How To Lie With Charts

Reading Time: 11 Minutes

Some of the best examples of data visualization today can be found in journalism. In recent years many if not most global news networks have embraced empirical storytelling and have made significant investments in both the people and technology needed to tell compelling visual stories powered by data. 

The New York Times and Bloomberg are two great examples. The Times was an early player in the data journalism and visualization space, and today the NYT Graphics Department have produced some of the most innovative and stunning interactive data graphics around. I have more than a few favourites, but their recent graphic that visualizes the history of Olympic Games medal wins by country is amazing!

Bloomberg, on the other hand, has focused more on making data visualization part of their DNA. From the famed Bloomberg Terminal, to the ways in which they visualize stock data throughout their website and app, to on-air segments like Joe Weisenthal's Killer Chart, visualizing data in clear and creative ways has become part of Bloomberg's core offering. 

Unfortunately, Fox News seem to have missed the memo on visual storytelling entirely. That's not to say they don't produce their share of charts and data graphics for on-air segments. They do, they're just really bad at it.

Case in point. Here's a fun little graphic produced by Fox for a recent segment they aired on the U.S. economy.

You may look at this chart and say, 'hey, the U.S. economy's rate of growth sure has slowed since Obama took office'. I mean, just look at that bar chart. The average rate of economic growth during Obama's presidency is less than half the rate of growth the country enjoyed during the previous time frame. Clearly Obama's presidency has been an economic disaster, right?

Well no, not exactly. This chart is both poorly designed and quite deceiving, as it strips away important context that you would need to meaningfully interpret the data and what it means. Let’s break down why exactly this chart is such a flop.

1 - Too much non-data ink

Edward Tufte introduced the concept of data-ink in his book, The Visual Display of Quantitative Information. When it came to defining the concept, Tufte had the following to say (emphasis mine):

“A large share of ink on a graphic should present data-information, the ink changing as the data change. Data-ink is the non-erasable core of a graphic, the non-redundant ink arranged in response to variation in the numbers represented”.

If data-ink represents the non-erasable, or essential elements of a chart, non-data-ink (or what Tufte calls the data-ink ratio) represents any unnecessary textual and graphical elements and formatting that might distract or impede the viewer’s interpretation of the data.

Now if Tufte had devoted an entire chapter in his book to showcasing examples of bad charts with too much non-data ink, I'd like to think it would be chock full of Fox News graphics. Just take a look for yourself.

Fox News Charts

My eyes are burning!

What the folks over at the Fox News graphics department don't seem to understand is that, when it comes to visualizing data, you can actually say a lot more with less. Their graphics are often loaded with a dizzying amount of decoration and graphical formatting that make it hard to focus on what the data is actually trying to say.

Joey Cherdarchuk of Darkhorse Analytics wrote a great post about this very topic a few years back. His article covered Tufte's data-ink concept and explored how you can make data graphics more effective by stripping away unnecessary graphical fat. To illustrate his point, Cherdarchuk created a wonderful animation (below) that demonstrates how removing unnecessary formatting and clutter can make your graphics more impactful and easier to interpret.

Joey Cherdarchuk Data-Ink Animation

He also created a Speaker Deck version of his animation, which you can find here.

2 - Poor chart labeling

Another issue with the Fox economy chart is that the y-axis has not been declared. It’s always good practice to use descriptive axis labels so your audience knows exactly what data is being shown and how it is quantified.

But beyond the y-axis labeling issue this chart doesn't do a good job of communicating what data is being represented. The chart title simply states "Average Annual Economic Growth" without clearly indicating the underlying data set or metric used. Sure, savvy viewers might know that GDP is a standard measure of economic growth, but there are many different ways to quantify economic growth (e.g. GNP, MFP, to name a few). So it's still important to declare this information.

And the reality is that, when it comes to mainstream news networks like Fox, their audience is broad enough that most viewers probably won’t know how economic growth is measured. Sure, it's not Fox's responsibility to teach their viewers economics, but they should still equip them with the information and data they need to make up their own mind.

When creating a chart you should always be mindful of the audience it is being created for and ensure that you include all the relevant information they would need in order understand and interpret your graphic. This generally involves using descriptive and meaningful labels for your chart title, axis labels and legend.

3 - Unequal time intervals

The two issues mentioned above have more to do with basic chart formatting rules that, if followed, can help ensure your charts are easily understood. However, the biggest problem with Fox's economy chart is not with how it's formatted, but with its deception.

Now I didn't actually see the segment this graphic aired on, but I don't need to to know what they were trying to say - that Obama's presidency has been a disaster for the country's economy. That conclusion, or bias, is baked right into the graphic for all to see.

The graphic shows the difference in annual economic growth as a percent change between two distinct time intervals. The first time interval covers 1950 to 2008 while the second time interval covers 2009-2016 (i.e. Obama's presidency).

So what's the problem? Well first of all they're comparing two time intervals that aren't equal. The first calculates the average annual GDP growth for a 58 year period while the second interval only covers 7 years. This isn’t fair, as there are 51 more data points (i.e years of data) that contributed to the average for the first interval. 

