Fans of our office dog already know that Bella has just written a book based on the German version of her blog. The title translates to "“Bella consults – 75 rules for better visualization". Next Wednesday, the first copies will be delivered to our office hot off the presses.

Gerald Butterwegge, Simone Appoldt, Nicolas Bissantz and Bella check the final proofs. Pictures below: Carlos Da-Aira explains the color control process. Helmut Hofmann with Bella.

Bella’s book is based on her blog and, therefore, first originated online. The Web, however, is a medium with a very limited resolution. All amateur pictures and scanned images, every drawing and screenshot, all of our own and quoted material worked well in the Internet. In our first attempt to transform the blog into a book, our greatest fears were confirmed: On paper, everything seemed so rough, awkward and lackluster. That meant that we had a lot of work ahead of us. We had to rescan the material, search for the original sources, edit the photos and use all of the advanced print options that our own software supports to give the book the desired effect. We painfully experienced firsthand how large the gap between the print and digital world still is. The explanation is quite simple. Monitors are Pixelland with maybe 1024 x 768 pixels. Paper, however, is Panoramaland which is generally available with over 1,200 different dots per inch. On the UV press at our printer Hofmann in Langenzenn it was even 2540 dpi.

Panoramaland is fascinating. With a good dose of creativity, craftsmanship and competence, we can create an immense resolution with highly impressive quality. Of course, we truly missed the real-time and interactive functions that come naturally in Pixelland. The ink drops in Panoramaland sit in their exact position down to the last micrometer. But if you want to place them somewhere else, that means trouble. For the future, we wish that we could have both – a type of interactive Panoramaland.

In management reports and business magazines, we often encounter visual metaphors which stem from road or traffic signs. When we see these old, familiar signs in their original ‘habitats’, however, we can learn a lot from them. Here are some examples:

Robust perception: Triangles show the direction. Plus and minus signs show changes.
We recently discussed the redesign of the Wall Street Journal. For the past few weeks, the journal has applied a new coding for plus and minus signs: little red and green triangles which show how indexes and currencies have changed compared to the previous day. These triangles appear at the top corner of two different pages and either point up or down.

I recently encountered similar triangles while hiking through the forests of New Zealand. A trail, which left the main path and headed into a dense, tropical jungle, was marked by plastic triangles that hung vertically every few meters. These signs were more than just helpful. They were essential because a person could easily get lost after the first few steps in this labyrinth.

If someone had just nailed plus and minus signs on the trees, finding the right path wouldn’t be any easier. In fact, hikers might get so caught up wondering what their purpose was that they literally might not be able to see the wood for the trees.

The triangles, which were long and steep, sometimes pointed right or left and sometimes up or down. Depending on the perspective of the approaching hiker, the same symbol was cleverly displayed in many different ways so that relatively few triangles were needed to clearly mark the challenging off-road path. The color of the triangles stood out because it was not otherwise found in the forest. That, too, was a major help. Two triangles turned towards each other also signalled the stop points for counting offspring, which was the actual purpose of this trail. They showed the hikers that they had reached their destination; the path didn’t go any further.

Robust colors: Forms. Robust form: Colors.
Permanent right-of-way laws can dramatically improve the general safety before one-lane bridges, tunnels and other tight areas. The earlier the driver knows which side has the right of way, the better. The ability to differentiate colors and forms from increasingly larger distances, however, changes in various ways. In most cases, the human eye can differentiate forms for a longer time – especially when individual limitations in vision also come into play.

This sign is quickly identified from afar – even with red-green color blindness. Whoever is travelling in the same direction as the large arrow has the right of way. The differentiation between red and black makes the message even more noticeable.

When presenting information on a computer screen or printout, we need to overcome a similar situation. We can easily lose finer color nuances in two separate LCD displays, which have very low color certainty. Plus, we will oversee even larger differences in colors, at the very latest, when making black and white printouts. A robust design takes these circumstances into account.

Robust perspectives: Doubling
On a very curvy street in the mountains, a summit can hide a warning sign if it is mounted at a standardized height. One way to solve this problem is to ‘double up’, in other words, mount the same sign at different heights. The driver can see the higher sign at the ‘usual’ height before the summit and the lower one afterwards. Upon approaching it, the seemingly redundant design surely brings a smile to the faces of design-sensitive souls.

