
Robert W. Ressler, Brandeis University
What’s the issue?
A mission statement is like a window into the soul of an organization. For nonprofits, the need to communicate with many diverse stakeholders creates an incentive to craft a mission statement that is reflective of an organization’s purpose, values, and on-the-ground activities. That makes a nonprofit’s mission statement a uniquely reliable source of information regarding actual organizational characteristics, especially ones that are not elsewhere easily categorized in a single activity code on a tax form, such as organizational identity, target audience, or theory of change. In addition, nonprofits must provide a mission statement to the IRS. But until recently, the technology simply didn’t exist to easily access and process nonprofit mission statements (for example, high-speed computers running text processing software on digitized tax forms).
My recently published NVSQ article with Brad Fulton and Pamela Paxton, titled Activity and Identity: Uncovering Multiple Institutional Logics in the Nonprofit Sector, takes advantage of technological developments to illustrate how mission statements can be used to categorize nonprofits on new, dynamic, dimensions. Specifically, we estimate how many religious organizations span the nonprofit sector in fields like education and health care. When we started the project five years ago, we chose a dictionary approach, or using a list of specific terms selected by content experts, to process mission statements. But recently, more complex and resource intensive classification methods have been developed using machine learning. So as our article has approached publication, I’ve had to spend a little time thinking about the relative benefits of a dictionary approach to a machine learning one. I’m sharing those takeaways here.
Continue reading “Classifying Nonprofits Based on Mission”