When displaying time-based data, it’s generally best practice to use time intervals that are equal and adjacent (i.e. month over month, quarter over quarter, year over year, etc). I’ve pulled the full dataset from the U.S. Bureau of Economic Analysis, and this is what the data looks like when you use an annual time interval.

U.S. Annual Percent Change in Real GDP

This is how Fox should have presented the data, though they didn't necessarily have to go back as far as 1950. This actually stood out to me, as one of the first questions I had after looking at the chart was why did they need go so far back in time for the first interval? I'll tell you why. They did this because it helped support their narrative by creating the biggest possible margin to compare against Obama's average economic growth rate. I went through the BEA dataset, and if you calculate the average for every possible combination of years going backward from 2008 (e.g. 2008 + 07, 2008 + 07 + 06, etc), the range from 1950 to 2008 gives you the biggest possible figure. So what does this tell us about the people who made the chart? For one thing it tells us that they already knew what narrative they wanted to spin before they even looked at the data. And second, it tells us they the massaged the presentation of the data to fit that narrative, or bias. 

Working with meaningless data

Now around the same time I saw the Fox economy chart I came across another article from the NY Times that was attempting to do something similar - which is to calculate the average GDP growth for a presidential term. Now the author doesn’t actually visualize the data, though he does present it in a basic table (note. Some of the GDP values are outdated as the article was written some time ago):

The problem with both the Times column and Fox's chart is that GDP growth is not something that can or should be wholly attributed to a president. Can a president and his/her policies have an impact on economic growth, of course. But are these policies the only factor that can have an impact on the economy, absolutely not.

Let’s consider a few examples. Gerald Ford, the Republican president who took over the oval office after Nixon stepped down, was in office for just 2.5 years. This is hardly a fair comparison when it comes to two term presidents that had up to 8 years to craft their economic legacy. He was also building on an economic program and fiscal policies that were proposed and set in motion by Nixon, so it's even less fair to fully attribute any growth outcomes to Ford. Also, Ford’s own economic initiatives and policies were largely considered to be a joke. One notable effort by his administration was an initiative to "tame runaway price increases by urging Americans to wear round red lapel pins emblazoned with the initials WIN, for "whip inflation now." It didn’t work. 

Take Obama’s presidency as another example. The full effect of the 2008 housing crisis hit the books by 2009, resulting in the biggest economic contraction since the Great Depression (-2.8% growth). The 2008 financial crisis had nothing to do with Obama or his policies, yet this still pulls down the average growth figure for his term, substantially. The fact is there are many other variables that affect economic growth outside of the presiding administration and their policies, such as a recession or the actions of the Federal Reserve.

Using notation to tell your story

Suffice it to say, calculating the average GDP per president isn’t a meaningful exercise. However, I do think you can still convey this information in a chart without jeopardizing the quality of the data. The major problem with the Fox chart was that they repackaged the data in way that made it impossible for the viewer to see the full picture (the 58 and 7 year intervals). However, you could still visualize the GDP data by president using notation. Here's an example of what that could look like:

Click the image for Larger Version

With this example, the president, their political party and the duration of their presidential term are all included in the chart. But more importantly the full data set is still intact, we avoid calculating a figure that is potentially dubious (i.e. average GDP growth per president), and it includes enough context and granularity for the viewer to make their own judgments. 

The 'New Normal' of economic growth

There’s even more context to consider on this topic before making a final judgment about a specific president’s impact on the economy. Which is that lower rates of economic growth in the US (and across the globe) are fast becoming the “new normal”.  Many economists now believe that the days of seeing 3% annual growth in the economy are long gone, and that the new benchmark for annual growth going forward is actually closer to 2%, or even lower. There's a great deal of debate about why and how this is happening but I’ll spare you the economics lesson for now. But if you do want to learn more you can read about it here, here, here, here and here.

The point is, showing a chart on the economy and trying to pin slower rates of growth on one presidency is downright deceptive. Yet many people will still watch this segment on Fox News and believe every word. This chart and everything Fox want it to say will become the viewer’s reality. After all, the underlying data source is credible, and data never lies right? I wish that were true, but data can be used to lie or mislead when packaged and presented in certain ways. It's graphics like this that breed ignorance and narrow minded thinking. It fuels a false narrative and view of the world that make it difficult to have meaningful and constructive conversations about important issues we face and how to fix them.

And if you think this is just some harmless chart, and that most people will be smart enough to see through its deception, think again. After Donald Trump tweeted the graphic to his followers it's now gained more than 26K likes and 11K retweets. Yikes.

I like to think that Fox News will eventually realize they need to grow up and stop pulling these tricks. But truth be told, I just don’t think they're that concerned with empowering their viewers with a fair and balanced view of the world.

And to be honest, this isn’t a problem that's limited to Fox News (though they are a serial offender). I’ve seen poorly constructed and deceptive data graphics featured on many major global news networks. Here's NBC failing at basic chart formatting in the best way:

So the next time you’re watching the news and you see a data graphic or chart, don't take it at face value. Take a closer look, be critical of what you see, and decide for yourself what's true and what's not.