We often face similar situations with management information. The same data deserves the readers’ attention from completely different perspectives. A single visualization rarely can provide such robust perspectives. Therefore, it always pays to ask if we shouldn’t display the same data in multiple variations. What may appear redundant at first is probably the safer alternative when navigating through the twists and turns of your data landscapes.

Each of these examples was a sign to me as well. Symbols are designed to reduce insecurities and these do it adequately. We can learn a lot from their construction but we can’t apply much from their imagery without the risk of oversimplifying or even manipulating our data. How a share price changes compared to the previous day is no real guide through the modern investment jungle. And a traffic light signal cannot determine if standing still or moving forward will improve the situation.

Professor Ed Galea, who teaches fire safety and mathematical modeling in London, investigated the risks of flying. He says that life-threatening plane crashes are extremely rare, crashes with fatalities are even rarer, and crashes without survivors are even rarer than that. Where you sit plays a certain role.

Flying is pretty safe. This machine arrived without its roof – but it still arrived. (Source: AP/Wide World Photos, Aloha Flight 243, 1998–04–28)

The following graphic wants to say exactly that. Instead, it just makes business and first-class passengers sick to their stomachs. Of course, the three-color logic exaggerates; the differences between the numbers aren’t as great as the colors suggest. And no one would die as easily as the graphic implies either. According to Professor Galea, the role of the assigned seat depends on the type of the accident. And then again, the distance to the next exit is more important.

Poor design and a vague message: In this chart, the differences in color are much greater than the differences in numbers. And in reality, of course, everything is a little more complicated than in this drawing. (Source: Gil Ahn/seatguru.com, Survival rates on seat levels of all commercial plane accidents in the USA since 1971)

The next illustration attempts a bit more differentiation. It is so confusing, however, that you probably won’t want to sit next to its author in case of an emergency. In addition, the data in these first two graphics seem to contradict themselves. You would probably be aggravated that you just traded your business-class ticket for economy. This graphic varies in two aspects: both the colors and column heights change. However, it seems that the color in each row remains identical and independent from the column height. It is hard to imagine what the source data looked like.

What do different heights in the same color mean? (Source: BILD am Sonntag (BamS), 2008–08–24, “Wissenschaftler klären Frage – Fliegt man vorn sicherer als hinten? [Scientists explain – is flying safer in the back?])

The third chart takes a similar approach and appears to combine the data from the first two charts. Overall, however, it is even more confusing. The risk by location (aisle, window, bow and tail) is listed in percent. The risk based on the distance to the exit is in color. At my next check-in, I am going to demand an aisle seat – up front, light green. I hope that they’ll still let me on board.

Colors and percentages stand for different data. (Source: WAMS, Nr. 9, 2009–03–01, p. 76)

A small improvement would be to get rid of the traffic light colors, which we have criticized here on more than one occasion. The most important argument is that the human eye cannot intuitively sort red, green and yellow colors in a natural order. That works much better with shades of gray or another single color as our redesign below shows.

An attempt from our side – but it, too, has its limits.

The problem is, however, that you cannot use a single color to visualize any number of different levels. Differentiation is lost. In most cases, five levels are the limit. The darker the tone is, the harder it is to differentiate it. In our redesign example above, you can easily see that in the color scale. The so-called ‘edge fluting’ makes the whole thing even more difficult. Imagined color differences occur on uncontrolled color transitions which need to be suppressed on the borders with additional color changes.

For many types of data, it’s important to be able to differentiate between positive and negative values. This, of course, limits the use of a gray scale which is very helpful when printing hard copies. In our DeltaMaster software, we generally manage this with two-color scales: different shades of blue for positive values and red for negative ones. In this case, however, there are still difficulties with print-outs.

These examples show just a part of the problem. Over all, I think that no visualization can treat this topic fairly. You can quickly see how thin and inaccessible the necessary data are when you try to look for sources. All of the graphics suggest, however, that a person can differentiate the already relatively low risk. That’s nonsense. What we have learned is that it’s better not to draw anything if you can’t produce anything reliable and legible in the process.

