Skip to main content

Database Reporting


Derek Willis, news applications developer at ProPublica , formerly of The Upshot at The New York Times, served as the College’s second Innovator-in-Residence. Willis worked with students in Assistant Professor Alison Bass’ Investigative Reporting class and Assistant Professor Bob Britten’s Interactive Design and Data Visualization class to collect data about West Virginia voter behavior and compare it to the campaign messages residents receive.

Willis’ work focuses on elections and politics, congressional behavior, campaign finance and other topics. Formerly of The Upshot, Willis was a member of The Times’ Interactive News desk during this project, where he built and maintained political databases used in web applications.

Willis spent the Fall 2014 semester working with students to analyze data in the West Virginia Senate race between Republican Shelley Moore Capito and Democrat Natalie Tennant.

The project allowed students to gain a critical and highly marketable reporting skill – the ability to parse data and use it to inform their elections reporting – and was designed as a hyper-local model that could also be used to provide a big-picture analysis of an entire election.

“We hope to understand which areas are critical in an election,” Willis said. “Ideally, we’ll be able to see how one candidate or another can hold together a coalition of voters – or how they failed to do that.”

He stressed the importance of collecting this information at a “micro” level.

“If you don’t use this kind of data, you run the risk of missing a candidate’s approach,” he said. “Because of the ability to target smaller and smaller slices of the electorate, reporters can have a thorough understanding of what the campaign is doing.”

Students in the participating classes were divided into teams to cover five West Virginia counties: Kanawha, Monongalia, Preston, Harrison and Marshall. They identified voters in each area and interviewed them about whether they had recently shifted their political affiliation, what issues mattered to them, and which political campaign and party messages they received. Then, the students analyzed the effectiveness of messaging on a highly targeted level to understand how it impacts voter behavior.

Student teams developed information graphics as a way to explain complex data to the public. Stories and dataviz products from the class were picked up by West Virginia media organizations.

Many of the students decided to take the class because they’re interested in investigative journalism, a significant aspect of which involves collecting data and understanding its implications. Other students are motivated by an interest in politics.

“I’m excited about this class because, by learning to filter this kind of data, I will learn teaching techniques that I can pass on to future students,” Bass said.