Misleading election campaigns? Let’s create more awareness

In the last couple of weeks, the entire world was following the presidential elections of the United States of 2020. The Democratic candidate Joe Biden was running against the current president of the U.S., Donald Trump, and he won the elections. The weeks building up to the elections, many different graphs and maps were shared by news- and media channels, mostly by people on social media. However, many of these visualizations could be misleading, and give a poor representation of reality. In this blog, I would like to show some examples of misleading data visualizations used in the presidential elections, and argue why they are misleading. 

Putting the reader on the wrong track
As many of you know, information and data visualizations can be misleading. But what does this mean? According to Cairo (2015), it means that any type of visualization of data can put the reader on the wrong track, without conscious intervention of the designer. This could happen intentionally, or as the result of naïve mistakes from the designer. In this last case, it does not mean that a designer’s obligation of being truthful is violated. 

How do misleading data visualizations look like? Let’s take a closer look at some examples of the elections in the US – both from 2016 and 2020.

Mapping the US: is it mostly republican?
Firstly, take a look at this map, initially tweeted by Lara Trump. In this tweet, she posted a map including the statement “try to impeach this”. With this visualization, she suggested that the US is almost entirely Republican, with only a few places being Democratic. 

Tweet from @LaraLeaTrump on September 28th, 2019 | Twitter

However, Karim Douïeb – co-founder of a data science company – picked up this tweet and wanted to change the map, since he thought it was a misrepresentation of the truth. He created a new visual of how the election map should look like. A transition emerged between the first map and a new map, which is showed below.

Tweet from @karim_douieb on October 9th, 2020 | Twitter

Obviously, the first map did not show an accurate representation of the data. It showed large areas in the US where only a small amount of people live, whereas the big cities – who are mostly Democratic – were not represented well enough. Therefore, this is a clear example of how data visualization can mislead: Lara Trump showed a graph that was framed, which made it untruthful.

According to Harper (2004), viewing data graphically allows the reader to see a quick overview and the trend of the data. However, it could influence how people interpret it. Depending on the goal of the graph, sometimes presenting detail of different areas in a map (second map) might be more appropriate, because extreme simplification of data (first map) could bias people (Cairo, 2015).

Bar chart from Trump’s 2016 campaign: deceiving?
Another example of how data visualizations can mislead is the following bar chart, tweeted by Donald Trump himself on October 6th, 2016. According to Muyskens (2016), Trump has used over 40 bar charts in the elections of 2016, which showed misleading data.

Tweet from @realDonaldTrump on October 6th, 2016 | Twitter

In this bar chart no zero-baseline was added, causing a distortion of data. Therefore, the two bars cannot be compared with each other. This is also called y-axis truncation. Muyskens (2016), indicated that this bar chart visually shows that Trump’s lead was increased by 11 points, instead of the actual 2 points.

Let’s check if this bar chart is indeed misleading, using Tufte’s lie factor (1983). The closer the lie factor (size of effect shown in the graphic / size of effect in data), is to 1.0, the more accurate the graph is. Trump’s bar chart scored a 7,7 on the lie factor, indicating that this graph is indeed misleading. Thus, this is a clear example of how a bar chart is used in an inappropriate way.

According to Yang et al. (2020), who investigated the practice of truncating the y-axis of bar graphs to start at a non-zero value, it was found that 83,5% of all participants in five studies showed a truncation effect. Even people who were given a warning about the truncation effect beforehand, or people with substantial experience working with data and statistics, were still vulnerable towards the truncation effect.

The fact that people who have experience with data and statistics are still vulnerable towards a truncation effect, says a lot about what we should expect from common people. In this example, the graph was presented on Twitter, where a huge amount of people have access to. We cannot expect them to see behind the poor misrepresentation of data. 

Conclusion
Data visualizations are meant to communicate and persuade, but can also be oftentimes deceiving (Cairo, 2015; Yang et al., 2020). Whether it is with positive or negative intentions, people have to deal with a poor representation of what is actually happening. What can we do to deal with this in a better way?

Klein (2020), suggests that people in the field of science should better take into account the difference between levels of manipulation. Furthermore, it is important for scholars to see the difference between poor and well presented data (Klein, 2020). According to Cairo (2015), we could better prepare and educate those people who communicate data visualizations, and whose goal is to present truthful and accurate data visualizations.

In line with Cairo (2015) and Klein (2020), I firmly believe that people – especially those in the field of science, communication, and journalism – should be educated better, in order to spot misleading data visualization, and to prevent communicators of data visualizations to deceive and mislead common people. Because we do not want people to draw the wrong conclusions, based on misleading data visualizations, right?

To conclude, I am curious about your opinion. Therefore, I have some questions to you:

• Do you agree with the statement that people in the field of science, communication, and journalism should be educated better?
• How would those people be educated better, in order to spot misleading data, and to prevent communicators of data visualizations to deceive and mislead common people?

References:

Cairo, A. (2015). Graphics lies, misleading visuals: Reflections on the challenges and pitfalls of evidence-driven visual communication. In D. Bihanic (Ed.), New challenges for data design (pp. 103-116). Springer-Verlag, London

Harper, S. (2004). Students’ Interpretations of Misleading Graphs. Mathematics teaching in the middle school, 9(6), 340-343. Retrieved from https://www.researchgate.net/profile/Suzanne_Harper/publication/317570148_Students’_Interpretations_of_Misleading_Graphs/links/594037f4aca272371224d9df/Students-Interpretations-of-Misleading-Graphs.pdf

Klein, O. (2020). Misleading Memes. The effects of deceptive visuals of the British National Party. The Open Journal of Sociopolitical Studies, 1(13), 154-179. https://doi.org/10.1285/i20356609v13i1p154

Douieb, K. [karim_douieb]. (2019). Challenge accepted! Here is a transition between surface area of US counties and their associated population. #HowChartsLie #DataViz #d3js [Tweet]. Retrieved from https://twitter.com/karim_douieb/status/1181695687005745153?s=20

Muyskens, J. (2016). Most of Trump’s charts skew the data. And not always in his favor. Retrieved November 15, 2020, from https://www.washingtonpost.com/graphics/politics/2016-election/trump-charts/

Trump, L.L. [LaraLeaTrump]. (2019). Image: Try to impeach this [Tweet]. Retrieved from https://twitter.com/LaraLeaTrump/status/1178030815671980032?s=20  

Trump, D.J. [realDonalTrump]. (2016). Reuters polling just out- thank you! #MakeAmericaGreatAgain [Tweet]. Retrieved from https://twitter.com/realDonaldTrump/status/783808918824681472?s=20

Tufte, E. R., & Graves-Morris, P. R. (1983). The visual display of quantitative information (Vol. 2, No. 9). Cheshire, CT: Graphics press.

Yang, B., Vargas-Restrepo, C., Stanley, M., & Marsh, E. (2020). Truncating bar graphs persistently misleads viewers. Journal of Applied Memory and Cognition, in press.