In the process, we have almost nearly forgotten the dangers of flying. That’s good because there aren’t many. Our security depends on our own actions. Wear flat shoes. Know where the exits are. Inflate your swim vests outside of the aircraft. And before the crash, lean forward and protect your head. Have a safe trip!

If you visited our Christmas market (Christkindlesmarkt) in Nuremberg over the past few weeks, you probably had the opportunity to observe the ‘Männleinlaufen’ at 12 noon. This world-famous clock above the city’s main market (Hauptmarkt) is truly a piece of art. A herald appears and a marshal gives the beat for a group of trombone, flute and drum players who welcome the seven electors. Carrying the imperial regalia, they bow and circle three times around the Emperor Karl IV who greets them with a scepter in hand. These figurines, which are over 500 years old, survived two world wars and still run today.

Figurines that work: Bechtle AG shows its growth from one employee in 1983 to 4,250 in 2007 (closer view).

The visionary Otto Neurath, a highly educated man who was well versed in picture languages dating back to hieroglyphics, invented pictorial statistics back in the 1920’s. Figurines also played a key role here as well. As we can see in this example, pictorial statistics use minimalistic, standardized symbols that are free of unnecessary decoration. Another important aspect is that they use many similar symbols – and not simply different sizes – to represent numerical proportions. Pictograms, which we can attribute to Neurath, still help us find our way around airports, train stations, etc. today.

In the foyer of Bechtle AG’s corporate headquarters in Neckersulm, Germany, visitors can witness a piece of its corporate history. Using simple board game pawns, the company shows its astonishing workforce development in a highly effective yet humorous way. Equally impressive is the accompanying display that uses labeled hurdles to show its sales growth from €33,000 in 1983 to €1.4 billion in 2007.

In contrast a graphic from the magazine ‘Wirtschaftswoche’ (No. 11, 2009–03–09, p. 15) is a near caricature of Otto Neurath’s approach (I personally find the visualization confusing and feel that mixing terms like ‘debt’, ‘risks’ and ‘liabilities’ is completely biased – but that’s beside the point). Although the figurines resemble the standards developed by Neurath and the graphic designer Gerd Arntz, they vary in size and not in number. We’ve discussed the problems associated with spatial graphics on multiple occasions. What I particularly find annoying is that people can have debt but debt isn’t made up of people. Otherwise, we should think long and hard if we really want something today to look like it’s stuck in the 1920’s. After all, Neurath’s pictorial statistics were designed to give illiterate or poorly educated people the knowledge they need to successfully participate in a democracy.

The Wall Street Journal (WSJ, 2009–11–13, p. 9) recently asked 52 economists where they expect the next bubble to emerge in financial markets. The charm of the selected visualization lies in the transparency of the underlying data.

A treasure upon closer examination: The reference to opinion and person remains visible (Source: Wall Street Journal, 2009–11–13, p. 9).

The graphic designer could have closed the bars, labeled them or simply used percentages instead but chose not to. The fact that the tally is missing is a nice reminder that this graphic shows trends and nothing more. The summary of ‘Most likely’ and ‘Second most likely’ into segments makes other aspects of the data clearer as well. The approach that the Sunday edition of ‘Die Welt’ took in its panel of entrepreneurs is cryptic in contrast. Here, the opinions of twelve general managers from midsize companies are simply lumped together in an index (WAMS, 2009–03–16, p. 29).

The topic of people and graphical figurines is multifaceted. Even Bella has given it some thought – and quickly discovered what additional problems can occur when analog stick figures are supposed to represent percentages. Neurath’s abstraction in which one figure, for example, stands for 250,000 workers in the Soviet Union is a talent that needs to be learned. In general, stick figures work the best when one figure stands for one person. That’s why the figurines at Bechtle are so effective – and the ones high above the Christkindlesmarkt have been as well for over 500 years.

The Wall Street Journal Europe, which has been on the market since 1983, underwent a relaunch on November 17th, 2009. Its American parent company, the Wall Street Journal, often targeted the publication not to bankers but to their customers for a large portion of its 120-year history. Accordingly, it already got a facelift back in early 2007. One column was omitted and some content was moved to the online version. All this happened before Rupert Murdoch took over the helm on August 1st, 2007 and began to make the time-honored business and financial publication a ‘complete’ newspaper by adding sections on politics and culture.

