Tuesday, March 25, 2014

The danger of not adjusting for inflation in infographics

I'm teaching at the Malofiej news graphics workshop and summit this week. It's been fun so far. We have enjoyed a talk by John Grimwade already (see photos at the end of this post,) and I discussed common pitfalls in data graphics. The latest case I included in my slide deck came from Marca, a Spanish sports daily. It was published in August 2013.

The infographic compares the most expensive soccer players in history, up to that date. The headline means "The Hundred Million Euro man." It refers to Gareth Bale, now playing for Real Madrid. The second —from the bottom— is Cristiano Ronaldo, who signed a 96 million euro contract with the team in 2009. Impressive ranking, right? It is. But it's even more impressive that Marca got everything wrong, as these numbers are not adjusted for inflation.

You read that right. How's that even possible in 21st century journalism? Haven't we learned Numbers 101 yet? Not adjusting for inflation is like comparing the number of homicides in New York City (population: 8 million) with the homicides in Tucson, Arizona (population: 524,000.) It wouldn't be fair to do so. It'd be better to use homicide rates (cases per 100,000 people.) Absolute numbers are tricky. In most cases, it's advisable to normalize them somehow before jumping to conclusions.

The Marca folks cannot say they weren't warned against this blunder. Idoia Portilla, a professor at the University of Navarra, told me that a journalism student of hers, who was an intern at Marca at the time, tried to convince some people in the newsroom that their story was not really there. He calculated the real values and came up with the bar graph on the left. The reaction of the seasoned professionals he talked to was appalling: They argued that Marca's readers don't understand "complex" terms such as "adjusted for inflation." I believe that this a bad excuse, at least in part.* The reason why they didn't listen to the intern is that they wouldn't have had a story if they had told the truth.

(*I wrote "in part" because many journalists despise their audiences, thinking that readers are less intelligent and educated than they are. The opposite is usually true, of course.)

According to Malaprensa, an excellent Spanish media criticism blog, the case I described in my talk was not the last time Marca presented data without normalizing them. The graphic below is even worse, as the proportions are all wrong. Just compare 57 with 111,7.




And now, the photos from John Grimwade's presentation at Malofiej:


2 comments:

  1. There's another danger not so well-known: the danger of not adjusting for age structure of the observed population. That's called rate standardization and is used to make comparisons among populations in time or space. But you usually never see it in newspapers, they are used to raw rates, which are easier to calculate but deceptive.

    ReplyDelete
  2. True. Rate standardization may be hugely important in health stories, for instance, to make populations comparable:

    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3406211/

    I also forgot to mention that some of the original data they were using in the first infographic were in British Pounds. At first they calculated the values in Euros based JUST on the current exchange rate, rather than looking for the average exchange rate on each year

    ReplyDelete