Data Visualization as Critical Making

Examples of Data Visualization, made with Canva

Visuals like these have become a part of our everyday lives. If you clicked into the most recent COVID-19 article, chances are it will have some sort of data visualization of the number of people sick, how that compares to the past month, and what percentage have already been vaccinated.

Due to the rise of big data and improvements in computational power, graphs and charts like these are no longer time-consuming to produce, because we can have machines perform most of the tedious tasks for us. Therefore, they have become the new standard. Just like “pics or it didn’t happen”, if you want to tell someone that something is true or if you want someone to understand how important an issue is, you should have the numbers to back it up. As a society, we have been trained by graphs and charts to develop trust in numbers and statistics and understand the severity of something by how it compares to a reference point. For example, when the NOAA and NASA report annual global temperatures, they focus not on the temperature itself but on how they compare to the 20^th-century average.

However, the way that numbers are presented can be more nuanced than we realize. If somebody told you that back in 1991, 215 out of every 100,000 cancer patients passed away, but in 2019, 146 out of every 100,000 cancer patients passed away, would you have a good feel for what that means? What if they told you that the overall cancer death rate has dropped by about a third from 1991 to 2019? Or, if they added that this means we have averted about 3.5 million deaths due to cancer during that time period? Immediately, you have a much better understanding of the significant progress cancer research has made.

Here is another example, this time from the CDC:

Source: “When Data Visualization Really Isn’t Useful (and When It Is)” by Christopher Berry, May 11, 2021 on Old Street Solutions

At first glance, it seems to make sense. The states in darker colors are the states that have the most reported cases of COVID-19. Instinctively, we understand that the darker a color is, the higher the number it represents. However, if you look closely, the coloring is all out of order. States that have no reported cases are in a darker shade than states that have 1 to 100 cases. States that have the most cases (10,001 or more) are in a light yellow. It is also very difficult to distinguish states that have no cases and states that have 101 to 1,000 cases, because they are in a similar shade of orange. This is a good example of how data can be 100% accurate, but still lead the audience to the wrong conclusion because of a simple mismatch of colors.

I think data visualizations are examples of critical making and new media because they harness technology to extend our ability to share with others our own interpretations of data. They transform numbers into something that humans can interact with. Without visualization and interpretation, data is unintelligible and therefore meaningless. However, like in the example above, we must think critically about how best to present the information to an audience that wants to glean the most out of a visual from just a quick glance. Misleading information that produces the wrong conclusion is worse than no information. Also, not every correlation is meaningful, and not everything that data proves is logical. For example, hot weather may increase ice cream sales and the risk of sunburn, which statistically correlates ice cream sales to the risk of sunburn, but we realize that this cannot be logistically true.

Source: Data Viz Project

With the growth of big data and artificial intelligence, we will only have increasingly more data points to tamper with. Data visualizations encourage us to rethink what critical making means in 2022, because the message is entirely within the medium. Traditional bar graphs and pie charts are suitable for very specific data interpretations, allowing us to see simple trends and percentages. Less conventional data visualization methods have gained popularity in the last few years, due to their ability to convey meaning in ways that traditional statistical charts can no longer compete with. Alluvial diagrams represent changes in the composition of something over time. Hive plots highlight how well something can satisfy a set of criteria. Radial histograms make it easy to display more data bars without overwhelming the reader. However, each form also comes with its own disadvantages. It takes thoughtful implementation to ensure that the form of visualization does not get in the way of interpreting the data, and complements the message it is intended to convey.

Source: “The Unwelcomed” project on ALHADAQA by Mohamad Waked

Data visualizations can also be a part of nonviolent-yet-disruptive protests, similar to electronic civil disobedience, because numbers help to justify and raise morale for activism. Websites such as the Pew Research Center and ProPublica routinely employ data visualizations to unveil important findings on social issues such as gun violence and environmental pollution. You also have independent data visualization designers such as Mohamad Waked, who started his own data visualization lab. One of his projects was an online interactive story based on the number of migrants and refugees that have passed away while crossing borders. His data visualization map helps to put into perspective the significance of the issue around the world.

Although data visualizations are not usually thought of as an art form, I would argue that it can be very powerful because it has the potential to tell true stories with a single visual. Just as a poet chooses their rhyme scheme and meter, a data visualization designer carefully chooses the form of visualization that enables their data to speak for itself. Misunderstanding can lead to false insights and poor decision-making, all under the assumption that it was backed by real data.

Isabella Wang, Feb. 7 2022

VMS 290S Spring 2022