      
Hold on to the little table on your left. It is the last of its kind and could increase in value someday. As of November 17th, 2009 it became history and was replaced with a ticker that doesn’t even tick. (Title pages of the WSJ Europe from November 13th and 17th, 2009)

The number of relaunches is growing steadily. Every time, we hear about the great concern over the demise of our information culture. In the PR department, these efforts were called ‘investments’ while the finance department called this same procedure ‘cost-cutting’. In fact in the WSJ Europe’s case, the goal was to cut costs by €18 million annually.

This clear, easy-to-read table filled with facts and figures was second to none. The title page of each WSJ Europe listed the changes of important indicators compared to the previous day – until November 13th, 2009.

The relaunch of the FAZ on January 20th, 2009 was primarily attributed to cost-cutting as well. The stock section was trimmed down by an entire page. Some things disappeared completely while others such as range visualizations were added. How nice – in part because it bears a certain resemblance to one of our own inventions.

Since November 17th, 2009, WSJ Europe has replaced its front-page table with a ticker that uses triangles instead of algebraic signs.

Let’s go back to the Journal’s facelift. The dozen or so news briefs that graced the left column on the title page moved to page 2 where it now looks anorexic in comparison. The first issue after the relaunch only had room for five little news blips. What a shame…

Fortunately, the well-known standard charts that are drawn in that clean, clear design that I love and I often praise as good examples are still there. Almost. For a long time, I looked forward to seeing that functional table showing the most important index developments on the title page. After all, other publications use the front page to torment their readers with trend arrows where there are no trends to report in the first place.

That beautiful table was replaced with a paper (tiger) ticker – a broad jump for your eyes on the upper corner of the page. Tickers move so that your eyes can stay still – but that doesn’t work on paper. In the first issue every index moved in the same direction. I had to wait for the second issue to be sure – and my instincts were correct. Rupert Murdoch did away with pluses and minuses and opted for triangles instead. If the point faces upwards and is green, that is a plus sign. A point that is red and faces downwards is a minus sign.

Oh, Schumpeter…stand by us.

On a beautiful autumn weekend, I took a drive through the quaint Wispertal valley. During my trip I came across a road sign that was supposed to attract motorcyclists to a local restaurant. The only problem was that in two words on the sign there were four mistakes – including a horrendous grammatical error. I suffered.

Four mistakes in two words. According to Duden, the ultimate authority for German spelling and grammar, this sign should read “Biker willkommen!”

The German magazine Focus recently reported the growth of the VW Group since its takeover of Porsche. It showed that Skoda, a lower-priced Czech brand of cars generated €8 billion in sales for 62,600 vehicles. One reader noticed, however, that that couldn’t be right. Although Skoda produces cheaper vehicles, it would have generated higher revenues than Porsche – only with fewer cars. Focus admitted that the reader was correct. The numbers were wrong; a zero was dropped by accident. The right number was 626,000 vehicles.

Source: Focus 31/2009, page 25.

The Sunday edition of the newspaper “Die Welt” recently published two good diagrams regarding foreign aid for Ghana and Malaysia. Unfortunately, there was one small mishap. Someone wrote the word for source (“Quelle”) but forgot to list the source itself.

These are just all examples that someone should and could have prevented if the material were carefully proofread in the first place. Each text, each chart deserves a second glance. After all, wherever small mistakes are lurking, large ones surely are as well. Even the best data visualization will suffer if there are errors in the headline, legend or comments.

Some people may think it’s pedantic to emphasize this but I think it is common courtesy. Conventions make it easier to understand things. Even a pragmatic thug in language issues has to admit that correct and formal language is good marketing for our reports. After all, who should take our message seriously when we were too lazy to read it carefully ourselves?

Today, I have to travel almost exactly 1,000 km. I planned this trip a long time ago. I did, however, quickly register that it would fall on Friday the 13th. So, I prepared myself…

Fridays are dangerous
If you are superstitious, here is the good news first. Fridays are much more dangerous than other weekdays. The reason is anything but trivial. For most people, the weekend starts on a Friday. More traffic leads to more accidents. This is backed by our own small analysis of current accident statistics.

