Lying with charts is pretty easy
Charts are important because we use them for important things. They’re how we run businesses, manage finances, discuss societal issues, and debate politics.
But it’s so easy to lie with charts. And I can prove it.
A few years ago Vanity Fair did a poll to find the most popular batman: Christian Bale, Adam West, Michael Keaton, Ben Affleck, etc. As you’d expect, Christian Bale won, but I was surprised by how badly Ben Affleck did.
Now I don’t particularly like Ben Affleck’s Batman, personally I’ll always be a fan of Adam West.
But Ben Affleck is already one of the saddest people on earth, the last thing he needs is more bad news.
So let’s cheer him up by making a pie chart that makes him look more popular.
First we use Excel to turn that table into a normal pie chart with all the default Excel settings. Ben is very clearly in 4th position.
Next let’s make it 3-D.
Now that we have a 3-D perspective, let’s move Ben Affleck down to the bottom so that his segment will be closer to us and will therefore appear larger.
Next we just tweak the perspective to make Ben’s pie segment look as big as we want.
I feel like the chart on the right is pushing the perspective just a bit too far. Let’s not be too ambitious, I think Ben would be happy if he was the second most popular Batman.
I tried to find a picture of Ben Affleck happy and this was the best I could do.
Now you might be thinking that this example is a bit contrived. It’s obviously ridiculous isn’t it? Surely no-one would ever try this in real life. Would they?
Well, take a look at this pie chart that Apple used in their 2008 keynote:
Notice how the 3-D perspective makes the 19.5% “Apple” share of the market appears much larger than the 21.2% “Other” section. They made their Market share appear larger using the exact same technique I just used to make Ben Affleck’s Batman appear more popular.
For my next trick, I’m going to make global warming disappear
NASA has published the average global temperatures for the past 130 years so I’ll just use Excel to put them into a line chart.
As you can see, global temperatures were pretty stable until the 1930s when they started rising steadily. Now watch what happens when I change the minimum value on the Y-Axis scale from 55 to 0.
Global warming disappears! But surely no-one would do this in real life would they? Well…
Manipulating charts isn’t just for climate skeptics either. We can go the other way too, we could make global warming appear even more extreme. We just tighten the Y-Axis and add a trendline:
Why is this so easy?
The problem is that although charts look objective and mathematical and scientific, they’re actually none of those things. Charts take something that’s abstract and turn it into something that’s visual. They take numbers and turn them into an idea, or a feeling or an opinion.
Every chart gives you a perspective on the underlying data, and like any perspective, you never get the full picture.
When we create charts, we often choose a perspective based on what we want to see. So if we’re not careful, we can literally hard code our biases into the chart.
Does this mean charts are bad?
No of course not. Charts are amazing tools, but like any tool, they can be used well or misused. All of us create or consume charts because they’re awesome. If anything I think the world needs more charts, but they have to be more good charts.
When we create charts, we simply need to be aware of our own biases. When we see charts made by someone else, we need to be aware of the biases that may be subtly encoded into them.
I actually don’t think everyone out there making charts is intentionally lying to us. The marketing team at Apple made a misleading pie chart because they’re biased to think Apple is a great company. So what? Of course they’re biased! I’d be more surprised if Apple’s marketing team didn’t think Apple was a great company.
I still use charts all the time, I love them. I think they’re useful and valuable. I even made charts to help me take care of my kids when they were babies.
Even right now I’m exploring new ways in which radar charts could be used to help teams adopt agile practices.
We need to recognise that charts are a powerful tool for interpreting data but they only show one perspective of that data. And that perspective may say more about the biases of the author, than the data itself.
This has nothing to do with the rest of my blog post. I just think it’s a great nerd joke and more people need to see it.