Updating the Famous Hannah Fairfield Connected Scatterplot
PolicyViz Newsletter #52
Hi all,
Back in 2012, Hannah Fairfield from the New York Times published this connected scatterplot that showed the evolution in the number of miles traveled versus deaths per 100,000 population.
So, in Fairfield’s graphic, the dot on the far left is 1950. Ten years later—near the bottom of that downward line—fatalities had declined but miles driven had increased. Ten years after that, in 1970, deaths and miles driven had both increased. By 2011, the very bottom dot in the bottom-right part of the graphic, deaths were markedly lower than they were 60 years earlier and miles driven had increased by about 6,000 miles per person.
Not everyone loves connected scatterplots. Some find them confusing or hard to read. Personally, I like them; yes, it can take a few minutes to fully understand what’s happening, but I like being able to see the co-movement of two series simultaneously. In a 2016 paper, Steve Haroz, Robert Kosara, and Steven Franconeri found that, in their sample, the connected scatterplot was the most difficult chart for people to read though were also the most effective at engaging readers.
The results suggest that low-complexity connected scatterplots can be understood with little explanation, and that viewers are biased towards inspecting connected scatterplots over the more traditional format.
But I do find that I create connected scatterplots I usually end up in one of two spots: Either the line is basically a straight diagonal because the trends are nearly perfectly correlated (e.g., program participation and spending) or the trends generate a confusing hairball.
But I love the Fairfield graphic. It’s a great example of data working out perfectly to make a clear and interesting graphic. So, what happens when we update it
Pretty cool, but the post-2010 period is a bit of a mess—and the 2020 COVID year is clearly an outlier. Let’s zoom in.
Okay, not that much more interesting. I do wonder what Fairfield would have done if she were publishing this graph now. Is the 2012-2016 jumble too confusing or would she add a little annotation telling us not to worry about it?
In my update of the graphic, the vehicle miles data I pulled from the St. Louis Federal Reserve Economic Data (FRED) data site didn’t match Fairfield’s original (fatalities matched). Steve Haroz had provided me the data several years ago when I was interested in using it in my data visualization classes. As you can see in this chart, the 2024 FRED series is pretty much a parallel upward shift from the Fairfield data.
What to do? Well, for this example, I calculated the average difference in the two series in the 1970-2011 period (equal to 986 miles) and subtracted it from each year in the post-2011 period (see next graph). I then used the Fairfield/Haroz series in the 1950-2011 period and the FRED-adjusted numbers in the 2012-2022 period.
It’s not a perfect solution to the mismatch, but it enables me to update the original graphic. If I were doing an in-depth analysis for the Department of Transportation or something, I would spend more time trying to figure out what’s going on, but I think my adjustment works well here.
Thanks,
Jon
Podcast: Charting New Horizons: Amanda Makulec on Leadership, Community, and the Human Touch Behind DataViz
Amanda Makulec is the current Executive Director of the Data Visualization Society (DVS), and in this week’s episode of the PolicyViz Podcast, we discuss her journey and the DVS’s evolution as it approaches its fifth anniversary. Amanda shares her experience starting as a volunteer all the way to leading the entire organization. With her second term coming to an end, she emphasizes the importance of term limits and her commitment to ensuring the organization’s sustainability by focusing on operational systems, finances, compliance, and community responsiveness.
Things I’m Reading & Watching
Books
Teaching Accessible Computing, edited by Alannah Oleson, Amy J. Ko, Richard Ladner
Practical Charts by Nick Desbarats
Data By Design by Lauren Klein and others [I think this is still in draft]
Articles
A Tool for Equity in Community Engagement and Collaboration, Daly et al.
An Opportunity for the Census Bureau to More Accurately Estimate the Disabled Population in the US, Hermans et al.
Data Visualizations
What would have happened to friends and family if Gaza was home? from Washington Post
A sports stadium boom is coming to America. Is that a good thing? from Washington Post
Enabling Circular Economy in Used Water Management in India from Council on Energy, Environment and Water
TV, Movies, Music, and Miscellaneous
Foxcatcher, Hulu
Three Body Problem, Netflix
The Regime, Hulu
Note: As an Amazon Associate I earn from qualifying purchases.
Sponsor: Ant Design System for Figma
Design and develop Ant Design projects faster than ever! The powerful UI kit for Figma based on the most popular React UI library - Ant Design. Create and implement well-documented Ant Design apps in no time!
Variables, Auto Layout, Variants, Component Properties
Dedicated Figma Plugin to copy style settings from Figma to code
Hover and click interactions for easy prototyping
Switch between light and dark theme
Developer-friendly components