More accidents take place on Fridays than on other weekdays. (Source: The Federal Statistical Office of Germany, Accidents with Personal Injuries Between 2000–2008, Chart: DeltaMaster).

Thirteen is irrelevant
Using accident data collected in Western Germany from 1964 to 1982, Müller showed that more accidents do not take place on Fridays that fall on the 13th than on other Fridays*. Wunder confirmed this observation in Germany from the years 1985 through 1999**. And we made the same conclusion based on the data from the years 2000 to 2008:

More accidents do not take place on Friday the 13th than on other Fridays (Source: The Federal Statistical Office of Germany, Accidents with Personal Injuries Between 2000–2008, Chart: DeltaMaster).

That means that we should be able to cope with the dangers of the current day. This belief, by the way, is widespread. At the very least, most people make occasional jokes about the date. Friday the 13th, however, primarily takes place in the media. Today, it will be fun to count how many times someone talks about a century-old superstition*** that is backed by current accident statistics. That, too, is a myth. Friday the 13th is a modern fabrication. In Germany, the first references date back to the time following World War II. And probably we will see some therapeutic nonsense again that the number thirteen actually brings luck.

We all want to believe that our times are more enlightened than those of the past. But they are not. To the contrary: What some claim to be enlightenment is often just a new superstition in disguise.

And if you still don’t want to leave your home today, try practicing your pronunciation of paraskavedekatriaphobia. That will surely occupy some of your time. This horrible word describes the fear of days like today. But does anyone really suffer from it?

Sources:
* Müller, A., Mehr Verkehrsunfälle am Freitag, dem Dreizehnten?, Zeitschrift für Parapsychologie und Grenzgebiete der Psychologie 30 (1988) without No, p. 226 ff.
** Wunder, E., Die Folgen von ‘Freitag dem 13.’ auf das Unfallgeschehen in Deutschland, Zeitschrift für Anomalistik, 3 (2003) without No., p. 47 ff.
*** Dreischer, M., Heute ist ein Glückstag, Rheinische Post, 2009–02–13, p. C8.

Hearing is one of our most powerful senses*. We can close our eyes but
not our ears. We can see something without realizing that it is even there. In other words, we can oversee things and see through them as well. Our ears, however, are relentless. When someone turns down the volume, we strain our ears even more. When it comes to recognizing and understanding things, I have a theory that hearing and seeing go hand in hand. In data analysis, however, we rarely use them together adequately. But now we have made one small step in the right direction.

As Bella already reported a few months back, the artist Johannes Kreidler created a grotesque graphic showing the effects of the financial crisis. He placed the falling stock prices in line charts with music and mixed them with scenes from children’s ballets. I won’t comment the artistic aspect but the businessman in me keeled over in pain. Kreidler’s flippant interpretation received considerable attention. Dozens of radio stations and newspapers covered his story.

Up until now, our analytical version of audio-animated time series was only known among business professionals in the fields of finance and management accounting. The idea to use our senses of hearing and sight in data analysis is too valuable and fruitful to see corrupted as an entertaining gag in the public eye. In fact, a tone sequence controls our eyes so that we can understand the pattern in a time series more quickly.

One possibility to show data audio-visual. Press start, than each row will be played “twice”.

Here is one possibility how we can visualize data using audio-visual features. Simply click on “Start” to begin; each row will be played twice. It seems to work that way: Just because our eyes observe something doesn’t mean that we understand what we are seeing. That depends, in part, on a train of thoughts that stimulate our brains. This, however, doesn’t happen through sight alone. In this case, the effect must exceed an alarm threshold – which only happens in the case of extreme patterns. Audio effects, however, can produce the necessary stimulation. If you would like to see for yourself in person, we would like to invite you to the EVA 2009, an Electronic Media and Visual Arts convention in Berlin where we will show a live example.

We already introduced audio time series back in the year 2000. The higher the value was, the higher the tone it had. I also think that we business professionals have already shown that it makes sense to combine sight and sound for analytical purposes. We are still working on it. Seriously.

* Timo Frasch describes what the melodic components of their products mean to automobile manufacturers in a recent article in the Sunday edition of FAZ (“Die Herznote des Klangs“, FAZ No. 41 from 11 October 2009, p. 14).

