some of the problems with data visualisation

That title might be a bit misleading, though I can’t quite think of the right wording… what I am actually thinking about are the problems of the poor folks who have to present social data visually. Even though I have seen some cringeingly horrible things produced, I am aware that most people are not doing this on purpose!

We see lovely graphics like the public spending one I mentioned earlier, and we admire the work of Edward Tufte, David McCandless and others… but these are unusual cases. These are people who specialise in the visualisation part of the problem and who have time to develop a single beautiful graphic for a single dataset — and that is not what life is like for most people dealing with social data, who are swamped with shedloads of the stuff and juggling a hundred other tasks as the decision-makers breathe heavily down their necks. And you can’t really blame the decision-makers for their impatience, since they in turn are being berated in the press for not “doing something” on every conceivable issue before breakfast.

It takes time to ponder a problem from many angles and devise a way to communicate it clearly. If the data was released this morning and the councillor or minister wants it on their desk yesterday, then you don’t have that time. You are unlikely to have very sophisticated tools, either: chances are, you’ve got Excel and maybe a couple of add-ins — if you are lucky enough to have had the time to research and identify suitable add-ins. And the thing you produce will never be a one-off. It has to be robust to whatever happens in umpteen other hospitals/schools/local authorities/nations as appropriate to the issue… and there is probably far too much data swirling about anyway, with a lot of it of dubious quality.

So, yes, it is very hard. And, yes, I still get cross when I see it done badly… Because the very beautiful visualisations are not really the benchmark. Basic mathematics is. As in the example I keep going back to of a superlatively shocking bit of averaging. You only need the simplest of graphs to tell you how silly that is. Beautiful graphics are a treat, but this is bread-and-butter. My crossness comes in large part because so many people, even those who are hired to do this work, can’t cut a slice of bread and butter it*. How can you hope to “speak truth to power” without the basic skills? I’m not going to talk about this aspect, but I reckon that has a lot to do with it.

Enough. Have a cat picture instead. I know the lids are nice and warm, but surely one would be comfier than two?

* Of course, more of that public spending could go on hiring people who have not only the bread-and-butter but also the fancy cake-icing skills, but then that would be just another stick to beat them with, wouldn’t it?

some of the problems with data visualisation” was posted by Cally on 11 July 2012 at

Creative Commons License

  1. neki rivera

    hanks for the new statesman read.ood points to ponder.