Bella consults

William Playfair invented bars, columns, and pies. Over 200 years ago. And drew many timelines.

People still draw the same way. Not always a good idea. For example, William drew this. Above: England buying in Asia. Below: Asia buying in England. People in England are angry.

Exports and imports.

You think: Gap widening, widening, widening, staying wide, slightly closing, closed. Starting in 1755, parallel at last, parallel, parallel. Starting in 1765, widening again, widening, widening. But that’s not the case.

Orange lines by me. Click to enlarge.

You don’t see the hump. Because we see a horizontal parallel line. And ought to be looking vertically. Rather arduous. Our eyes do this only if we know the trick. That’s what he says as well.

The FAZ newspaper in Germany drew it like William. And we don’t see a lot of the humps. Troubled him as well. The FAZ noticed. And created a second graphic. But it’s not clear there either. Because the red columns start at blue. Instead of at zero.

US trade dispute with China (including Hong Kong). Left: US goods trade with China in billions of dollars. Red = imports, blue = exports. Right: US balance of trade in billions of dollars. Red = China, blue = other countries. Source: FAZ, 2010–09–27, page 13. And his article.

The balance is missing again and you think it shows gaps where there aren’t actually any.

Source: His article.

Don’t resort to William all the time. Just because something’s been around for a long time, doesn’t mean it’s good.

Looking into the future is hard. Look: Forecast down, remained the same. Forecast down again, gone up. Forecast up, gone down. Forecast a little down, gone down a lot. And so on and on. Up and down.

Source: Wall Street Journal, 2003–02–07.

Therefore: If you forecast, say that. Clearly. There are a lot of nice means to do so:

Paint in the front.

Paint besides.

Fill and paint besides

Paint a box around

Very important for lines: How many points have you forecast?

The climax: Paint range

So always mark. Because: Unmarked speculation is manipulation.

Brazil, Russia, India, China, and South Africa are developing. Die Welt shows how. Memorize the color. Find the column. Read the height. Jump to the right. Find the next column. Memorize the height. Jump back to the left. Compare the heights. Jump two times to the right. Find another column. Note the height again. Jump to the left again. Compare the heights. Oh. Forgot the color? Start over again.

Economic growth and inflation of the BRICS countries. Source: Die Welt, 2011–04–15, page 13.

I see. They’re trying to show the development of the BRICS countries. In detail. The development, though. You only see this if the years are side by side, not the countries.

Redesign: me.

Funny: If you want to see developments, it’s sometimes better not to sort by time.

Not so funny: Two out of three values are speculation, but not marked as such. 2011 isn’t over yet. 2012 is still to come.

The Handelsblatt website now does sparklines. Nice actually. If only they really were sparklines.

Lines in the Handelsblatt articles. Source: handelsblatt.com/4066396, 2011–04–15.

Funny: The line for Bayern Munich (a soccer team) is the stock price for Munich Re (an insurer). And the one for Bayer Leverkusen (also a soccer team) is for BMW (the automobile manufacturer).

The line details. Try it yourself or look at the whole page.

Pity: No number after the line. The first rule in my guide to sparklines.

Great pity: The line always shows a stock price. No matter what the article actually refers to.

This one took me a while. Looks harmless. The biggest one on bottom, the smallest one on top. Together they are decreasing. Got it. The strongest drop has Chrysler. Ford has the second most. The least drop has GM. Got it wrong. Measure it. Otherwise you don’t get it.

US-Market share of American automobile manufacturers, cumulated in percent.
Source: Financial Times Deutschland, 2008–11–21, page 8.

I played with that. The smallest one on bottom, the biggest one on top. That lifts Ford and lowers Chrysler.

When you put Ford on top, then it really falls.

Watch out: The first of stacked lines is determining. Look what happens if the first line raises and the others fall. I turned around the data of GM. No changes for the others. Chrysler also raises now, although it´s falling.

Market share of GM the other way round with data of WardsAuto.com.

Therefore: Beware of flats. Don’t stack lines. Read graphics carefully. By the way: It would have worked like this:

Every line is one manufacturer. For the total amount another line. That’s all logarithmic.

You already know this: Don´t chart things in a map that don’t belong in a map. In 1945 The Economist tried it.

Distorted territory size, proportional to the Gross National Income, in The Economist 1945.

Eduard was annoyed about this already in 1972.

„These or similar cartoons of diagrams are among the worst sins in cartographical and statistical drawing.

The Worldmapper is still amused by that.

Distorted territory size, proportional to the Gross National Income in Worldmapper 2011

I´m angry about this like Eduard. He is too. Distortion, however, can be essential and correct. Look at this. There is no territory in this map. Because: For tramming original is irrelevant. The tram just follows the tracks.

Timetable of tram line 8 in Nuremberg. Travelling time in minutes. Source: VGN.

Take another look at this. Here original is almost irrelevant. But there´s more territory in this map. Because: People drive a car themselves. The territory is linear here. Reduced to distances. No left, no right, no up, no down. It works. The maps show what will happen. A series of turnings and crossings, every few miles.

Road map of South Africa. Click to enlarge. Source: Camp & Live Guide, www.caravanparks.com, South Africa

I´m feeling weak. My beautiful nose is dripping. I’m coughing. Am I sick? I take my temperature. With my new digital thermometer. Perfectly readable. 38.6 degree Celsius. I google how sick I am. It’s ok.