Napoleon’s Russian offensive was a catastrophe. He started with 422,000 men in June 1812 and returned with less than 10,000 soldiers on October 7, 1813. His army had already lost 135,000 men two weeks into the campaign – although there were no major battles at this point. Napoleon wanted his troops to feed off the land during their advance but the enemy left nothing but ‘scorched earth’ during its retreat. Since they had no alcohol to sanitize the water, there was a rapid outbreak of dysentery. Before the battle of Smolensk on August 17th, disease, weakness and desertion had already decimated the troops to 175,000 men. Napoleon arrived in Moscow with 100,000 soldiers. He had already lost two-thirds of his main army – not to mention many horses.

Undoubtedly a graphical milestone: This 1869 visualization drawn by the engineer Charles Joseph Minard shows the data from Napoleon’s disastrous Russian campaign from 1812–1813. Without further explanation, however, it is more of an appeal than an analysis. Click on the image to enlarge.

Napoleon’s army faced many battles during its retreat. It lacked horses to pull the loads. The soldiers torched their wagons and left their dismantled canons behind. When winter arrived, they had no warm clothing. Since the horses had the wrong shoes, the number of accidents rose on the slick paths. They even burned the pontoons that they carried to build bridges just a few days before they reached the Beresina River. Lice thrived in the appalling hygienic conditions and transmitted typhus fever. Napoleon returned with less than 10,000 men.

Total: 213 words.

Minard’s graphic also describes the disaster. We see the troop size at each stage in the campaign – including its direction and location – on a two-dimensional map. Below the image, we also see the temperature scale. This visualization is often quoted as a masterpiece. If it stands alone, we can interpret a lot from – but also into – it. The temperature scale could lead us to believe that it was the harsh Russian winter that defeated Napoleon. But that is only part of the truth. I gave this picture 213 words. In my opinion, however, that still isn’t enough. Nonetheless, these words helped the graphic because they shed more light on the circumstances.

Napoleon lost for many reasons. However, we can visualize very few of them in a graphic drawing. And that is exactly the point. Our world is difficult, beautiful, complex and colorful. Although graphics are very important in the world of abstraction, they can only capture a piece at a time. Words provide the necessary means to complement our graphical argumentation. In other words, he who draws must write as well.

Let’s assume that we want to climb a mountain. Before our journey, we study its profile. If the slope remains constant, we calculate that our assent will be constantly difficult and strenuous. Experience shows that our intuition is usually correct. I assume that this ability is an instinct that dates back to the Stone Ages. Hunting and gathering food, escaping danger, or spotting optimal rest areas all required that people could assess paths accurately. And indeed, we are very good at interpreting upward and downward slopes in analog and, therefore, absolute form.

The silhouette of Mont Vinaigre in Southern France. Time series often look similar but they are not. Is that why there are so many misunderstandings?

Charts are often based on analogies to real-world objects. Take line charts, for example. Line charts – especially when displayed as area charts – often look like a mountain summit. Let’s take a look why that misconception is so fatal.

Let’s assume that we are looking at a time series with 24 periods. In the first ten periods, each value increases by ten units apiece. It then drops once by twenty units. Afterwards, each value drops by 5 units a total of 13 times. That looks like this:

A time series or a foothill of Mont Vinaigre?

A mountain climber will clearly see that the ascent is constant. The first step is just as steep as the last. A business professional, however, is more careful. He rarely focuses on absolute changes in time series. Linear scales have a constant upward slope when relative increases decline or performance drops. A logarithmic scale quickly sheds light on that:

Logarithmic visualization of the same data.

Logarithmic scales are not common. As a result, most readers of daily newspapers or managers who analyze reports are on their own when it comes to avoiding misinterpretations. I personally believe that our brains are so molded by stone-aged instincts that they can’t read the linear chart above like the logarithmic one below it. We follow the ascent with undying optimism – and can’t see the warning signs behind the drop in growth. Linear axes don’t give early warning signals.

Let’s climb another mountain – but this time, let’s choose the one in the second chart. Now, our stone-aged instincts help us feel with each step that our ascent won’t take much longer. And sometime later, the decent will get steeper and steeper. Therefore, managers who want to understand their data landscapes have to hike them in logarithmic boots.