Illness as indicator

Illness as indicator

Organisation: The Economist Newspaper (United Kingdom)

Publication Date: 04/11/2017


Size of team/newsroom:large


The victory of Donald Trump in the 2016 US presidential election was one of the most surprising and unlikely electoral outcomes in recent memory. In its aftermath numerous analyses were undertaken by news outlets and political scientists in an effort to better understand how Mr Trump won. As a data journalist at the Economist Newspaper my aim was to try to contribute to that discussion with some original and distinct analysis. In the first instance, I created a set of county-level demographic data using the American Community Survey’s 2014 micro data sample—matching geographies tor the results file and estimating demography for counties with fewer than 100,000 inhabitants (the cut-off for the Census Bureau data). Second, I ran regressions of the result against a set of demographic factors for each of America’s 4,000 counties. This demonstrated that education, age and race were all significant and powerful predictors of Republic swing: elements that were understood before the election thanks to pollsters’ microdata.

What makes this project innovative? What was its impact?

A statement on Twitter by Patrick Ruffini, a pollster, prompted me to probe further. He challenged people to “find the variable that can beat % of non-college whites in the electorate as a predictor of county swing to Trump”. A reporting trip to a Trump rally in Manheim, Pennsylvania in early October proved my inspiration. At the rally I observed that many of Mr Trump’s supporters were unable to stand long enough to listen to Mr Trump’s speech. Instead, they sat together at the very back of the sports hall. I took a photograph which I’ve attached as evidence. That observation gave me a hunch that ill health, despite Obamacare’s efforts to increase access to healthcare, was a likely factor in Mr Trump’s election victory. Using county-level health metrics from the Institute for Health Metrics and Evaluation at the University of Washington, I discovered that collectively, a set of health data (obesity, diabetes, heavy drinking and physical activity) explained 43% of the swing to the Republicans (thus beating the 41% explained by non-college whites). I presented the data with a simple scatter chart, showing the weighted index of health metrics and percentage swing to the Republicans, with regional colours. I think this simple illustration helped disseminate the idea. I believe that the article significantly contributed to helping understand the proximate causes of Mr Trump’s victory. The article has been cited by policymakers, academics and health practitioners in America and Britain.

Technologies used for this project:

R for analysis and data harvesting; D3 for interactive visualisation; Adobe for static print visualisation
Follow this project

Comments (0)

You have to be connected to contribute

You have to be connected to follow

Leave this project and no longer be informed about this project

By joining this project, you will be informed by email when an update or a new contribution is posted on the website.

Thank you for your active participation !

The GEN Community Team