Thanks for taking the time to read my thoughts about Visual Business Intelligence. This blog provides me (and others on occasion) with a venue for ideas and opinions that are either too urgent to wait for a full-blown article or too limited in length, scope, or development to require the larger venue. For a selection of articles, white papers, and books, please visit my library.

 

“Sources of Power” in Data Visualization and Decision Making

April 19th, 2007

I have sometimes been amused during my attendance at high-level business meetings in American industry, amused at the discrepancy between the way we are told that important decisions get made and the truth. (Donald A. Norman, Things That Make Us Smart: Defending Human Attributes in the Age of the Machine, 1993, Basic Books, New York)

The logical-rational decision making models that we are taught in college are worthwhile and necessary, but they are rarely used in the course of everyday business. Although there are times when they ought to be used but aren’t, it is appropriate that they are used seldom. One reason for this is that these processes require a great deal of time — something we rarely have. The other reason is that if you have expertise in a domain, you are able to make decisions that are usually just as good, but require little time.

Gary Klein, Chief Scientist at Klein Associates, Inc., has written an informative book about decision making that approaches the topic quite differently from most other books and articles. In the introductory chapter of Sources of Power: How People Make Decisions (MIT Press, 1999) he sets the stage for his treatment of the topic as follows:

During the past twenty-five years, the field of decision making has concentrated on showing the limitations of decision makers — that is, that they are not very rational or competent. Books have been written documenting human limitations and suggesting remedies: training methods to help us think clearly, decision support systems to monitor and guide us, and expert systems that enable computers to make the decisions and avoid altogether the fallible humans.

This book was written to balance the others and takes a different perspective. Here I document human strengths and capabilities that typically have been downplayed or even ignored.

Klein has studied decision making for years and has filled the book with research findings of his own and others, along with story after story of decision making in action.

Despite the fact that we at times and for good reason use “deductive logical thinking, analysis of probabilities, and statistical methods” to inform decisions, Klein reports that in natural settings decisions are rarely analytical, but are usually informed by intuition, mental simulation, metaphor, and storytelling.

The power of intuition enables us to size up a situation quickly. The power of mental simulation lets us imagine how a course of action might be carried out. The power of metaphor lets us draw on our experience by suggesting parallels between the current situation and something else we come across. The power of storytelling helps us consolidate our experiences to make them available in the future, either to ourselves or others.

If you are new to a field, having not yet developed expertise in the domain, decision making must be informed by a relatively slow process of information gathering and evaluation. If you are an expert, however, this research-laden and brain-taxing process is necessary much less often. According to Klein, experts usually make decisions based on what he calls the “Recognition-Primed Decision Model” (RPD). It combines two processes: “the way decision makers size up the situation to recognize which course of action makes sense, and the way they evaluate that course of action by imagining it”. Experts can often look at a situation and quickly recognize it as familiar, knowing intuitively what’s going on and how to respond. When multiple courses of action seem possible, they can take the first from their immediately prioritized list and evaluate its merits by running a quick mental simulation. If problems are discovered during the mental simulation, they proceed to the next possible course of action, until they find one that works, and then without further delay, take action. Are they always right? No, but if they’re experts in the field, they’re right most of the time.

How does this relate to data visualization? People who analyze data, if they are experts in the domain, usually know what’s important and make sense of it based on pattern recognition. They’ve seen it before, or something similar. Nothing presents meaningful patterns that reside in data better than properly chosen and well designed visual representations. More than any other tool, data visualization software can support meaningful pattern recognition for a broad range of people and can enable them to clearly present what they’ve found to others. Good visual analysis software pares information down to its essence in the form of a picture, removing the noise to enable clear focus on the signal. It translates abstract patterns of meaning in the data into images that can be easily perceived by our eyes and discerned by our brains, thereby serving as an external tool of cognition.

Good decisions must be based on clear presentations of data that allow experts to bypass time-consuming and often unnecessary mental gymnastics and software mechanics so they can spend their precious resources assessing the situation and responding while there’s still time. If you don’t have good visual analysis and presentation tools to support this process, you’re wasting valuable time and working partially blind.

