Top 5 types of graphics most often misread

4 February, 2020
Episode notes

Guest

Alberto Cairo

Knight Chair at the University of Miami. New book: “How Charts Lie” (WW Norton). Previous: “The Truthful Art” (2016). Consultant and freelancer.


Mark gets political and data-visual with data visualisation expert, Alberto Cairo.

Alberto has a journalism degree and worked as a graphic designer for many years. He now teaches the subject, and quite literally wrote the book on it. Although a lot of what’s covered in this episode is US-centric, Alberto maintains that the misuse of graphics isn’t a partisan problem linked to whether someone’s in a blue state or a red state, but essentially that the right is guilty of bullshitting.

Bullshit is a word you’ll hear a lot in this episode, and especially refreshing in a Spanish accent. But in this context, it has a very specific meaning, with its roots in a book by Harry Frankfurt.

In Alberto’s latest book, How Charts Lie, he reminds us that “a chart shows only what it shows”, and nothing else.

Advisory

In the discussion around Mark’s fourth pick — about 38 minutes in — Mark and Alberto discuss how the question of abortion is tackled within the US, in order to bias survey answers. They don’t get into the actual topic of abortion obviously, but some of the language around it is a little bald.

Alberto’s picks

In order of discussion:

Choropleth map

This type of map which uses colour to represent data was used especially heavily in the 2016 Trump election campaign, to argue the amount of popular support he had. It was recently brought back out of storage with the tagline “Impeach this”. Here’s a piece from CNN debunking this map.

Manipulation of line charts

In his book, Alberto cites an example of a climate change chart, representing temperature change from 0 to 100℉, and ignoring the impact of a change of a single degree.

Perspective effects

3D charts might look nice, but you can manipulate them very easily based on where you place the virtual camera. Even a subtle shift in perspective can make something look a lot bigger than it is, or heavily emphasise one bar in a chart over another. Alberto’s advice: if you’re showing data on a 2D service like a screen or in print, don’t be tempted by 3D charts.

Projection

Alberto references this tongue-in-cheek article from the New England Journal of Medicine, whose scatterplot graph seemed to suggest that the more chocolate a country ate (kg per year per capita), the more Noel Laureates that country had per 10 million people. The reason the chart is misleading is that it doesn’t go into why a country consumes more chocolate, and why a country has more Nobel laureates than another (since these two things aren’t necessarily actually related at all).

Sourcing of data

Alberto gives an example of a map showing homelessness in Florida, which seems unaccountably high due to that particular state’s definition of homelessness, which demonstrates how important it is to understand where a set of numbers comes from. The numbers in a chart or a graph might be entirely accurate, but it’s the context behind those numbers: how they’re gathered and how they’re interpreted, that can make a difference.

Mark’s picks

In order of discussion:

Omitting the baseline

This particular practise is useful in comparing two percentage.s If there’s a < 2% difference in something, a bar chart whose scale starts at 0 and ends in 100 isn’t going to show that difference, but you zoom in and start from, say, 85%, you can make the difference look much more dramatic.

Distorting the X axis

You might be forgiven for assuming that an X or Y axis in a chart should go up or down in regular intervals: 1, 10, 20, 30, etc. But if that doesn’t suit your case, and you want to emphasise a particular band of numbers, or that particular segment doesn’t fit your argument, turns out you can just omit that data from your axis, and all you need to do is add a little squiggle to the axis to show you’ve made an “edit”.

Using real-world objects as bars in a graph

Similar to Alberto’s point about 3D charts, Mark wanted to bring up the use of real world objects as a replacement for the bars in a bar chart. The sport-based example Mark was reaching for came from a YouTube video on reading misleading graphs.

Reversing the Y axis

Mark brings up a well-known example of a misleading chart on people killed by guns that reverses the Y axis, showing how the number changed once the state enacted its self-defence law. The chart makes it look like gun deaths dropped after the law was enacted, when they in fact did the opposite. And just to prove how influential this kind of tactic can be, Mark gets the meaning of the chart entirely upside down on mic.

Dive deeper

Mark mentioned a move by the Lib Dems in 2019 that raised a few hackles online, which was covered in the Guardian.

More of Alberto Cairo

You can follow Alberto on Twitter, check out his website, and read his blog, The Functional Art.

Links