5 takeaways:

➀ “Don’t tell me what you believe. Show me your checkbook.” Data —whether from individual bank statements or government spreadsheets – is an accountability mechanism, said Derek Willis, a former ProPublica data journalist turned professor at the University of Maryland’s Philip Merrill College of Journalism. He recalled a pastor who used to say, “Show me your bank statement, and I’ll tell you what you believe.” Unlike sources with something to hide, data can’t run away from you when you ask it questions, Willis said. So keep asking until you get exactly what you want. Data “gives us an opportunity to really test out some theories, to really try to see if what people are saying to us is what is actually going on,” Willis said.

➁ Manual repetitive tasks mean you’re doing it wrong. At the beginning of the COVID-19 pandemic, Willis and his ProPublica colleagues found themselves repeatedly searching the Federal Procurement Data System for the same types of information about who got contracts. That was their cue to automate. [The tracker they built is here and Willis teaches journalists how to use it here.] “If you find yourself going back to a site or doing anything that’s manually repetitive with a computer, there’s a better way to do it because computers are built to automate tasks,” Willis said.

➂ Turn spreadsheets into stories by finding a character. ProPublica journalists knew they wanted to zero in on first-time contractors trying to cash in on the pandemic spending bonanza. They started combing through the names and digging into their backgrounds. That’s how they found Zach Fuentes, a new contractor who had been former deputy chief of staff in the Trump White House. “He left the White House, started up his company called Zach Fuentes LLC, and then won a $3 million federal contract to supply masks for the Indian Health Service,” Willis said. The IHS told ProPublica that nearly a quarter million of the masks Fuentes supplied may not be suitable for medical use. Thus, what began as a line in a spreadsheet turned into a very human story of waste, abuse and its effects on a marginalized community.

➃ But which programming languages to use? Much of the work Willis did for ProPublica required programming, but it need not be intimidating. “If you can follow a recipe, you can do programming because recipes are basically a set of instructions in a particular syntax … and just like with cooking, if you can read and follow instructions and you can practice, you’ll get better at it.” The main programming languages used by journalists — Python, R, Ruby, JavaScript — can do the same things. “There are a lot of people on the internet who will be happy to engage in holy wars over my programming language is better than yours; you should mostly ignore those people,” Willis said. “Pick the language that makes the most sense to you. The one that looks most like English to you or whatever your native language might be.” And be prepared to Google error messages.

➄ Don’t let Atom feeds intimidate you. Data journalists are used to working with CSV files but shouldn’t overlook Atom feeds, Willis advised. Atom feeds are XML – “structured data, like a spreadsheet, but it’s on the web” and has a permalink. Atom feeds also include more granular data than CSVs. For example, information on who was a new federal contractor was not available on the government website, but reporters found it after Derek used Ruby to read the XML of the Atom feed and pull the data into a database “so that we could ask questions of it,” Willis said.

You may also like Derek Wills’ other talks to journalists: Documenting Federal COVID Contracts, Who Got Those Federal Contracts?, Using Data in Investigative Reporting and Reporters Share Tips for Documenting Campaign Cash.

Speaker: 

Derek Willis, Lecturer, Philip Merrill College of Journalism, University of Maryland

This program was funded by the Evelyn Y. Davis Foundation. NPF is solely responsible for the content.