs the eyes to move, tunes up part of itself to receive input, and extracts exactly what is needed for our current thinking activity, whether that is reading a map, making a peanut butter and jelly sandwich, or looking at a poster. Our impression of a rich detailed world comes from the fact that we have the capability to extract anything we want at any moment through a movement of the eye that is literally faster than thought. This is automatic and so quick that we are unaware of doing it, giving us the illusion that we see stable detailed reality everywhere. The process of visual thinking is a kind of dance with the environment with some information stored internally and some externally and it is by understanding this dance that we can understand how graphic designs gain their meaning. (Preface, pp. ix and x)

With these words, Colin begins his own elegant dance of explanation, taking the reader step by step through the process of visual perception, pointing out in practical terms along the way how this knowledge can be applied to information visualization and also other thinking processes, such as scribbling on a napkin to explore ideas.

It is always tempting, when reviewing a book this good, to say too much out of sheer excitement. I’m not going to give into this temptation, because I don’t want to spoil any of the fun that you’ll have reading this book from beginning to end with little advance knowledge beyond the fact that it is an extraordinary work. Everyone interested in information visualization should read this book. Not just read it, but mark it up with lines, comments, and even diagrams in the margins, and then keep it close at hand for easy reference and review.

Take care,

Richard H. Thaler and Cass R. Sunstein, both of the University of Chicago, have written a thought-provoking book full of practical suggestions for improving the decision-making process, titled Nudge: Improving Decisions About Health, Wealth, and Happiness. Based on the title, you might assume that this is a self-help book. It is not. It is treatise on the importance of building into products and services “nudges” that help people make decisions that best serve their needs. Products and services require people to make decisions and—intentionally or not—the way the experience is designed pushes (or “nudges”) people in a particular direction. Where products are placed on store shelves and where foods are placed in a school cafeteria line are examples of “choice architecture” that make it easier for people to choose some things over others. Thaler and Sunstein explain the omnipresence of choice architecture and how it works, then go on to argue that, because it exists, whether planned or not, we ought to design choice architectures that make it as easy as possible to make choices that meet people’s needs. They are not saying that any choices should necessarily be removed or that someone’s idea of a good choice should be forced on anyone, but merely that the providers of products and services ought to consider the needs of their customers and make it as easy as possible to make good choices.

One of the prime examples of choice architecture is the existence of defaults in software and other systems. Given the fact that many people will never bother to consider their options, why not make the default choice one that is most likely to work well for them. Many and perhaps most software vendors put little thought into the defaults that they build into their products. A software engineer’s notion of how something should work is insufficient consideration of the matter.

Many of the most visible problems that people face today, such as rising foreclosures and poor health coverage, are in part the results of poor choice architecture. Often they’re the results of choice architectures that were designed to benefit some people at the expense of others, such as disreputable mortgage brokers and health insurers trying to maximize their own interests alone.

Thaler and Sunstein don’t raise awareness of the problem and stop—they go on to propose many practical solutions to important issues, including environmental protection, prescription drug coverage, financial debt, and even marriage rights.

All of us who have opportunities to nudge people in the direction of good choices have a responsibility to do so. To some degree, every one of us is a choice architect—some with influence over many lives and others with influence over a smaller population of family, friends, and co-workers . We can nudge in ways that do not limit freedom of choice. To nudge wisely, we must understand the real needs of people, not just their superficial or momentary preferences, which they often live to regret.

The June/July 2008 issue of Scientific American Mind includes an article entitled “Your Inner Spam Filter” by Andrew W. McCollough and Edward K. Vogel of the University of Oregon. In it they explain that superior abstract reasoning appears to be related to better use of working memory and that the difference between those who reason effectively and those who don’t might be due to differences in the brain’s ability to filter out information that is not relevant to the task. Research has demonstrated for years that working memory is limited to about four chunks of information at a time. You might be surprised by how little we can hold in working memory, the area in our brains where information is temporarily stored while we’re thinking about something. Better use of working memory’s limited capacity results in better abstract reasoning—the kind of reasoning that handles data analysis. So, the analytical process is improved by the brain’s ability to filter out irrelevant information, a function that works a bit like working memory’s spam filter.

Data analysis software, in an effort to support the process, should eliminate all non-essential content, thereby reducing the need for an analyst’s brain to filter it out. By doing so, good software will help to level the playing field between analysts whose internal spam filters vary in quality. The current trend in data presentation and analysis software to display information using flashy visual effects, such as 3-D charts with lighting effects to make them look photo-realistic, rather than displaying the data alone in simple, clear, and meaningful ways, is working against the needs of analysts. As Edward Tufte wisely wrote in back 1983, “Above all else show the data.”

