The Storytelling Sleuth: Using Data Science to Make Your Case
Three non-profits that utilize data to tell stories.
Also co-authored by Miriam Young, DataKind’s communications specialist.
For detectives investigating a crime, suspect interrogation, witness interviews, and advanced forensics like DNA fingerprinting, voice & gait analysis, and biometric identification techniques have made the investigation process much more powerful by allowing us to see beyond what we can observe and accurately remember.
Robin Deutsch Edwardstwitter.com/robindedwards2
Robin Deutsch Edwards is the owner and principal for High Mileage Communications, which provides communications and marketing supports for mission-driven businesses. For DataKind, Robin leads strategic communications and public relations.
Miriam Young is DataKind’s communications specialist. She previously worked for Taproot Foundation and NPower’s skilled tech volunteerism program, The Community Corps.
Storytelling is like detective work. For mission-driven organizations, stories are our evidence. They make our case proving the difference our efforts make in the world. And as forensic science profoundly shifted the field of criminal justice, so too has data science given us the tools and techniques necessary to see hidden patterns and reveal new insights about our work. Data science uses the power of computer programming and statistics to organize, analyze, manipulate, and present data in a new way. It allows us to combine and use different types of data, like spreadsheets and satellite images, to see new patterns and gain a deeper understanding of how our work fits in.
If we limit our storytelling investigation to only what we can observe and count—the number of people our programs serve, event attendance, or dollars raised—then we’re like detectives pursuing an investigation with 18th century tools. In our quest to tell stories with purpose, we need the right tools to extract clues and valuable insights. Data science is a powerful way to tap into the full range of available data about the work we do and in turn identify our strongest stories. Whether your organization’s data comes from your website analytics, your programs, or publicly-available data sources, the fast-emerging field of data science makes it possible to go simply reporting on things we can see and count to leveraging a much wider set of data to make our strongest case. Here are three nonprofits—working with DataKind volunteers—that used the latest in data science to tell stories with purpose:
‘There is nothing like first-hand evidence.’ (Sherlock Holmes, A Study in Scarlet)
In many cases, there are untapped sources of data right at our fingertips. GlobalGiving, the world’s first and largest crowdfunding community for nonprofits, used data from their past work to identify specific project features (title, photo, etc.) that led to increased donations. DataKind volunteers used text analysis to examine the words used on each project page revealing a correlation between specificity of language and project success.
In other words, the analysis revealed that organizations using the GlobalGiving platform were less successful when they used generic words like funding for the “arts” versus a specific project like “a photography exhibit.” GlobalGiving’s analysis of their own data uncovered critical information that wasn’t possible to glean anecdotally. This information is empowering GlobalGiving with a compelling story to share with their constituents about how to raise the most money for their cause.
“The world is full of obvious things which nobody by any chance ever observes.” (Sherlock Holmes, -The Hound of the Baskervilles)
The gem hiding in your data might just be the powerful hook you’ve been hoping for in your next story. During a four-month project for DonorsChoose.org, a nonprofit organization that provides a platform for teachers to post classroom project requests for funding, DataKind volunteers helped the organization create a taxonomy to categorize requested products so they could be grouped and analyzed in the future.
By digging into the data (from more than 80,000 projects in 2014), it turns out that schools in high poverty areas were more likely to request basic office supplies, like paper. This indicates that students in some of the highest poverty schools do not have access to supplies and are forced to crowdsource basic educational tools. The new finding uncovered by the data scientists became a central storytelling point in the organization’s biggest annual campaign held in conjunction with the Gates Foundation.
'What one man can invent another can discover.’ (Sherlock Holmes, The Adventure of the Dancing Man)
There are many free and publicly-available sources of data just waiting to be used to enrich our efforts and help maximize impact outside of our four walls. DC Action for Children is a nonprofit organization that works to fight poverty and support child well-being in Washington, DC.
Like many organizations, DC Action for Children produced static reports in PDF format to aid in their advocacy efforts. With the help of DataKind volunteers, the organization took advantage of a variety of publicly available data to create an interactive visualization to illustrate the various factors that lead to child poverty in Washington. Using the outside data sources helped paint a bigger picture—showing for example, that one out of every three kids in DC lives in a neighborhood that doesn’t have a single grocery store. The volunteers then helped visualize that larger story in an interactive way, which helped draw attention to the organization’s advocacy efforts.
Not every organization has the resources available to hire a data scientist but we can start by recognizing the opportunity data science offers to add depth and context to our efforts. After all, we aren’t all legal experts but we know when to call a lawyer.
At the end of the day, the lesson is to think like a detective when searching for your most powerful stories and ensure that you are using all of the tools and techniques available to make your best case.