A typical investigation might proceed like this: a journalist spends months (or more) reporting, keeps it fairly hush-hush, writes several stories, and then monitors comments and the social web as reaction unfolds.

But for our ongoing investigation into patient safety, spearheaded by reporters Marshall Allen and Olga Pierce, we’ve launched community and crowdsourcing efforts long before we’ve published a single traditional story. We’ve developed a database of sources that is helping to inform the investigation as we report it. And we are using Facebook to create a space for patients, providers and journalists to discuss patient safety issues openly.

So far, these efforts include:

• A Facebook group with over 1,350 members, who share stories, swap practical advice and engage in constructive dialogue.
• A questionnaire for patients and their families, which has generated nearly 260 tips since May
• A series of discussions with medical providers and a questionnaire that has generated 50 tips since September
• A community page on ProPublica.org featuring all of the above, plus Q&As with other health care reporters, resources for patients and related coverage.

Why go through all this effort? Because we are interested in using social media to source our journalism instead of just promote it. As we look back at 2012, here are a few key things we’ve learned about using social and community tools for investigative journalism.

Building a Database of Sources

Every social component of the patient harm investigation is ultimately meant to drive people to two key documents: a questionnaire for patients or family members who may have experienced medical harm, and a separate questionnaire for medical providers.

Our questionnaires are one of our most valuable sourcing tools, because they turn a jumble of stories and comments, from a range of social networks, into structured data that our team can sort and annotate. We almost always use Google Forms, which feed answers into a shareable, searchable spreadsheet. The responses are private. But as patients and providers respond, our reporters can begin to spot patterns, story leads and sources

The key to getting people to fill out the questionnaires boils down to networking and repeated promotion. Whenever someone shares a story in our Facebook group, we ask them to fill out a questionnaire.  We link to them in the majority of the health care stories (and news apps) we publish.

And while social networks can provide great new opportunities for community building, don’t forget good old-fashioned networking. Consumers Union, the nonprofit arm of Consumer Reports, has a large existing email listserv of people concerned with patient safety issues, so we asked them to link to the questionnaire. We also spoke at one of their conferences last month, and saw a spike.

Different Spaces for Different Audiences

Most reporting projects aim to reach a range of readers. People concerned with patient safety issues, for example, fall into three broad buckets: patients or families who have recently experienced harm, established patient advocates and medical professionals.

We established our Facebook group as a way to connect with all three groups, but quickly learned that each required a different engagement strategy. Many of those recently harmed are (justifiably) angry and vocal, and tend to use the community as a support group. Meanwhile, medical professionals and advocates were looking for constructive conversations about patient safety policy and possible solutions.

Patient stories quickly buried other voices, and we didn’t see the level of provider interaction we were hoping for when we created the Facebook group.

So, we pivoted. Instead of steering the Facebook conversation toward providers, we created a community page on ProPublica.org focused on balancing perspectives and increasing responses to our provider questionnaire. Since then, we have received 50 responses from providers with a range of expertise.

Ultimately, it’s important to remember that although you might create an online community, once it reaches a critical mass, it belongs to the group – not to you. While we work to guide and highlight constructive conversations (and maintain community guidelines), we see ourselves as facilitators, not minders.  

Facebook Groups: Give, and Ye Shall Receive

For any group to reach critical mass it needs to provide value. It’s key to create content and resources specifically for the community.

In the early stages of our Facebook group, we operated on an editorial schedule, tailoring a mix of expert Q&As, posts commissioned from community members, discussion questions and resources. We listened. When people expressed frustration that the media wouldn’t tell their story, we created a two-part series with tips on dealing with journalists.

Our challenge was to maintain a constructive environment around a very emotional topic. As the group grew, cliques formed and we began to see some bullying and blanket allegations against the medical industry (which we discourage in our guidelines). We stepped up our moderation efforts and became more stringent about deleting inappropriate comments and re-posting community guidelines.

But our most successful strategy to date? We engage with people directly, and occasionally give them social media etiquette tips. If someone crosses the line, we message them, sometimes privately, sometimes in public, with a warning and explanation of our community guidelines. We also invited group member Regina Holiday, a well-respected patient advocate, to write a post on social media etiquette.

The response was outstanding and with time, the group self-corrected. In fact, one member — whom we had to ban for bad behavior, and let back in after coaching — has become one of our super users. Now, she often welcomes new members to the group and re-posts community guidelines to remind others of the group’s standards.

The success of this community also hinged very much on teamwork in our newsroom. Myself, Marshall, Olga and our editor Tom Detzel worked closely together to manage the group, reducing the overall time required to moderate, grow and network the community.

Whenever Possible, Structure Social Data

We chose Facebook for hosting our group for a few reasons: millions of Americans are already using it, notifications can help our team track contributions and members can add their friends; in September, one advocate added 250 of her Facebook friends to our group overnight (which also led to a spike in patient questionnaire responses). 

But it has some drawbacks. The group search function makes it difficult to mine contributions for potential story leads, and moderators have limited ability to permanently highlight quality conversations.

And that is why Facebook’s Group API is your best friend. If you have an access token for a group, you can download posts with the related author, timestamp and comment URL into a CSV file. Al Shaw, one of our news apps developers, set up a script for our team to download contributions to our Facebook group in a matter of minutes, and the spreadsheet makes it much easier to search for specific keywords, posts or people.

How To Download Your Facebook Group Data

Here are instructions for Mac users to download Facebook group data, courtesy of Al Shaw.

• Get an access token for your group from the Facebook Graph API Explorer. A box will pop up. Make sure you click the user.groupspermission, and then click Get Access Token again. Copy and paste this into a text editor.
• Copy and paste your group ID (example: https://www.facebook.com/groups/209024949216061/) into a text editor. You can also test your group ID in the Facebook Graph API Explorer by entering “GROUPID/feed” in the get field. You should see all of your group posts.
• Grab the script for scraping the feed here: https://gist.github.com/4322279/
• Click “download gist” and decompress the file on your computer.
• Navigate to where that file exists in the terminal (example: iMac:/ myname$ cd /Users/myname/documents). Here are instructions on how to use the terminal. Stack Overflow is also a good place to go with questions.  
• In your terminal, type: “sudo gem install crack fastercsv rest-client” and hit return. You’ll have to enter your network password.
• Type: “ruby fb_scrape.rb YOUR_ACCESS_TOKEN YOUR_GROUP_ID” and hit return. Make sure to note that this is two pieces of information: your access token, and your group id – both of which you pasted into a text editor earlier.

You'll see all the Facebook data fly across the screen. A file should show up in your gist folder (check in your Finder) called fb_posts_GROUPID.csv (with that long number being the group ID you put in). Click to open, and you have your spreadsheet!