What happens when two or more newsrooms join forces on a beat? A journalistic collaboration, of course!

Welcome to the 37th edition of Conversations with Data, featuring Rachel Glickhouse, the driving force behind ProPublica’s recently launched Guide to Collaborative Data Journalism.

ProPublica is one of the field’s leaders in data-based collaborations, after bringing together over 1000 journalists to work on Electionland and more than 170 newsrooms for Documenting Hate. But, as Rachel tells us, there’s a lot that goes into making large-scale data collaborations work, and countless lessons to learn from.

What you asked

What types of projects are best suited to collaborations? Are there any types that aren’t well-suited?

Rachel: “You can collaborate on anything -- it just depends on how many newsrooms you want to involve. A collaboration can be as small as just two newsrooms, working together to co-report and publish a story. You can have a group of digital, print, radio, and broadcast newsrooms in the same city or state work together to tell stories about the same issue but on different platforms. A large group of newsrooms can work to report stories independently using the same shared dataset. The one thing you'll need regardless of the size of the collaboration is trust, which can be a challenge given the nature of our industry. For that, you'll need good communication, a clear set of expectations and guidelines, and someone managing the project.

If you have a very narrow, specific story to tell, it's more likely going to be suited to a small number of partners. Those newsrooms should be the ones best equipped to tell the story and also to deliver it to the right audience, whether that's determined by geography or the beat. If you want to focus on a specific issue but work independently, you can still band together and co-publish each other's work, or centralise your work on a landing page. If you have a large dataset spread across cities or countries, there's likely opportunity for a larger number of newsrooms to work together on it. (Think Panama Papers, for example.) Ultimately, it comes down to how you want to share resources and divide the labour. We have more tips on this in our collaborative data journalism guide.”

What are some of the biggest mistakes you’ve seen in collaborative projects and what can be learnt from these?

“One thing I often hear is that collaborations require at least one person to coordinate and manage the project. I can certainly attest to that!

I can also tell you about some of the challenges I've encountered. The first is to assume potential or new participants understand the project and how to use your resources. It's important to have a structured onboarding process, trainings, and written materials to make sure partner newsrooms are clear on what the project entails and how to use the data. It's critical to tell people what the data can be used for, and what it shouldn't. Set realistic expectations about requirements for participating in the partnership. If the requirements are technical in nature, make sure you communicate them clearly. If you're going to do a large partnership, it's best to keep barriers to entry low. Finally, the big challenge of any collaboration is ensuring that work will get done. In a big partnership, don't assume that every single partner will be productive or be able to find a story.”

You also launched a new tool with the guide, called Collaborate. Can you tell us what the tool does and some examples of how it can be used by journalists?

“This tool is meant to make it easier for journalists to work together on a shared dataset, whether that's within their own newsroom or between newsrooms. It's especially useful for crowdsourced projects, and can also be used for data projects.

Collaborate helps you track the status of each individual data point in a dataset. It lets you divide up a dataset in order to see which journalist is working on each data point or tip. It allows you to create filters and labels to parse, search, and sort the data. You can keep track of which tips were verified, and which data points have been reviewed. You can add notes to each data point, and export the entire dataset with all of the information you've added. You can create a contact log to track each time a tipster or source has been contacted, and by which journalist. You can also redact sensitive data. You can also limit access to each project by creating users with specific permissions.

Now that Collaborate is up and running, ProPublica is going to use it as our main tool for crowdsourced investigations, allowing us to coordinate between ProPublica staff and partner reporters. We're also planning to use it to open up some crowdsourced data sets to local newsrooms, since we sometimes share tips when we don't have time to tackle them all.

I think Collaborate can be really helpful in helping newsrooms tackle crowdsourced projects in order to track each tip or submission. It can also be useful in parsing through large datasets to find qualitative patterns or to select specific data points that would make for good stories or sources.

And if you're a developer, you can tweak Collaborate to meet your newsroom's needs; the code is available on Github.”

What are some of your favourite lesser known collaborative projects and why?

“Some of my favorite collaborations are:

• Resolve Philadelphia, which works with basically the local media market in Philadelphia to report on prisoner reentry (a past project) and poverty (their current project). Some of their funding is shared among partners and their whole model is really cool. The fact that they convinced so many competitors in a local market to work together for so long and have been able to produce so much reporting is really remarkable.
• Six newsrooms in Florida are working together to report and co-publish stories on climate change. The initiative was announced this summer. It's smart, timely, and a no-brainer.
• The Bureau Local in the UK does absolutely incredible work carrying out national investigations and getting local journalists to do their own reporting. They are the gold standard for local-national data collaborations.
• Comprova is a project in Brazil that grew out of an election fact-checking collaboration into an ongoing collaboration to fact-check and combat misinformation. There have been a number of these election projects around the world, but given the spread of misinformation beyond elections, I was really heartened to see this project was continued.”

What makes a collaborative project successful?

“There are many ways to evaluate the success of a collaboration, but a few I'd point to are output, reach, and impact. The larger the collaboration, the more difficult to manage the production of stories. If at least one story comes out of a collaboration, that's a success, and even better if it's a well-reported, well-told story. I'm partial to quality over quantity.

Next, it's useful to track the size of the collaboration and the audiences you've reached, which you can count by the number of newsrooms involved and their geographic coverage, metrics on stories produced from the project, and social media reach of posts related to the project. That can help you determine if your reporting made it to the audiences you were hoping to reach.

Finally, you can look at the impact of the reporting. I consider output and reach part of the impact of a collaboration, but if you can also determine changes that were a result of the reporting -- anything from a meeting held to a person arrested to a law changed -- that's success.”

Read ProPublica’s Guide to Collaborative Data Journalism here.

Our next conversation

From corporate watchdogging to political corruption, there’s plenty of stories to be found by digging into money trails. In our next edition, we’re looking for your advice on nailing ‘follow the money’ investigations. Comment with your tips!

As always, don’t forget to comment with what (or who!) you’d like us to feature in our future editions.

Until next time,

Madolyn from the EJC Data team