In general, I’m sceptical about signals. In the way Jürgen shows them, however, they work.

Source: MedizInfo.

Jürgen says: „The values of the graphic could serve as a guideline and for evaluation.“

The diagram in my last posting is a pitfall. Do you remember? Japan is too old. Look at it again. If you use columns, you must not use a logarithmic scale. If the values spread widely, you must use a logarithmic scale – and you cannot use columns.

All redesigns: Me.

Less years would have been enough. To compare the differences between dark grey and light grey on the left side with dark grey and light grey on the right side the eyes have to jump. Still quite okay here.

Stacked is okay. Just as well. In this case.

With filled columns you can see the change of the proportion very well. But only that.

The first redesign is my favourite.

Standard & Poor’s gives Japan bad marks. AA‑ instead of AA. Japan spends more than it earns. And it is too old. That’s what this diagram in the Journal wants to show.

Quelle: Source: Wall Street Journal Europe, 2011–01–28, page 15.

Count the pixels. The older population grows three times higher. The total population just 20 percent. Even so both series look equally steep. Or flat. Because: Comparing two developments over time is hard. On different levels even harder. We’ve known that for a long time. It applies to proportions, too.

Here we can understand what S&P meant. Because: This diagram shows the proportion only. In comparison to other proportions. And not to itself.

Quelle: Statistics Bureau of Japan (Hrsg.), The Statistical Handbook of Japan 2010, Chapter 2: Population. Here Sixty-Five-Year-Olds and a longer period. Doesn’t matter.

Rupert deletes out algebraic signs in the Wall Street Journal. And draws triangles instead. Red and green. We already had this. Not good.

Source: Wall Street Journal Europe, 2010–10–25, page 1.

However, the triangle is not the cause. It can be very good. As here, for example.

Source: International Herald Tribune, 2010–12–18, page 1.

Because: The triangles are located below each other. One understands right away. They are black. They do not judge. The algebraic signs are still there. They do not substitute.

Design is fine.

A value has changed by 2 %. In total. A few effects have improved it. Others have worsened it. The effects and their differences should be visible. And must not be squeezed. Therefore 10 billions are cut off. One ponses.

E.ON Group – Drivers of group Adjusted EBIT 2009 vs. 2008. Source: E.ON AG (ed.), Performance and streamlining, April 2010, page 17; PDF. Dotted lines by me.

Ugh. If anything does not really fit: omit it. Or think twice: why? Perhaps the form does not fit to the data? Or one has only miscalculated. Like here. Everything had fit in neatly.

Drawing pictograms is difficult. The Financial Times divides humans into four and amputates arms. Horrible. A little man represents 500 people. A loose arm 150. Strange currency.

Source: Financial Times, 2010–08–27, page 5.

Here it works. The pictograms are ships which were attacked by pirates. Green: The pirates wanted to hijack the ship. Blue: They did hijack the ship. Each pictogram is one ship. Exactly one. And every time it’s about one ship. Just one. Not many or parts. Ship is ship.

Source: Wall Street Journal, 2010–08–17, page 3. I took it from him.

A pictogram works like: Count, what you want to show. Draw, what you want to count. Draw one of it. One per thing. If it doesn’t fit, no pictogram.

Last time: Spoiling shadows. Today: enlightening shadows. Sometimes a column is more imporant than another. Then it is highlighted. That works good with shadows. As we see here.

Source: Wall Street Journal, 2010–11–30, page 25.

Shadow reduces contrast. That’s why sometimes bold is used additionally. Or we differentiate. Shadow and bold: very important. Shadow without bold: nearly very important. As we see here.

Source: same.

I like figures. And tables. And figures also like tables. A chart has to prove first, that it is better.

But some tables look like zebras. The stripes group. The contrast changes. The readability suffers. So does the understanding. Those few figures would not have needed the pattern.

Source: Financial Times, 2010–08–23, page 2.

Here we have many columns. We better divide. Otherwise, the eye can loose the scope and fall off the row. The shadow is neat. But groups again. And makes it less legible. The shadow is good for emphasis. Not for separation.

Source: Neue Zürcher Zeitung, No. 261, 2010–11–09, page 36.

The zebra is better slim. Unlike the Labrador. Take thin lines after every third through fifth row. For wide tables after three. For small tables after five. Then nothing that’s not a group will be grouped. The contrast remains the same. Also the legibility.

Source: Same newspaper. Same page.

A table with few columns does not need anything.

Source: FAZ, Nr. 59, 2010–03–11, page 25.

One of my rules for all charts: My rules are not for all charts.

In an issue of the USA Today: How do youth hockey players injure. Very often somehow. And quite often others. Intentionally. A decogram, he would say. First I thought: The chart is not serious. The subject isn’t either. The chart doesn’t want to be serious. It doesn’t have to. Then I pondered: Injuries are not funny.

Source: USA Today, 2010–10–18, page 1.

In the same issue of the USA Today: America has spent a lot of money. And saved accounts and cars. America wants this money back. A serious subject. Shown seriously. Good.

Quelle: USA Today, 2010–10–18, page 7A.

Again in the same issue of the USA Today: a carved baseball player. Also a decogram. Again I thought: The chart is not serious. The subject isn’t it either. The chart doesn’t want to be serious. It doesn’t have to. Then I pondered: carving pictograms of humans? Not so funny.

Source: USA Today, 2010–10–18, page 1B.

Serious has to be serious. Funny may be funny. Taste helps always.