Data visualisation is trendy and has been for a while. This is almost completely a Good Thing. A good viz really can help people understand things that would otherwise be hard to get, or notice things that would otherwise be easy to miss. I’m also selfishly pleased about it given that my wife works on an information visualisation product and I’m working for an artist who specialises in a powerful variety of infoviz.
The “almost” is because just as with every flavour of the month, people are now rather too quick to jump on the infographic bandwagon, and get praised much too readily for having done so, even if the graphics don’t help us understand anything. A good infographic is either simple or piggybacks on something we can expect to be culturally well understood like a map of the world. I’ve been seeing more and more things turned into infographics without adding value, and occasionally actively harmful ones that confuse the viewer. Here’s an example I just saw blogged:
The basic idea of using liquid containers to show how changes in birth and death rates, out of sync with each other, have caused population growth is kind of nice, though I’m not sure how much it really helps us. Showing the population of each region as a bar in a chart is certainly nice, but tying it to the liquid containers idea gives us this:
I see at least three problems, possibly four:
- Where are Greenland, the Asian part of Russia, Australia & Oceania? (Admittedly not that many people, but they’ve still been populated for the whole time represented by this video)
- Are China & India included in Asia? The maps seem to imply so, but if the scale’s constant (also implied but not specified) then they can’t be included in Asia’s bar.
- This one’s a bit of a stretch, but: do the colours mean anything? A rainbow has an intrinsic ordering, while the continents don’t – seeing the rainbow suggests to me at least that there is something meaningful to the order.
But the thing that’s really bothering me is the size of the vessels. I look at this and immediately read the blue vessel for Asia as “almost full” while the red, orange & green for the Americas & Europe say “plenty of spare capacity”. I don’t think it’s what the makers intend to say, and if it is then they’re wrong in very important ways, but it’s very much the message I get from that image.
So please be careful if you’re making infographics! It’s just as easy to obscure your message and say things you didn’t intend to as it is to educate usefully.
Excellent post, Eldan. There is actually quite a gap between statistics people and info-viz people, which is unfortunate. Andrew Gelman has talked about it a lot — here’s a sort of meditative post on the topic. http://andrewgelman.com/2011/08/infovis-infographics-and-data-visualization-where-im-coming-from-and-where-id-like-to-go/
I adore maps, but they have two real problems you have to work around. First, the unequal areas can give awkward impressions of weight — picture those red state and blue state maps from 2004: they made the entire country look red, basically, despite the lower population of the middle states, just because they were larger. There was a cool web essay about that at the time, which I’m guessing maybe you saw back then, because I am sure somebody posted it to ReallyI’mWorking, but just in case I’ll link it here. http://www-personal.umich.edu/~mejn/election/2004/
And second, it is very hard to figure out how to display variability on maps, and variability is key to interpreting data correctly. I think this is an active area of research, actually. There are some people who do spatial statistics at UW, and in my “speed dating” with the biostats department on visiting weekend last year, I had a brief conversation with one of them about this problem.
Thanks for the kind comment.
I think Gelman’s absolutely right that the majority of infographics people aren’t coming from a stats background, and that this is a problem. It’s not that they have to be experts, but they do need a solid understanding of what they’re presenting. An example of getting it right, to my mind, would be the Economist’s recruitment of science correspondents, which makes very clear that they want science & engineering grads with decent writing skills, not professional journalists who think they know science. I was also disturbed by the part about the infographics community just not liking to criticise – it’s hard to improve with that sort of culture.
Those are both really important points about maps. The series of red-blue and red-purple-blue maps is now one of my favourite illustrations of how graphics can mislead, and I hadn’t even thought about the issue of variability. To some extent knowledge of what’s being mapped can decrease the problem, but granularity is a huge deal – for instance if anything the red-blue counties map is even more dominated by red, but I’m familiar enough with the shape of the US to understand that all the big concentrations of people are little blue patches, so as soon as I saw that map I understood the election results a lot better.