How to Lie with Charts

In this section, you’ll learn how to avoid being fooled by misleading charts, and also how to make your own charts more honest, by intentionally manipulating the same data to tell opposing stories. In the first half, you will exaggerate small differences in a column chart to make them seem larger. In the second, half, you will diminish the rate of growth in a line chart to make it appear more gradual. Together, these tutorials will teach you to watch out for key details when reading other people’s charts, such as the vertical axis and aspect ratio. Paradoxically, by demonstrating how to lie, our goal is to teach you to tell the truth and to think more carefully about the ethics of designing your data stories.

In the first half of this tutorial, we’ll examine data about the economy, a topic that’s often twisted by politicians to portray it more favorably for their perspective. The Gross Domestic Product (GDP) measures the market value of the final goods and services produced in a nation, which many economists consider to be the primary indicator of economic health. (Interestingly, not everyone agrees because GDP does not count unpaid household labor such as caring for one’s children, nor does it consider the distribution of wealth across a nation’s population.) We downloaded US GDP data from the US Federal Reserve open-data repository, which is measured in billions of dollars and published quarterly, with seasonal adjustments to allow for better comparisons across industries that vary during the year, such as summer-time farming and tourism versus Christmas-time gift shopping. Your task is create a misleading column chart that exaggerates small differences to make them appear larger in the reader’s eye.

  1. Open the US GDP mid-2019 data in Google Sheets, and go to File > Save As to create a copy that you can edit in your own Google Drive. We’ll create charts in Google Sheets, but you can also download the data to use in a different chart tool if you prefer.

  2. Examine the data and read the notes. To simplify this example, we show only two figures: the US GDP for the 2nd quarter (April-June) and the 3rd quarter (July-September) in 2019. The 2nd quarter was about $21.5 trillion, and the third quarter was slightly higher at $21.7 trillion. TODO: DECIDE whether to add this: If we calculated the percent change, or (21747 - 21540)/21540 = 0.0096 = 0.96%, that’s a growth rate of just under 1 percent.

  3. Create a Google Sheets column chart in the same sheet using the default settings, although we never automatically accept them as the best representation of the truth. In the data tab, select the two columns, and go to Insert > Chart, as you learned when we introduced charts with Google Sheets in Chapter 7. The tool should recognize your data and automatically produce a column chart, as shown in the left side of Figure 15.1. In this default view, with the zero baseline for the vertical axis, the difference between $21.5 versus $21.7 trillion looks relatively small to the reader.

  4. Reduce the vertical axis. Click on the three-dot kebab menu to open the Chart editor and select the Customize tab. Scroll down to the vertical axis settings, and change the minimum to 21500 and the maximum to 21800, as shown in the right side of Figure 15.1. Although the data remains the same, those small differences now appear much larger in our eyes because you’ve manipulated the vertical scale. Only people who read charts closely will notice this trick. The political candidate who’s campaigning on rising economic growth will thank you!

The default GDP line chart on the left, and the reduced-axis chart on the right.

Figure 15.1: The default GDP line chart on the left, and the reduced-axis chart on the right.

TODO: Placed images above side-by-side to allow for easier comparison. Rethink if “zero baseline” and “truncated baseline” are best titles.

However, your chart is clearly wrong because you’ve violated one of the cardinal rules about chart design in Chapter 7. Column (and bar) charts must start at the zero baseline, because they represent value using height (and length). Readers cannot determine if a column is twice as high as another column unless both begin at the zero baseline. By contrast, the default chart with the zero baseline is truthful. But let’s move on to a different example where the rules are not as clear.

In the second half of this tutorial, we’ll examine data about climate change, one of the most pressing issues we face on our planet, yet deniers continue to resist the new reality, and some of them twist the facts.[TODO: Cite Cairo examples]. In this tutorial, we’ll examine global temperature data from 1880 to the present, downloaded from the NASA, the US National Aeronautics and Space Administration. It shows that the mean global temperature has risen about 1 degree Celsius (or about 2 degrees Fahrenheit) during the past fifty years, and this warming has already begun to cause glacial melt and rising sea levels. Your task is to create misleading line charts that diminish the appearance of rising global temperature change in the reader’s eye.23

  1. Open the global temperature change 1880-2019 data in Google Sheets, and go to File > Save As to create a copy that you can edit in your own Google Drive.

  2. Examine the data and read the notes. Temperature change refers to the mean global land-ocean surface temperature in degrees Celsius, estimated from many samples around the earth, relative to the temperature in 1951-1980, about 14°C (or 57°F). In other words, the 0.98 value for 2019 means that global temperatures were about 1°C above normal that year. Scientists define the 1951-80 period as “normal” based on standards from NASA and the US National Weather Service, and also because it’s a familiar reference for many of today’s adults who grew up during those decades. While there’s other ways to measure temperature change, this data from NASA’s Goddard Institute for Space Studies (NASA/GISS) is generally consistent with data compiled by other scientists at the Climatic Research Unit and the National Oceanic and Atmospheric Administration (NOAA).

