Skip to content
Digital Impact

Digital Impact was created by the Digital Civil Society Lab at Stanford PACS and was managed until 2024. It is no longer being updated.

Nonprofit Ecosystem Research and Visualization

Team

[ess_grid alias=”drexel”]

Challenge

Can a data-driven and data-visualized approach to understanding the breadth, range, and scope of nonprofit organizations within a community, combined with important community and individual metrics, increase opportunities for collaboration among nonprofit organizations and enhance donor knowledge of the nonprofit ecosystem?

Project Overview

This project was supported by a 2016 Digital Impact Grant. In the one-year grant period, we undertook and completed the three phases of work.

Phase 1: Stakeholder Input and Field Scan of Data Sources/Community Metrics

Based on a review of the existing academic and practitioner literature, we solicited additional input from a range of stakeholders in the nonprofit and philanthropic sectors. Through semi-structured interviews and conversations, we identified both the  challenges they face when seeking to use data for strategic decision-making as well as the types of information they seek.

The common themes identified from the literature and stakeholder interviews were:

  • Challenges: Nonprofit Leadership
    • Lack of easy access to useful data on community needs
    • Lack of understanding of nonprofit ecosystem
    • Lack of training and expertise in gathering and analysis
  • Challenges: Philanthropic Leadership
    • Lack of easy access to community-level metrics
    • Inability to make data-informed case for support/grants
    • Lack of training and expertise in gathering and analysis
    • Lack of sector-wide, regional metrics

Based on these findings, we then identified key public data sources on both the nonprofit ecosystem and community demographics. In order to create a tool that could eventually be scalable and national, we determined that the initial data used would be sourced from the Internal Revenue Service (IRS) and U.S. Census Bureau. The IRS data would provide in-depth insights into the nonprofit organizations while Census data (more specifically data from the American Community Survey) would provide insights into individual and household demographics. Of particular interest to this work was the newly-released open 990 data of the IRS. We believe that the prototype tools developed for this project are most likely the first online tools to integrate this new data source.

Phase 2: Data Gathering and Analysis

Using the city of Philadelphia as the use case for the project, we compiled all of the necessary IRS and Census data needed to power the prototype tools. Through this process, we gained deep insights into the challenges of using and integrating the IRS open 990 data. We identified three key issues:

  • Representativeness: The IRS released open data only on those nonprofit organizations that opted to electronically file their Form 990. This only reflects approximately 60% of the nonprofit sector, and use of this data would need to be supplemented by organizational data from other IRS sources (e.g. IRS Business Master File).
  • Technical Issues: The IRS released the open data set with an inconsistent XML schema, no data dictionary, and no documentation. As a result, there is no effective way to automate the process of compiling the data in a consistent manner.
  • Consistency: Because the three different Form 990s that are included in the open data set have very limited data fields in common, there is no standardized way to cross-reference data from one form to another. This is a function of inconsistency in the way that each form collects data.

Despite these challenges, we were able to utilize the IRS open 990 data and combine it with other IRS data to identify the 2,327 nonprofit organizations within the city of Philadelphia. This figure is based only on nonprofit organizations that file either the IRS Form 990 or the IRS Form 990EZ. For individual and household demographics, we compiled a range of data from the U.S. Census for all 384 census tracts within the city of Philadelphia. The core fields gathered from both data sources:

  • US Census Data Fields: Population (Children), Poverty Rate, Race/Ethnicity, Median Age, Employment Status, Per Capita Income, Number of Households, Median Household Income, Median Home Value, Household Type, Languages Spoken
  • IRS Data Fields: Organization Name, Street Address, Mission Statement, Nonprofit Type (NTEE), Year Founded, Budget Size, URL, Phone Number

All data were reviewed for accuracy and consistency, missing data fields were corrected, and the data were compiled into two separate databases, one for nonprofit organizations and one for individual and household demographics.

Phase 3: Prototype Tool Development

Based on the stakeholder needs and the data gathered, we developed two prototype tools to assess our core research question. The first tool served as a minimum viable product to test the use of the two data types and determine the visualizations. Using just the several basic data points from each data set, we created an online interactive map that allows users to identify specific communities based on userselected characteristics and to identify the nonprofit providers in those areas. Despite using just a few key data points, this tool proved to be very effective in visualizing the data and identifying specific communities based on user interaction.

Conversations

Grantee Profile: ImpactView

Digital Impact hosted a virtual roundtable conversation about Open 990 data, featuring project lead Neville Vakharia:

The Open 990 and US Nonprofits

Neville Vakharia presented ImpactView Philadelphia at the Data on Purpose conference at Stanford University in February 2018.

Outputs

Insights

Through the stakeholder input stage of the project, we learned that the needs of nonprofit and philanthropic sector leaders regarding data-informed decision-making were surprisingly foundational. Despite the rapid increase in the amount and availability of data, many organizations did not have the resources or capacity to use this data. While this was not unanticipated, it was helpful to understand the specific types of information that these leaders sought and how basic tools could provide better access to this data. It was also clear that these leaders wanted to make more data-informed decisions and acknowledged that they were limited in their abilities to do so.

Next Steps

We have proven the concepts behind our research question, and now must continue the work towards achieving our longer-term outcome of increasing opportunities for collaboration and philanthropic support within the field. This will require a multi-stage approach combining stakeholder feedback, an update of the prototype tools, and expansion of our geographic reach.

Connect with the project team: on Twitter at @NevilleVak or by email at nvakharia at drexel dot edu.