Looking out for Bad Practices:
On January 28th 1986 the Challenger Shuttle launch ended in disaster, killing all seven crew members and representing a tragedy for their families, NASA, and the nation. While some might know that the cause of this event was the failure of the O-ring, a gasket that sealed the right booster, many are unaware of the graph that led engineers to make the fateful decision to go for launch.
This graph, and the others depicted in the timeline created, all serve as examples of how bad data visualization can lead to dangerous results. At the heart of it, studying user experience design is a practice of looking out for bad practices and all the ways in which design can create dangerous or misleading circumstances for its users. In this assignment, I wanted to extend this practice in order to understand the consequences of bad data visualizations and provide a walkthrough of intentionally/unintentionally dangerous data visualizations.
Creating the Timeline – Materials Used:
In order to create the timeline linked below, the interactive tool TimelineJS was used in conjunction with Google Sheets. A template was provided and customized in order to create a custom timeline. While code, text, images, video, and sound could be input, text and photo links were the sources primarily used. After the Google Sheet was customized, the file was published to the web and a copy of the URL was input into TimelineJS where a link could be embedded or shared.
The design of the template was kept minimal in order to help the data visualizations stand out. A theme was created and kept consistent throughout the timeline in order to create a comprehensive narrative. When possible, corrected versions of the data visualizations were presented in order to help the viewer understand how they could be improved.
In total, five data visualizations were selected to be included in the timeline under the criteria that dangerous decisions could possibly be made should an individual interpret the data visualization without a critical lens.
- The Challenger Disaster (1986)
- Purdue Pharma OxyContin Marketing Material (1995)
- Florida Gun Deaths (2014)
- Global Temperature Change Graph (2015)
- Covid-19 Fatality Rates (2020)
Interpreting the Visualizations
Below are representations of the data visualizations used throughout this timeline. Below, each of these types will be discussed and compared with their improved versions.
1. Index Chart
Index charts were the most popular form of data visualization misused in the timeline created. While the way these graphs were misused varied, both charts appeared convincing to the untrained eye and represented instances of misinterpretation.
The graphs below represent the concentration of OxyContin in the bloodstream over time. The two graphs differ in an important way however. In the graph on the left, the graph produced by Purdue Pharma as part of their marketing package, a logarithmic y-axis is presented, making the rate of change look smaller than it actually was. When compared with the graph on the right, a linear y-axis, it is clear how this graph was used to misinform doctors of the addictive qualities of OxyContin.
2. Pie Charts
Pie charts in small multiples were used in order to help mislead the public about the dangers COVID-19 presented. In the example below, these pie charts were organized in such a way as to compare the fatality rates of COVID-19 and the seasonal flu and contrast these results against the fatality of SARS and MERS. One consequence of using these pie charts is that they don’t illustrate change over time and were used to minimize the severity of COVID-19.
3. Scatter Plots
The scatter plot below, which contributed to the Challenger Shuttle disaster, represents an issue with missing data. The graph on the left helped support the decision to launch since no direct correlation between temperature and O-ring failure was seen. However, if the entire data set had been included in the scatter plot, as seen in the graph on the right, NASA engineers would have been able to quickly realize that the O-ring was only successful when exposed to temperatures above 65°.
Takeaways
All visualizations presented in the timeline illustrate the power of data visualization to persuade an audience of a particular reality. Without substantial data literacy, it is easy to see how convincing figures and visualizations can be and how important it is to first look critically. The following are takeaways I’ll be taking forward in order to test the validity of data visualizations:
- Is only one side of the data being shown? If the data is only representing a particular case (i.e. failures) where can one access the positive instances in order to read the full story.
- Is there a good reason for standards to not be followed? If the author of the data visualization has decided to present the visualization in a logarithmic scale rather than a linear one, what advantage do they gain?
- Are the stylistic choices impacting my ability to read the data visualization?
- What scale is this data representing? Is there a good reason for the scale of this data to be larger than the actual rate of change?
- How can groupings influence the message of the data visualization: If the data is grouped amongst similar datasets is there a good reason, or is the only reason to minimize the message.
The final takeaway, and perhaps the most important is the following:
Does the visualization align with common sense?
By first answering this question, data visualizations can be assessed for their merits and their message can then be taken into consideration.
Personal Reflection
This exercise was a good introduction to the world of data visualization — the tool was simple to use and put the focus where it should be– on the data itself. If I were to do this assignment again, I would pick a topic that lent itself to the timeline format a bit easier as it was difficult to weave a cohesive story out of the individual timeline points I decided to focus on.