  3. Create a Google Sheets line chart in the data tab by selecting the two columns in the data tab, then Insert > Chart. The tool should recognize your time-series data and produce a default line chart, though we never automatically accept it as the best representation of the truth. Click on the three-dot kebab menu to open the Chart editor and select the Customize tab. Add a better title and vertical axis label, using the notes to clarify the source and how temperature change is measured, as shown in Figure 15.2.

Figure 15.2: Default line chart of global temperature change. Explore the interactive version.

Now let’s create three more charts using the same data but different methods, and discuss why they are not wrong from a technical perspective, but nevertheless very misleading.

Lengthen the vertical axis to flatten the line

We’ll use the same method as shown in the first half of this tutorial, but in the opposite direction. In the Google Sheets chart editor, customize the vertical axis by changing the minimum value to negative 5 and the maximum to positive 5, as shown in Figure 15.3. By increasing the length of the vertical scale, you flattened our perception of the rising line, and cancelled our climate emergency…but not really.

Misleading chart with a lengthened vertical axis.

Figure 15.3: Misleading chart with a lengthened vertical axis.

What makes this flattened line chart misleading rather than wrong? In the first half of the tutorial, when you reduced the vertical axis of the US GDP chart, you violated the zero-baseline rule, because column and bar charts must begin at zero since they require readers to judge height and length, as described in the chart design section of Chapter 7. But you may be surprised to learn that the zero-baseline rule does not apply to line charts. Visualization expert Albert Cairo reminds us that line charts represent values in the position and angle of the line. Readers interpret the meaning of line charts by their shape, rather than their height, so the baseline is irrelevant. Therefore, flattening the line chart for temperature change may mislead readers, but it’s technically not wrong, as long as it is labelled correctly.24

Widen the chart to warp its aspect ratio

In your Google Sheet, click the chart and drag the sides to make it very short and wide, as shown in Figure 15.4. Image measurements as listed in width by height, and we calculate the aspect ratio as width divided by height. Since the default chart is 600 x 370 pixels, its aspect ratio is about 1.6 to 1. But the stretched-out chart is 1090 x 191 pixels, and its ratio is about 5.7 to 1. By increasing the aspect ratio, you have flattened our perception of the rising line, and cancelled our climate crisis once again…but not really.

Misleading chart with a stretched aspect ratio.

Figure 15.4: Misleading chart with a stretched aspect ratio.

What makes this warped line chart misleading rather than wrong? Once again, since changing the aspect ratio of a line chart does not violate a clearly-defined rule of data visualization, it’s not technically wrong, as long as it’s accurately labeled. But it’s definitely misleading. Visualization expert Alberto Cairo states that ideally, we should design charts with aspect ratios that “neither exaggerates nor minimizes change.” What specifically does he suggest? Cairo recommends (but does not propose a universal rule) that the percent change expressed in a chart should roughly match its aspect ratio. For example, if a chart represents a 33 percent increase, which is the same as 33/100 or 1/3, he recommends an aspect ratio of 3:1 (because the fraction is flipped by placing width before height), or in other words, a chart that is three times taller than its width.25 Therefore, if we apply Cairo’s recommendation to our climate change chart, the difference from 0° to 1°C represents a 100% increase, which suggests an ideal chart with a 1:1 aspect ratio, or just as tall as it is wide, as shown in Figure 15.5.

Cairo’s recommendation for percent change (100%) to match chart aspect ratio (1:1).

Figure 15.5: Cairo’s recommendation for percent change (100%) to match chart aspect ratio (1:1).

However, Cairo clearly states that his aspect ratio recommendation “isn’t a universal rule of chart design” and there are several cases when you should ignore it and use your own judgment. For example, instead of global temperature change, which increased from 0° to 1°C, imagine that our chart displayed the global temperature, which increased from about 13° to 14°C (or about 55° to 57°F). When we express the temperature in absolute numbers, it doesn’t feel very significant, even though a 1°C change in average temperature can have dramatic global consequences. Using this different scale, the chart would represent only an 8 percent increase, or about 1/12, which under Cairo’s recommendation translates into a 12:1 aspect ratio, or twelve times wider than it is tall, as shown in Figure 15.6.

If we apply Cairo’s recommendation to a chart of global temperature (not temperature change), where percent change (8% or 1/12) matches its aspect ratio (12:1), the result is misleading.

Figure 15.6: If we apply Cairo’s recommendation to a chart of global temperature (not temperature change), where percent change (8% or 1/12) matches its aspect ratio (12:1), the result is misleading.