Take care,

Signature

Visualizing 360 Data Points

April 9th, 2007

In addition to this blog, Perceptual Edge provides a Discussion Forum for people to exchange and discuss ideas regarding data visualization. It tends to be a place where specialists in visual data analysis and presentation congregate, but from time to time the ideas that are shared cry out for a broader audience. A few days ago Jorge Camoes provided a dataset with 360 values — expenditures in 12 categories for 29 European Union countries, plus an average for all of these countries (30 rows by 12 columns of values) — and seeded a discussion by presenting several ways to represent this information in a single display. The discussion that has ensued presents a rich exploration of data visualization, not only in terms of the solutions that have been proposed, but also in its consideration of the questions one must ask to determine the best way to visualize quantitative data. If this sounds interesting, please take a look and perhaps join in the discussion.

Take care,

Signature

Things That Make Us Smart

March 30th, 2007

Donald Norman is well known in the field of design. For many years he has been helping designers become aware of the qualities that must exist in products for people to use them effectively. As a cognitive psychologist, Norman bases his recommendations on an understanding of the human mind—both its strengths and its limitations. Well designed products (including software) take advantage of our strengths and help us overcome or work around our weaknesses. They don’t complicate tasks by forcing us to do things that we aren’t good at. They help us do things without drawing undue attention to themselves. We who are involved in the field of business intelligence, whether we know it or not, are designers. The reports and dashboards that we create are products. Just like any product, they must be designed in a way that can be perceived and understood by the human mind. Every business intelligence professional can benefit from Norman’s work.

Norman is probably best known for his book The Design of Everyday Things, which is superb. When I first read it several years ago, my neck became sore from the constant nodding that occurred as I recognized one point after another that made so much sense I couldn’t help but nod and smile. Although I’ve read several of his other books and papers since, for some unknown reason I managed until recently to miss his 1993 contribution to design entitled Things That Make Us Smart: Defending Human Attributes in the Age of the Machine (Basic Books). I would like to encourage you to add this book to your library. Allow me to entice you with a few quotes.

Society has unwittingly fallen into a machine-centered orientation to life, one that emphasizes the needs of technology over those of people, thereby forcing people into a supporting role, one for which we are most unsuited. Worse, the machine-centered viewpoint compares people to machines and finds us wanting, incapable of precise, repetitive, accurate actions. Although this is a natural comparison, and one that pervades society, it is also a most inappropriate view of people. It emphasizes tasks and activities that we should not be performing and ignore our primary skills and attributes—activities that are done poorly, if at all, by machines. (p. xi)

The good news is that technology can make us smart. In fact, it already has. The human mind is limited in capability. There is only so much we can remember, only so much we can learn. But among our abilities is that of devising artificial devices—artifacts—that expand our capabilities. We invent things that make us smart. Through technology, we can think better and more clearly. We have access to accurate information. We can work effectively with others, whether together in the same place or separated in space or time. Three cheers for the invention of writing, reading, art, and music. Three cheers for the development of logic, the inventory of encyclopedias and textbooks. Three cheers for science and engineering. Maybe. The bad news is that technology can make us stupid. (p. 3)

My goal is to develop a human-centered view of the technologies of cognition. My theme is not antitechnological, it is prohuman. (p. 12)

The power of the unaided mind is highly overrated. Without external aids, memory, thought, and reasoning are all constrained. But human intelligence is highly flexible and adaptive, superb at inventing procedures and objects that overcome its own limits. The real powers come from devising external aids that enhance cognitive abilities. (p. 43)

A good representation captures the essential elements of the event, deliberately leaving out the rest…A representation is never the same as the thing being represented, else there would be no advantage to using one…Herein lies both the power and the weakness of representations: Get the relevant aspects right, and the representation provides substantive power to enhance people’s ability to reason and think; get them wrong, and the representation is misleading, causing people to ignore critical aspects of the event or perhaps form misguided conclusions…Representations are important because they allow us to work with events and things absent in space and time, or for that matter, events and things that never existed—imaginary objects and concepts. (pp. 49 and 50)

My concern is with the violation of psychological principles, whereby graphs are used in inappropriate ways, sometimes deliberately to confuse but more often out of sheer ignorance. My frequent complaints to friends, colleagues, students, and newspapers are commonly met with the excuse “My computer program did that for me automatically; I had no choice.” Poor reason: Ignorance of the law is not a valid excuse, whether it be a governmental law of a psychological one. (p. 94)

The things we are good at are the things natural to humankind. The things we are bad at are unnatural. And guess what? We can build machines that perform flawlessly many of the things we are bad at. As for the things we are good at, it is very difficult, today usually impossible, to build machines that can do them. “Why that’s wonderful,” you should be saying. “What a marvelous match to our abilities! Between us and our machines, we could accomplish anything, for the one complements the other. People are good at the creative side and at interpreting ambiguous situations. Machines are good at precise and reliable operation.”

