Infographics and activity strips: what a visual shows and what it hides

0:00Here are two pictures. A map with every region painted a different shade, and a ribbon of coloured boxes. Neither is a graph. A graph puts a value at a position on an axis, and you read it back off the axis. These two do something else. They turn a number into a colour, or into a length, and hand it straight to your eye. That is a trade, and it always costs the same thing. The pattern becomes instant. The individual numbers become approximate, or unrecoverable, or both.

0:36So the question to ask of a picture like this is never whether it is accurate. It is: which questions can it still answer? Start with the map. Twenty-two regions, and one number each, shown as a shade. The number is a difference. How many kilograms of rice a person eats in a year, less how many kilograms of bread. The scale runs from minus a hundred, mostly bread, through nought, to plus a hundred, mostly rice.

1:10Nine regions come out on the bread side. Thirteen come out on the rice side. And the picture has already done something a table of twenty-two numbers cannot. You saw that without reading. Now look hard at that far end - the most misread thing on the picture. Plus a hundred does NOT mean a region eats no bread. It means the gap is widest. Here is a person eating a hundred and ten kilograms of rice and ten of bread. The difference is a hundred. Plus a hundred.

1:45And here is a person eating a hundred and sixty of rice and sixty of bread. The difference is also a hundred. Also plus a hundred. Same colour, exactly. The first eats a hundred and twenty kilograms of the two altogether. The second eats two hundred and twenty. Nearly twice as much food, painted the same shade. A scale built from a difference says nothing about either quantity on its own. And it cannot be undone. Allowing whole kilograms up to four hundred, three hundred and one different diets give that same plus a hundred.

2:28Now the thing that makes this picture famous. One line, drawn diagonally across the country. It cuts the bread regions from the rice ones almost cleanly, and it is the first thing anybody notices. It is not data. Nobody measured it. It was drawn on top of the numbers afterwards, by somebody who had already looked at them. Rub it out and you lose nothing measurable. Rub it out and you lose most of the picture's punch.

3:00A line somebody drew and a line somebody measured are different things, and the picture does not tell them apart. Which questions does it answer well? Ranking ones. The five leaning furthest towards rice: Manir at a hundred, Nagel at ninety-nine, Mizen at ninety-seven, Trippen at ninety-six, Meran at ninety-five. And furthest the other way: Rosmere at minus ninety-three, Halden at minus eighty-one, Pellan at minus seventy-eight, Marchfield at minus sixty, Delvin at minus forty-five.

3:42You got both from the colours, without reading a number. That is the encoding earning its keep. Ask for the top nine, though, and it refuses. Three regions sit on exactly eighty, and a top nine would cut between two of them. There is no top nine, and handing you one would be an invention. And the balanced middle: Birch, at plus three. Four regions sit within nineteen of nought, and Highmoor is exactly nineteen.

4:15Now the interesting one. One region here is named and carries no number at all. Karn. Guess it. Its shade matches three regions that border it: Kelvin at seventy-nine, Tamar at eighty-five, Andover at ninety-two. So the colour licenses a range. From seventy-nine to ninety-two. Thirteen wide. Eighty-five sits inside it. Forty-five does not - however confidently anybody says it. A fourth neighbour argues the other way: Marran, at minus fifteen, sharing a border and nowhere near the shade. It loses. Karn is painted the colour of the other three, and the paint outranks the border.

5:02The honest answer is a range and not a number. That is not the guess failing. That is the encoding telling you what it spent. Here is a trap the picture sets without meaning to. Average the twenty-two numbers on the map and you get about twenty-seven. Firmly towards rice. The figure the country reports for itself is thirteen point four eight. Both are right. They answer different questions, and the map cannot show which.

5:35Because a map gives every region the same patch of ink, whatever number of people live in it. Take three regions. One large at minus sixty, holding four people in every five; two small ones at plus thirty each. Average the three numbers: exactly nought, perfectly balanced. Average over the people: minus forty-two. Same country, same three numbers. The map cannot tell you which average you are looking at. Second picture. Somebody records a whole day by colouring a strip.

6:13Forty-eight boxes. Each one is half an hour. Midnight at the top, midnight at the bottom. No axis, no numbers, no labels. Same trade as the map. Length is the encoding here instead of colour, and you can read the shape of a day from across a room. Three days, three strips, and nobody says what the colours mean. Working that out is the exercise. Six colours, and three things to argue from: where a colour sits in the day, how long it lasts, how often it comes back.

6:51The long block every strip begins and ends with has to be sleeping. Nothing else runs for eight hours twice a day. The colour that comes immediately after waking, every single day, is washing and dressing. The colour appearing three times a day on all three strips, always in single boxes, is eating. The colour that only ever appears in single boxes, and only beside the long blocks, is travelling.

7:22That leaves two. The colour filling the middle of two strips is classes and study. The one filling the third, and the evenings of the other two, is friends and hobbies and family. Every one of those is an argument from position, length and repetition. That is all the picture gave, and it was enough. Now which day is which. The second strip has a study stretch fourteen boxes long. Seven hours, from ten in the morning to half past four. A full school day.

7:59The third has six boxes. Three hours, then the afternoon free. A half day. The first has no study stretch at all. And there is a break inside those seven hours. Two boxes at one o'clock: half an hour eating, then half an hour free. An hour's break, of which half is lunch. One more thing hides in the first strip. A three-hour block with a single travel box immediately before it and another immediately after.

8:31Went somewhere, stayed three hours, came back, starting at two in the afternoon. That signature appears nowhere else. So what did it throw away? Anything shorter than half an hour. A twenty-minute walk cannot be written down. Anything done at the same time as something else. Homework in front of a screen is one box, and you must choose its colour. And the difference between one long stretch and several short ones. Here are two days with identical totals, activity by activity.

9:09One holds a single three-hour block. The other holds six half hours scattered through the day, never two in a row. Identical totals, completely different days. A box forces one label onto half an hour, and that is the price of being readable across a room. Last thing. Average the three days activity by activity, and draw the answer as a strip. Study comes out at eight boxes. Four hours.

9:41And here is the lesson in one number. The three days had one hour, six and a half, and three and a half. Not one of them had four. The average day is a day nobody lived through. That is not a flaw - that is what averaging is. But now try to draw it. Only two of the six averages are a whole number of boxes: eating, and study.

10:07Sleeping averages twenty-one and two thirds of a box, and you cannot colour in two thirds of a box. So the drawn strip has to round, and four of its six blocks come out ten minutes wrong - two too long, two too short. And they cancel. The picture still fills exactly forty-eight boxes, looking perfectly correct, while four of its six blocks are not. That is the trade, one last time. Fast to read. Never quite readable back.