Even Cairo points out that this significant temperature increase looks “deceptively small” if you follow his aspect ratio recommendation using this scale, so he advises against it.26 Furthermore, if you convert the scale from Celsius to Fahrenheit, the calculation changes once again, which doesn’t make any sense. Where does all of this leave us? If you feel confused, that’s because there’s no universal rule with aspect ratio. What should you do? First, never automatically accept the default chart. Second, explore how different aspect ratios affect its appearance. Finally, use your best judgement with aspect ratio to tell true and meaningful data stories, because there is no single rule that fits all cases.

TODO: Need feedback on whether the 2nd and 3rd charts and text make sense here

Add more data and a dual vertical axis

Let’s add more data to make your chart even more misleading! In the Google Sheet, go to the tab named temp+GDP, where you will see temperature change plus a new column: US Gross Domestic Product (GDP) in billions of dollars from 1929 to 2019, downloaded from the US Federal Reserve. To simplify this example, we deleted pre-1929 temperature data to match it up more neatly with available GDP data.

  1. Select all three columns and Insert > Chart to produce a default line chart with two data series: temperature (in blue) and US GDP (in red).

  2. In the Chart editor, select Customize and scroll down to Series. Change the drop-down menu from Apply to all series to US GDP. Just below that in the Format area, change the Axis menu from Left axis to Right Axis, which creates another vertical axis on the right side of the chart, connected only to the US GDP data, as shown in Figure 15.7.

Add another vertical axis to the right side of the chart.

Figure 15.7: Add another vertical axis to the right side of the chart.

  1. In the Chart editor > Customize tab, scroll down and you will now see separate controls for Vertical Axis (the left side, for temperature change only), and a brand-new menu for the Right Axis (for US GDP only), as shown in Figure 15.8.
Brand-new menu for the right axis.

Figure 15.8: Brand-new menu for the right axis.

  1. Finish your chart by adjusting Vertical Axis for temperature change, but with even more exaggeration than you did in step 5 above. This time, change the minimum value to 0 (to match the right-axis baseline for US GDP) and the maximum to 10, to flatten the temperature line even further. Add a title, source, and labels to make it look more authoritative, as shown in Figure 15.9. By lowering our perception of the temperature line in comparison to the steadily rising GDP line, you’ve misled us into ignoring the consequences of climate change while we enjoy a long-term economic boom! Furthermore, the GDP data is not adjusted for inflation, so you’ve double-misled us by comparing 1929 dollars to 2019 dollars. Finally, since you accepted the default colors assigned by Google Sheets, the climate data is displayed in a “cool” blue, which sends our brain the opposite message of rising temperatures and glacial melt. So we’ll count this as a triple! [TODO: Decide whether to break non-comparable data and color into separate paragraphs.]
Misleading dual-axis chart of US GDP and global temperature change.

Figure 15.9: Misleading dual-axis chart of US GDP and global temperature change.

What makes this dual axis chart misleading rather than wrong? Once again, since it does not violate a clearly-defined visualization design rule, the chart is not wrong, but very misleading. In fact, many visualization experts strongly discourage using dual-axis charts because they confuse most readers and create mischief. Even though both axes began at zero, the left-side temperature scale has a top level of 10 degrees Celsius, which is not reasonable given the context of the data.

What’s a better alternative if you wish to visualize the relationship between global temperature and US GDP over time? Consider a scatter chart, which we introduced in chapter 7, because it works best to show the relationship between two variables by representing them as XY coordinates. Furthermore, make a more meaningful comparison by plotting global temperature change versus US real GDP that has been adjusted into constant dollars.

[TODO: if we like this idea, insert a scatter chart here as described above….]

To sum up, in this tutorial you created four charts about global temperature change. None of them were technically wrong, only some were truthful, while many were unreasonably manipulated to mislead readers by hiding or disguising important patterns in the data. Furthermore, there are additional ways to deceive that we did not examine here, such as building 3D charts and tilting the reader’s perspective below the baseline.27

You may feel strange that data visualization lacks clearly-defined design rules for many cases, like we are accustomed to reading in our math, science, or grammar textbooks. Instead, remember that the important visualization rule is a three-step process: never automatically accept the default, explore how different designs affect the appearance of your interpretation, and use your best judgement to tell true and meaningful data stories.

Now that you’ve learned about how to lie with charts, in the next section you’ll build on these skills to lie with maps.


  1. The inspiration for this specific tutorial came from a high school classroom activity created by the NASA Jet Propulsion Laboratory (JPL), as well as visualization expert Alberto Cairo’s analysis of charts by climate change deniers. NASA JPL; Cairo, How Charts Lie, pp. 65-67, 135-141.↩︎

  2. Cairo, p. 61.↩︎

  3. Cairo, p. 69.↩︎

  4. Cairo, p. 70.↩︎

  5. Cairo, p. 58.↩︎