Hah! That isn’t what has happened. Instead, technology has decided that machines have certain needs and that humans are required to fulfill them. The things we are good at, those natural abilities, are hardly noticed. Machines need precise, accurate control and information. No matter that this is what people are bad at providing, if this is what machines need, this is what people must provide. We tailor our jobs to meet the needs of machines. (p. 222)

It will take extra effort do design systems that complement human processing needs. It will not always be easy, but it can be done. If people insisted, it would be done. But people don’t insist: Somehow, we have learned to accept the machine-dominated world. If a system is to accommodate human needs, it has to be designed by people who are sensitive to and understand human needs. I would have hoped such a statement was an unnecessary truism. Alas, it is not. (p. 227)

Convinced? Do yourself a favor and read this book.

Take care,

Signature

Sparkline Tickers—What would Tufte think?

March 16th, 2007

I was recently informed by the folks at Bissantz and Company, makers of the product SparkMaker, that they have added the ability to display sparklines as tickers in the new version of the product. They asked for my opinion of this new feature. While watching their examples crawl across the screen, I couldn’t help but think that Edward Tufte, the inventor of sparklines, might not be pleased by this expression of his idea. I certainly cannot speak for him, nor do I wish, in expressing my own opinion, to suppress any views to the contrary that any of you reading this blog might wish to express.

Sparklines can work wonderfully in various contexts, including dashboards, to communicate rich trend information very simply and in a small amount of space. Think of them as an enhanced, much more informative substitute for the trend arrows that often appear on dashboards. Tickers are information displays that usually occupy a strip of horizontal space on a screen, just tall enough to show a single line of text. Information in tickers is constantly changing as it crawls across the screen from right to left, eventually disappearing at the left edge of the ticker, but constantly being replaced by fresh information that crawls in from the right. Here’s an example of a sparkline ticker display that appears on Bissantz’s website:

Sample Sparkline Ticker #1

Here’s another example, truer in design to Tufte’s original, which crawled across the bottom of the screen as I read about SparkMaker Tickers:

Sample Sparkline Ticker #2

What’s missing from this static display that you’re now viewing, of course, is the movement of the sparklines across the screen. They never stand still.

Movement in our field of vision is something that grabs our attention powerfully. It is difficult to ignore objects that move across the screen. There are occasions when the use of movement to demand people’s attention might be appropriate, including on dashboards, but the only examples that come to mind involve urgent information, which viewers must not fail to notice.

Here’s the rationale that Bissantz gives for sparkline tickers, which I’ve extracted from their email to me and their website (emphasis is mine):

Tickers show numerical data in a most compact way. In an endless loop an arbitrary number of values is passed before the eye of the onlooker, without any need for user interaction. The marquee consumes little space and leaves plenty of room for other or additional displays. The presentation within the marquee is clear enough to allow for perfect readability. Yet it lets the user stay with whatever she does.

SparkTicker’s information density releaves [sic] the user: He is not forced into heavy parameterization, page turning or query execution. Instead, he may remain focused on understanding and interpreting his data.

We…believe that dynamic displays of information can enhance data visualization in a lot of environments. For this purpose we provide a ticker option: SparkMaker users who want to go beyond static displays of sparklines can use the built-in ticker feature. It exports selected sparklines to an HTML ticker which can be easily integrated into any web publication. We are confident that such dynamic presentations can help to raise awareness for data and data visualization in surroundings that are typically reporting-averse. Since information in a sparklined ticker is distributed in a non-intrusive way and uses only a small portion of a screen’s real estate, people grasp information as a side-effect while e.g. browsing in the Intranet of their company.

The question we must ask about this and any form of information display is, “Does it work?” Are there any circumstances in which a ticker display of sparklines would communicate information more effectively than all other means of display? Although I am open to the possibility, I honestly cannot think of any.

You certainly wouldn’t want to user sparkline tickers routinely on a dashboard. The constant motion would be too distracting. Also, if the information contained in the ticker is not important enough to display on the dashboard, all at the same time, why not put it on a separate screen and make it easy to access when needed. A display of all the sparklines together would allow comparisons to be made that are not supported by a ticker display.