Take care,

Trevor Lott of www.DealerDiagnostics.com, a service that provides performance reports for auto dealerships, wrote to me recently. Here’s what he said:

I’ve been a fan of your work for several years and have always been impressed with your willingness to share, be it innovations, ideas or feedback. It’s only fitting that I had you in the back of my mind when I was testing the Google Charts API a few weeks ago to explore the potential of the web service to create bullet graphs. I’m happy to report that it’s possible but wanted to take it one step further: to promote bullet graphs and help to increase their accessibility. To that end, I wrote up a quick guide to creating bullet graphs using the Google Charts API. Please consider it a small tribute for what you’ve given to me and the data visualization community over the years.

I appreciate Trevor’s kind tribute—especially the fact that he delivered it in the form of a practical solution to a real need. If you’re interested in producing bullet graphs using Google Charts API, take advantage of Trevor’s fine work, which he’s provided as a simple series of clearly explained steps. You’ll find it in Trevor’s blog.

Take care,

I try to maintain a comprehensive library of books and articles about data visualization, so when I recently read that Springer published a new book entitled Handbook of Data Visualization, filled with chapters from respected experts in the field, I set out to get a copy. Ordinarily, I purchase books for my data visualization library—that is, I pay for them, rather than requesting complimentary desk copies, which are often offered to educators. This book’s $319 price tag, however, discouraged my normal practice. I decided this was an exceptional situation, so I tried to take advantage of my faculty position at the University of California, Berkeley. In response, I was told that I could get a copy to examine, but would have to return it if I decided against using it as a textbook for one of my courses. Once my hands touch a book, they don’t let go, so I tried a different approach. I offered to review the book in my blog, which earned me a copy, and today I am fulfilling my promise.

Last week, I spent many hours in airports and on planes (I was a victim of American Airlines’ inspections, which grounded thousands of flights), which gave me time to peruse all 920 pages of this book and read several chapters in detail. This book is indeed filled with data visualization expertise, but it isn’t clear for whom it was written. Contrary to the title, this is not what I would call a handbook. This is a collection of sophisticated academic articles that cover broad territory, but do not provide an overview and introduction to data visualization that the term handbook suggests. Unlike Readings in Information Visualization: Using Vision to Think by Card, Mackinlay, and Shneiderman (1999), Handbook of Data Visualization lacks informative and digestible introductions to the topics that it addresses. Despite my expertise in quantitative data visualization, I couldn’t follow much of the content, for it assumes an advanced level of mathematics and statistics well beyond my own. The fact that I couldn’t understand much of it certainly doesn’t make this a bad book; it simply suggests that non-statisticians should probably avoid it. Which brings me back to my earlier question: for whom was this book written? The obvious answer—for statisticians with expertise in data visualization—becomes less obvious in light of the price. I don’t know many statisticians who are able, or if able, are willing to plunk down $319 for a collection of articles, no matter how good they are.

I couldn’t help but wonder what could have possibly caused Springer to set the price so high. I assumed that its three editors—Chen, Hardle, and Unwin—would receive royalties from Springer for their work, but I doubted that contributors of individual chapters would be paid for their efforts. My curiosity led me to ask a friend who contributed a chapter along with three of his colleagues if he was compensated. He replied that his compensation was a single copy of the book, which he and his three co-writers were obliged to share. He was frustrated that the book’s high price would keep his work from being purchased, except by university libraries. Besides royalties to the writers, the only other significant expense a publisher usually faces is the cost of printing and binding, for they rarely spend much money to promote books. This book is hardbound, which costs more to produce than a paperback, but I estimate based on my own experience that printing and binding should cost less than $10 a copy, even if printed in relatively small quantities.

For $319, one would expect a book about data visualization to feature beautifully rendered color figures throughout, but it exhibits only one-color printing (black and shades of gray), except for a small insert of multi-colored pages in the middle. I believe that every book about data visualization should be printed in color, yet I’ve seen many examples of an author’s fine work that were undermined by a publisher’s decision to save money on production costs by going with cheap paper and a one-color printing process.

Another aspect of this book’s design that I found annoying and completely out of character with the concerns of data visualization was the placement of figures. In the chapter “Good Graphics?” by Antony Unwin, he wisely recommends:

Keeping graphics and text on the same page or on facing pages is valuable for practical reasons. It is inconvenient to have to turn pages back and forth because graphics and the text relating to them are on different pages.

Amen to that. Unfortunately, even the very page on which this statement appears and almost every other page in the chapter ignores this advice. I’m confident that Unwin cannot be faulted for this flaw in design, and that fault lies with the publisher, which took the easy, inexpensive path, despite the inconvenience to readers. I have had to fight hard to control the design of my books and articles, sometimes to the annoyance of publishers, in an effort to avoid problems like this. This shouldn’t be necessary. Publishers should be experts in these matters and respect their customers enough to do what’s required to make books work, even when it takes more time and costs a bit more.

I cannot recommend this book to most of my readers, who usually favor advice that is accessible to non-statisticians and can be more broadly applied. I am confident, however, that this book would be useful to statisticians who already know quite a lot about data visualization, if they could only afford to buy it. The failures of this book rarely stem from its authors, but instead from Springer’s near-sighted and dysfunctional publishing model.

Take care,