A bar chart with a distorted vertical axis isn’t very unusual. But what if that chart was posted by the White House and what if it was done on purpose – not to overstate the number shown, but rather to evoke a particular kind of response?
Last week, the Biden White House published this chart showing that U.S. GDP had grown by 5.7% in 2021. That’s the most in many years, with the chart showing numbers all the way back to 2001. Now one might argue that perhaps a new president won’t immediately have a huge impact on GDP, but the data in the chart appears to be legit. What is unfortunate, however, is that the vertical axis is being distorted.
People on Twitter quickly picked up on the fact that the axis scale goes up in whole percentage points until it reaches 5.0, but then there’s an additional 5.5 step before we get to the 6.0 value that should have been next. And since the growth number for 2021 sits between 5.5 and 6.0, that bar gets longer relative to the others.
A few days later, a correction was tweeted saying, “this is y you proofread.” The new chart has a correct vertical axis and brings the 2021 bar down to where it should be relative to the others.
Was this an honest mistake or was it designed to get more attention? And was the idea to overstate the change or get attention for the stretched axis?
The way the bars are spaced out and the choice of time window was clearly done to emphasize the 2021 value. While none of the bars even reach 4% in this chart, there were a number of years in the 1980s and 1990s with GDP growth over 4% (see this NY Times story with a different take on the same chart). Picking a timeline to fit a narrative is nothing new, and it’s not necessarily nefarious. The Obama administration did it to produce a very interesting chart showing job losses that I also discussed in a video. But that one did not distort any of the axes.
It’s certainly possible to make mistakes when designing charts like these, but I don’t believe that this was a mistake. This chart was probably not created in a visualization tool, but in Illustrator or a similar design program. If this were just a matter of the scale being wrong with an extra 5.5 gridline in there, that would be a reasonable mistake. But how did the final bar get stretched to fit into that scale correctly as well? That seems improbable.
It seems more likely that this was done on purpose, and there are two possible ways it might have been intended. The first and more obvious one is to simply stretch the bar to make the number look bigger and hope people won’t notice. That seems rather pedestrian, but it’s not like this hasn’t been done many times before. It does seem kind of futile though, since chart nerds will be quick to point out the issue, and it seems that there’s a low tolerance for bad and manipulated charts from official sources these days (remember “Sharpiegate?”).
The other theory is more interesting: what if this was intended to be noticed, to give it more exposure as people make fun of the mistake? And more than that, they would tsk-tsk the chart while pointing out that it wasn’t even necessary to distort the scale since the number was always going to be better than any in the 2000s so far anyway! That seems like a fun theory because it feels more like the kind of stratagem you’d expect from the kinds of people working in Washington.
Manipulating a chart is kind of boring, but the idea of somebody designing this as a sort of social engineering attack that would get people to discuss it in a particular way appeals to me. That doesn’t mean it’s true, of course, but it is a lot more fun than somebody trying to impress the boss by making a bar taller.