Marian Steinbach: Blog

Where Bar Charts Don’t Make It

2005/10/16

Dear reader, this is kind of a novelty: I actually wrote something more than five lines long, especially for you. I hope you enjoy it. Feel free to drop comments at the end of this page — bare in mind though that I have to aprove them manually due to comment spamming.

Now that it’s almost 60 years after Otto Neurath passed away, it still seems to be worthwhile to look back at his work. While Neurath’s influence has definitely been huge within the information graphics society, it’s more often questionable if the most basic ideas of statistical visualization have made it’s way into practice.

I came across this chart a few days ago, published by Originalgrafikservice, a subsidiary of dpa, the largest german press agency:

OGS/Danklorix chart
Number of bacteria per ten square centimeters (original Hi-Res version)
(c) obs/DanKlorix

The chart’s purpose is to visualize the number of bacteria in different places of a household. From left to right: wet mop, rinsing sponge, garbage can, fridge, carving board, doorknob, toilet lip and shower base.

Let’s not discuss all the ambiguities of counting bacteria within/on/at a fridge or a garbage can. Getting into this would proably cost way too much effort and would probably tell us that the company providing that data should rather have burnt their money, which would at least have helped them to save time and trouble. Let’s instead discuss the chart.

  • On first glance, it appears as if mop and sponge were equally contaminated with bacteria, approximately two times more than the trash can.
  • At closer inspection, the bars for mop and sponge seem to broken… oh, that could mean they hit their heads while skyrocketing the scale.
  • Indeed, their numbers are much larger than those of the rest.
  • Their numbers are not the same! Mop’s seems to be larger, but how much larger?
  • Wow! It’s ten times as large! That’s huge.

If viewers really take the trouble of trying to understand this chart and if they by chance come to the conclusion above, they are in fact only half way through. The relation between the 2nd (rinsing sponge) and 3rd (garbage can) column is still not obvious at all.

In fact, the sponge hosts 1.000 times as many bacteria per unit as the trash can. That’s quite a difference. And it’s not easily displayed in a bar chart, as the creator of the chart above might have experienced. He/she has decided to sacrifice the understanding of the dimensions within the upper range in favour of the precise visualization of the lower range numbers. That rather incomprehensible, since the communicator of that data has an interest to show that wet mops are a menace to people’s domestic hygenia.

Now, is there a better way of visualizing these figures within the boundaries of a restricted space like a sheet of paper? I humbly reckon: Yes, there is. And — beware! — it comes in ThreeDee!

Imagine the bar chart with bars in the correct scale. Suppose the lowest bar (Duschwanne or shower base), representing 20.000 germs on 10 square centimeters, is 1 cm high. The wet mop bar would have to be 500 Meters high.

Suppose, instead of the one-dimensional bar chart, the designer would have used areas in order to depict the figures. If the smallest number was represented by a 1×1 milimeter square, the largest square would have to be almost 22.4 centimeters in width. Still pretty large for a newspaper.

bacteria per square centimeter: 3D chart
Click for hi-res PDF version (673 KB)

Above is my approach to the problem. In contrast to the original, the third dimension is used for a purpose here. The chart shows several piles of little cubes, grouped to larger cube clusters of 1.000 (10x10x10) and 125.000 (10x10x10x5x5x5) units.

One could argue that it’s not possible to read accurate numbers from the chart, which can be overcome by labelling the piles accordingly. It can be argued though that it’s unlikely the objective to communicate exact numbers here. Nobody knows what 20.000 germs per 10 square centimeters mean, or do you?. Instead, viewers should rather get a sense of the dimensions in comparison to each other.

So far. If you have any comments, feel free to drop in.

3 Comments

Tim on 2005/10/24 at 12:23h GMT:

hey marian,

this is a very nice example of a discussion I titled “design & wahrhaftigkeit” / “design & truthfulness”. It’s all around us ;-) . Bad informationdesign that only requires formal aspects for describing contents.

what’s the need? a design that centers usability? I don’t think so. That wouldnt be enough. Maybe a design that centers epistemology and sociology? Still thinking…

Dave on 2005/11/10 at 01:00h GMT:

I like your attempt of visualizing the number of bacteria but now the writing below the graphic is hardly readable anymore.
I also agree with you that “Nobody knows what 20.000 germs per 10 square centimeters mean” — including me — but I miss the numbers anyway. Numbers in combination with an excellent visualization enhance the feeling of something as being concrete and reliable information and even if I do not really understand it, I at least want to be given the possibility to do so.

Marian Steinbach on 2005/11/10 at 11:45h GMT:

Dave, thanks for your comment! You are absolutely right about the labels not beeing readable. And I can understand that you want the figures visible. If the information graphics would have to be sold to a newspaper, I agree, it should be in. The focus here however was rather the schema of visualization.

Your comment

Note: Due to issues with comment spam, your comment might not be published immediately.

Subscribe to new comments via email