Graphs, and what they don’t tell you


Statistics are amazing. They can tell you anything. In fact, given a data set, you can find statistics that can conclude absolutely anything you want, regardless of the data itself. And then you can display the results neatly in a graph, in order to convey your point concisely and clearly.

Bearing that in mind, today I looked at the Economist’s Daily Chart, which claims to be a comparison of the number of death row prisoners in countries around the world.

I say “claims”, because that isn’t really what it is at all.

The graph is put together, as all such graphs are, with a nifty piece of software (I’m assuming) that includes a grim background of what is presumably meant to be a cell interior, and a graphic of handcuffs locking together two clenched fists. It’s very evocative. But the slick presentation only serves to mask the graph’s inadequacies.

If you’ll look closely at the graph, it shows Pakistan, the USA and China topping the graph. The first, and most obvious, odd point is that the bars for Pakistan and the USA are truncated. China’s bar tops out under 2,000 (with all countries below it under 1000). The bar for the USA looks a little longer, as if the figure was around 2,500 – but closer examination shows the figure to be 3,263. And while the bar for Pakistan appears to fall at around 2,800, the actual figure is 7,346.

I understand the need to make the graph clear for a range of data. But the lowest figure is for Cuba, which looks to be around 50. But the scale is at the top, so it’s hard to tell. If they were going to truncate the top two bars for reasons of scale, why not the third (for China) too? That would have allowed them to make the rest of the bars twice as big, and from the fourth bar down (Thailand), there is a reasonable range. It would certainly have been clearer.

I can’t help wondering if this was due to oversight, or whether it was trying to display a particular message – perhaps that the numbers of inmates on death row in the top three countries dwarf the numbers in other countries? Thailand at number four has 1000 such inmates. China, the USA and Pakistan have almost twice, over three and over seven times as many respectively.

But this isn’t even the most glaring issue with the graph. Look closely at the little asterisks for China, Iraq (around 200 on death row) and India (around 100). It turns out that the figure for these countries isn’t actually the number of prisoners on death row – it’s the number of people sentenced to death in 2007.

Which is completely different. Without any data on length of time spent on death row, let alone the number of sentences that aren’t carried out, these figures can’t possibly be compared. This isn’t just a little bit misleading, it’s complete rubbish. Given that this graph ran in the same slot just a few weeks ago (check the date), whoever compiles this section really ought to know better.

In their defence, the blurb above the graph acknowledges this fact.

China is third [in the figures], although the figures are not strictly comparable [emphasis mine]

I’ll accept that sometimes in statistics, data is not quite comparable, and that sometimes you have to use it anyway, with a disclaimer. But as someone in the comments section points out, it’s like comparing the number of iPods sold in 2007 with the number in warehouses (except that there is even less correlation – even the most twisted of judges will not increase the number of death row sentences just because more people are being executed). So why not leave out the data for these countries, and only include the ones for which it makes sense to compare the data?

An answer to this may too lie in the blurb.

China’s seemingly lowly position may well reflect the fact that it carries out executions swiftly rather than allowing a long judicial process of appeals. Such a long process may explain the relatively large death-row population in America.

To me, this implies that whoever composed the graph wanted to include China, and wanted it to be at the top. This opinion is further backed up by the fact that the cropped “teaser” for the graph on the front page cuts off the top two bars – a casual glance shows China at the top, and it’s not until one clicks through to the article that the bars for the USA and Pakistan are revealed.

China is notoriously secretive about its death penalty process, but recent Amnesty International figures put it at the top of known execution figures. If the composer of this graph wanted to draw attention to the country’s use of the death penalty, why not use these figures instead of the numbers of prisoners on death row (especially since the article actually mentions this report)? While they are not entirely reliable (since the countries that use capital punishment are often countries not keen on international observation of any form), they do at least all measure the same thing, and can be said in some sense to be comparative.

If the intent was to focus on prisoners on death row, then why not include figures for China that reflect this, with the disclaimer that China has fewer appeals processes than other countries. For that matter, why is it only China’s and the USA’s appeals processes that are noted? I can’t believe that the average length of time spent on death row is uniform in all other countries, with their widely varying legal systems. Is it because figures aren’t known, or because acknowledging them would weaken the appearance of the graph still further?

I don’t read the Economist often enough to know in detail its position on China and the death penalty, but this seems like deliberate skewing of the data to draw attention to the country. I absolutely feel attention should be focused on China’s consistent abuse of human rights, not to mention the use of the death penalty in countries worldwide, something that I personally find repulsive. Amnesty International and other organisations do sterling work in investigating and reporting human rights abuses and campaigning against the death penalty. This year, when the Olympics draw attention to China, is a good time to focus campaigns on that country.

Graphs, used properly, can be used to convey a great deal of information, as the Economist has noted in the past. But used sloppily, they do no good at all.


One Response to “Graphs, and what they don’t tell you”

  1. 1 Some pie charts, of questionable utility « Gin&Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: