There are so many bad charts published every single day that redesigning some of them would be a full time job. Many are involuntarily funny. And there are some dangerous and stupid ones.
I live in Portugal, a small country that only gets international press coverage when something goes wrong. Like now. Public debt is a serious problem, and, judging from this chart above, we are doomed.
What can we do when debt starts skyrocketing by 2015? Apparently not much. Unless…
Unless there is something wrong with the chart. Let’s take a closer look at the x axis:
The scale is not proportional, and that explains the exponential growth. Now, let’s correct the scale and compare the trend:
While it still looks bad, the message both chart send is quite different: “doom and gloom” on the left, “maybe we can clean up this mess” on the right.
Apparently, the data was too boring, and the graphic designer decided to spice up the story a bit. But the chart was published in Expresso, a very popular Portuguese weekly newspaper, and given its high standards, I want to believe that’s not the case. So, only unforgivable ignorance and/or negligence can explain this.
Your thoughts?
Jorge, ever heard the old chestnut “If it bleeds it leads”? I say this is two parts deliberate sensationalism 1 part visualization ignorance.
Newspapers are in the business of selling newspapers and advertising. The more shocking a story it is, the better it will sell so there is an inherent incentive (conscious or unconscious) to make any story as — I shall be nice and say — compelling as possible. The designer likely, IMHO, did as instructed by the editor and/or author to dress up the data a bit and was probably uneducated in why what s/he was doing is very bad from an information visualization standpoint. If the designer is trained in information visualization and also did this on purpose then s/he is complicit in the lie.
Nice deconstruction by the way.
I agree with Clint in that there is probably something to the idea that the editor took a little poetic license to make the story more “interesting”.
Nice post!
Hi Jorge,
It´s easy to see bad charts everyhere. In March-April, when companies (big companies) start to explain their results to shareholders, it´s like a leasson of “don´t ever do this!”. We can´t be so hard on them. Sometimes it takes time to do it right. When I see Stephen Few writing about good and bad charts, I always think of Armani trying to make everyone use the right clothes. We have to be patient because not everybody has the know-how (technical and visual) to elaborate the best graphics.
About the chart in “Expresso”, maybe the political tendencies of some journalists leads them to use graphic skills to present results in a special way or are we just seeing some ignorants in charts?
Thanks for your blog, it´s a piece of art.
We have do have a coffee at the “paredão/marina de Oeiras”, one day.
Jorge Caldeira
This is the classic illustration of the difference between X axes in Excel’s line chart and XY chart. The line chart spaces data points evenly, leading to the hockey stick in the publication. The XY chart spaces data points according to X value, leading to the more gradual increase in the last chart.
In fact, it looks like Jorge drew his two charts using the exact same data but with the two different chart types.
@Jon: As you know, if you use a date, Excel applies the right spacing even if you use a line chart (here both are line charts)
Of course, it is quite possible that the future figures have been predicted based on projecting the current growth rate forwards in a linear fashion with some minor adjustments (eg for changing age groups in a population, working or drawing a pension for example), so the (almost) straight line chart simply shows these *predicted* figures *predict* a roughly straight line. All of this is speculation – I bet charts from ten years ago looked very different from the actual figures we see today around the world.
I would give the chart designer the benefit of the doubt and accept that they might just be stupid, not mendacious.
They probably fell foul of simply drawing the points one at a time, labelling them in their graphics drawing package as they went along, and simply lost sight of the context. This is akin to doing this in Excel with text labels for years, instead of real dates (as Jon/Jorge point out in different ways) so the context of a timeline is lost into simple categories.