Data visualistion – a quick lesson

Have a look at this.  What do you think it is?  Go on, take a guess.

Keep thinking while I write about why it’s relevant…  I went to a seminar last week on data visualisation given by a very clever visual perception and processing expert from the ONS, Dr Steven Rogers.  He’s part of the data visualisation team (nice article about them here).  He talked about how to present statistics (actually anything numerical) for a general audience.  He warned that there are approximately 15 million adults in the UK who couldn’t get a G in GCSE maths.  So, when you present numbers to the general public, they won’t necessarily understand what you mean by a mean, or even an average.  Graphics are therefore our friends.  But not all pictures are equal.  Some pictures are hard to interpret without a bit of extra knowledge – just like the picture above.  So let me give you the equivalent of a GCSE in maths.  It’s a cow.  Ta da!

OK, so that wasn’t actually of any practical use…  But it’s still pretty cool.  This, on the other hand, is worth looking at.  The chart below is pinched from one of my new favourite blogs

It’s actually another chart that Rogers showed in his presentation.  It belongs to Branko Milanovic, a Lead Economist at the World Bank.  (BRIC refers to Brazil, Russia, India and China, by the way – no idea what happened to China on this chart…).  On the face of it, it looks as though it’s showing that all Americans, even the poorest, earn more than the richest Indians – the red line is above the orange one for its entire length.  It shows purchasing power parity (PPP) which is supposed to allow you to compare the amount you can buy across countries.  A US dollar in India buys more than a US dollar in the US.  The fact that the chart shows this (internationally comparable) measure only increases my confidence in it.

But, hang on a minute…  Let’s think about this more carefully.  In fact, the bottom left hand of the red (USA) line shows that the lowest 1% of earners earn an amount comparable to Russians on the 21st percentile (in their own country), Brazilians on the 45th percentile and Indians on the 96th percentile.  The red dotted line is the 63rd percentile of world income.  Milanovic has picked out the 63rd percentile because it’s where the bottom 1% of US earners sit in terms of world income.

The chart does contain useful information – but it’s hardly intuitive.  It’s all too easy to see the chart and think, “Aha, Americans richer than Indians” and just move on to the next thing.  Even if you read the chart carefully and realise that it shows that the poorest 1% of Americans have the same percentage of world income in terms of purchasing power as the 96th percentile of Indians, I’m not sure that it tells you very much.  What does that mean?  Do rich Indians feel as poor as poor Americans?  Seems highly unlikely.  If you thought even harder and realised that Brazil is an extremely unequal country (spanning the entire world income distribution) you could be forgiven for concluding that the USA must be a more equal place than either Russia or India – but a quick look at the Gini coefficients for these countries tells a different story (Gini coefficients are 56.7 for Brasil, 45 for the US, 42.2 for Russia and 36.8 for India).

So the take-home?  Make sure that your graph means what you think it means (to paraphrase the great Vizzini).  Edit: Inconceivably, idiocy reigns.  Of course the quote belongs to Inigo Montoya…

Another edit: If I had paid proper attention to my own blogroll, I would have noticed Junk Charts doing this and saved myself a post!

This entry was posted in Other blogs, Pedagogy and tagged . Bookmark the permalink.

2 Responses to Data visualistion – a quick lesson

  1. Essi Lindstedt says:

    Each time an individual presents data in anything other than its raw form, some degree of interpretation happens, as in the much quoted use of visualisations by Florence Nightingale as a campaign tool to raise awareness of healthcare problems. The ONS may be looking to visualisations as a way to get past ‘fear of numbers’ but political and statistical literacy are still required for a critical understanding of data, unless they are happy to admit that they only want people to take away their interpretation of the data. The government’s open data initiative is undermined by their lack of commitment to citizenship education, which empowers people with the skills and aptitudes to begin to make sense of the volumes of data that are beginning to be available.

    • Vicki Bolton says:

      I have often thought that the national curriculum should be replaced with two words: “critical thinking”. Open data implies that more interpretations should be available in future, so what we need is the skill to choose between them. Having said that, I’m broadly in favour of ONS using graphics. Their population pyramids (here for example) are great. I must agree that they are lacking certain key information, though – most notably how the future projections are made…

Comments are closed.