Let’s consider some of the specific claims that Bissantz is making about sparkline tickers. Is it true that “dynamic displays of information can enhance data visualization” and that sparkline tickers distribute data in a “non-intrusive way”? As I stated earlier, movement might be useful at times to grab a viewer’s attention to some new information that is urgent. There are also times when motion is necessary to provide a meaningful picture of information that itself involves change or motion. Hans Rosling and his colleagues at gapminder.org use motion to show change through time in a way that actually makes sense and brings information alive for people. The movement of data in tickers, however, doesn’t fall into this category. And please don’t confuse dynamic displays with dynamic data. The information that is contained in these sparkline tickers comes from static Excel spreadsheets. It is not dynamic.

Do sparkline tickers provide an extraordinary degree of “information density”? They appear to do just the opposite. You can see very little at a time and what you see reveals itself slowly as each sparkline plods onto the screen. When sparklines are genuinely used to create dense information displays, such as a screen or page full of them, that’s when they shine.

Do sparkline tickers allow someone “remain focused on understanding and interpreting his data”? Wouldn’t a display that stands still do this better?

Will they “help to raise awareness for data and data visualization in surroundings that are typically reporting-averse”? I’m not convinced that this is the kind of exposure that will present data visualization favorably and inform people of its true benefits.

Do sparkline tickers enable people to “grasp information as a side-effect while…browsing…the Intranet of their company”? Nothing that I’ve seen in the research about information perception and comprehension suggests that this would happen. This claim is somewhat akin to the old mistaken notions about subliminal advertising, which supposedly sent consumers running to the store by air brushing words such as “sex” and “buy me” onto ice cubes in a form so subtle that only the unconscious mind could detect them. I must admit, it would be nice if we could grasp information as a side effect based on some peripheral form of display while attending to other things, but I doubt this is possible, and am fairly certain that if it is, sparkline tickers are not the ticket.

If you would like to add your opinions and insights to this discussion, please do so by responding to this blog. I am always open to evidence that I haven’t considered, even if it proves me wrong.

Be as provocative as you wish, but back your opinions with substance

March 12th, 2007

Blogs, by their very nature, tend to be provocative, especially those that are written by industry leaders (sometimes called “thought leaders”). Opinions are often expressed with passion. This is certainly true of my blog. In fact, I tend to communicate in this way in all venues (blog posts, articles, books, presentations, classes, and even conversation over lunch). This seems appropriate. It is not appropriate, however, whether or not you’re communicating provocatively, to state opinions without substance. In other words, be as passionate as you please, but say something meaningful—something you can back up with solid reason and validate with evidence. I’m committed to this form of communication. If you ever catch me expressing opinions that are not well reasoned or that lack evidence, you should challenge me. And if you disagree with something that I say, by all means take me on, but do so thoughtfully. I don’t respond to emotional attacks that lack substance. State what you object to, why you object to it, and then back it up with reason and evidence. This is the only type of discourse that will get us anywhere.

People are sometimes offended by my opinions. More often than not, the people who take offense work for a company whose products or services I have criticized, or are loyal to them for other reasons. I know that criticism of something that you care about feels like an attack and that it is our natural response to lash out emotionally. Just as I put myself out there every day by stating opinions or producing work that are available for public critique, however, any vendor who produces software or services is similarly exposed. Public discourse and critique (both positive and negative) is a powerful means to improve the quality of what we do. I critique the business intelligence industry, not as an outsider, but as a committed insider who strongly believes in what we’re doing—so much that I’ve dedicated my professional life to improving it.

When people react to things that I say, I try to respond with substance, whether or not their reaction contained substance, as long as they made an honest effort to explain themselves. There is a threshold, however, below which I won’t respond at all. Anyone who thoughtlessly and cowardly posts a quick anonymous jab doesn’t deserve a response. I resist the juvenile temptation of schoolyard taunts. (By the way, you won’t find any examples of this type of comment on my blog, because on those rare occasions when they appear, I routinely delete them.) Vendors that try to suppress me by using political pressure to censor my work find that such attempts have no effect at all. (I currently publish my blog and most of my articles through my own organization, rather than through a publication that is financed by money that comes from BI vendors, to remove anyone’s financial interests from having control over my work.)

Just as our government leaders often get bogged down in unproductive bickering, we in the business intelligence industry (or any industry, for that matter) are prone to the same. Nothing good comes of this. We are better than this—or should be. Let’s make good use of our powerful minds to think critically. Only when we do this will we build a business intelligence industry that deserves its name.