Datafest, a two-day contest to analyze campaign finance data that I organized last month at Stanford, was one of the best experiences of my wonderful year as a Knight Fellow. The event was a lot of fun and a great opportunity to learn. It produced very interesting insights and illuminated possible ways forward for data journalism. It was also great to see how much can be accomplished in just a day and a half when you have the right people in the room.
Ten teams of journalists, scientists and engineers worked incredibly hard and with a lot of enthusiasm, some of them through the night. A few teams focused on original reporting using political influence datasets. Others made tools to help journalists investigate their data, worked on a prototype of a consumer oriented app to explore campaign finance data, or on visualizations with very valuable insights. Some projects show good potential for further development. Below is a summary of their original project ideas and equally original team names!
We had four winners. One of them, Team Most Excellent, sought to answer a basic and pertinent question: Is the influx of out-of-state money in Wisconsin’s governor’s race unusual? Their conclusion? No, it’s not. One of our judges called it a “most compelling insight” that suggested that “Scott Walker is not an anomaly in Wisconsin.”
Team Gophers’ work also won an award and was an excellent example of combining data from diverse sources to analyze a single issue. They examined the political influence of one company: Lockheed Martin. Their conclusion was, as one judge put it: “whoever is in power is the one who gets the money.”
Team KeyStoners focused on a single issue: the Keystone Pipeline authorization in Congress. This winning team reached interesting conclusions. Among them: a less divisive issue is more susceptible to financial influence, and the best explanation for a Democratic member’s switching votes is his or her hailing from a predominantly Republican state.
Team Frienemies was awarded the first prize for its network graphs. These graphs produced some quite non-obvious insights. One example is revealing partisan alignment of non-partisan organizations. Their project can be of practical value to anyone who wants to explore political networks.
Team Awesome decided to answer the question: Can you buy votes? They created a model that examines how campaign finance contributions influence election results. They also worked on a prediction model to forecast the outcome of the 2012 presidential race based on the 2008 campaign data. A very interesting idea. Alas, the prediction model wasn’t completed due to time constraints.
The theme of Team Z’s project was: “Big donors of elections past—where are they now?” They created a tool that tracks patterns of political giving by individuals over time. For instance, the tool shows that some big donors have shifted their contributions from federal races to state elections. Others have sharply cut their contributions altogether. This raises the interesting question: What’s driving these changes?
Team Donkey Versus Elephant$: Their project emphasized the availability of Big Data from various government agencies and a variety of social networks. Their presentation focused on the idea that this data could be used for understanding and leveraging on election dynamics. Our judges liked that the team took a variety of approaches and visualizations to the same problem.
Team G8tor focused on anomaly detection and geographic exploration. Their visualizations revealed, among other things, that the top donor states are not the swing states.
One of our judges called Team Black Cobra “the most altruistic group.” Their goal was to create a general tool that can be reused by others. As another judge remarked: “Super useful tool for journalists, interactive and reusable. You can imagine journalists exploring their own questions around timing of votes on controversial issues in specific states and how that affects donations. One of my favorites in its power as a tool potentially serving a broad pool of people with varied interests.”
Team Eagle worked on creating an app to visualize the connections between industries, the politicians they fund, and the bills on which those politicians vote. Judges thought this was “a very promising direction.” They also commended the team for “best design and best potential to have a visually appealing, interactive and inviting app to explore the data.” And the verdict was: “Should become a classic for iPad.”
Overall, the Datafest proved both enjoyable and valuable. The event not only produced interesting analytical results, it also illuminated possible ways forward for data journalism. In addition, I learned a great deal about organizing and managing intensive, project-focused events. I’m planning another Datafest in October. It’ll be bigger, better and more interesting. Some very smart people are helping me to prepare it. I hope all our participants will take part in it. Stay tuned!