Friday, November 27, 2009

A weird new type of chart

I came across this new type of chart by way of the PST blog. It's a survey taken in connection with the upcoming climate meeting in Copenhagen:

The chart is square divided into 10x10 smaller squares, each representing 1%. I can't understand how this really adds anything to the data, since I count the points to see which area is the largest. It would be way easier just to read the numbers from a table. And to make things worse, they arrange three of these charts into three faces of a cube. And the whole survey is presented as three (none aligned) cubes. In my opinion, a total mess that in best case just adds noise.

In his blog, Jon Peltier suggests another chart. The solution is pretty good, the only improvement I can suggest is formating the questions better (to make them stand out more) and aligning the question with the start of the columns.

But looking into the questions and options, it's not only percentage there. The last question is "How high a priority (0-100) do you think the goverment should place on addressing climate change". First of all, you can't measure that in %.(how do you calculate those who haven't answered or have no opinion?). Second, have a scale from 0-100 is just forcing it into a formate that would fit this particular graph. A scale more with more then 10 steps are too finely divided to have make sense.

Also, I keep thinking of whether the questions could have been planned better. For instance, wouldn't it be better to have a similar scale on each question, for instant a scale from 1-5? To avoid confusion and make the process of answering the survey faster.


Wilhelm said...

Both approaches result in a wall of text associated with each image, although the first case was more brutal.

Agree on the lack of logic behind a 1-100 scale. Increase of the variable space without quantifiable outcomes amplifies the noise to no end.

Anders said...

Both approaches result in a wall of text associated with each image, although the first case was more brutal.

Yes. That's why I was thinking about the design of the sruvey. A lot of these questions could have been rephrases to a simple "from a scale of 1 to 5..." or something like that, and thus make the result easier to explore and the chart cleaner. As it is now, I can't think of a way of presenting this without a lot of text. And second best option is to use design to separate the questions from the alternatives, in my opinion.

Wilhelm said...

Yup. With the range of possible answers, the resolution of the survey got squashed right out of the gate