About the data
What this site shows, how to read it, and what it can't tell you.
Who this is for
- Foundations deciding where and how to invest in civic life
- Universities building community partnerships
- State and local governments planning civic engagement strategy
- Organizers who want to understand their community and bring others into the conversation
- Community members, reached through all of the above by shared links, QR codes, and printed profiles
What "civic opportunity" means here
A nonprofit organization offers a civic opportunity if its IRS tax return or its website describes at least one of four ways for people to take part:
- Membership: people can join.
- Volunteering: people can give their time.
- Public events: gatherings open to the community.
- Civic and political action: advocacy, organizing, voter engagement, and similar activities.
Each organization is also assigned one of 15 kinds (religious, youth, arts and cultural, community, social and fraternal, education, healthcare, economic, housing, professional, unions, foundations, hobby and sports, research and think tanks, and political) by a model trained on organizations' own descriptions. When no kind could be identified, the site shows it as "Type not identified". It never leaves these organizations out, so the numbers always add up.
How to read the numbers
- Counts are numbers of organizations. Local chapters of the same national network in one place count once.
- Per 10,000 residents rates are smoothed. Very small places get pulled toward the national rate so a handful of organizations doesn't produce an extreme number.
- Similar places are the 10 places most like this one in population size, rurality, poverty, education, broadband access, unemployment, single-parent households, and racial and ethnic makeup. They are never matched on civic opportunity itself.
- More / fewer than similar places means at least a third more, or a quarter fewer, per resident than the typical similar place. For a single kind of organization the bar is higher (twice as many, or half). A comparison needs about five organizations on one side; below that the site says the numbers are too small.
- Confidence on each observation reflects how many organizations it rests on and how many have no identified type.
What the data can't tell you
- What a community wants or needs. Fewer organizations listing something isn't the same as people wanting more of it. Only conversations with people there can tell you that.
- Activity that isn't written down. An organization that doesn't describe an opportunity in its tax return or on a website counts as not offering it.
- Informal civic life. Neighbors, mutual aid, congregations without nonprofit records, schools, and local government aren't captured.
- Change or impact. The data is one snapshot. It can't show trends or whether any program made a difference.
- Specific organizations. Records are de-identified (no names, tax IDs, or street addresses), so the site shows kinds of organizations only. Organizations that use a P.O. box as their mailing address are left out, because their location is uncertain.
How the site is designed for learning
The place pages draw on these learning patterns:
- orientation: the Look โ Compare โ Discuss path at the top of each place page, with why each step matters.
- teaser and aha moment: "Guess first" asks which kinds of organizations you expect to lead, then shows the answer, so expectations meet the data.
- reveal: "How these numbers are made" opens up the steps from IRS records to websites to the comparisons you see.
- compare & contrast: every place is set beside the 10 places most like it.
- open ended questions: every observation comes with a question that has no single right answer.
Pattern names are from Learning Patterns by Tim Klapdor, licensed under CC BY-NC-ND 4.0. The descriptions above are our own and say how this site applies each pattern.
Data sources
- Civic opportunity and organization types: MapAgora datasets (Kim, de Vries & Han 2025), doi:10.7910/DVN/IRCA7C. MapAgora is built from the IRS Exempt Organizations Business Master File, IRS Form 990 filings, and text from organizations' websites. Counts on this site are recomputed from MapAgora's de-identified organization-level file using the same rules as its published county and ZIP code files.
- Population and community conditions (poverty, broadband, education, unemployment, single-parent households, racial and ethnic minority share): U.S. Census Bureau, American Community Survey 2017โ2021 5-year estimates, as included in MapAgora.
- Metro, small town, and rural labels: U.S. Department of Agriculture, Economic Research Service, Rural-Urban Continuum Codes (2013), from its county classifications file. Codes 1โ3 are shown as metro, 4โ7 as small town, and 8โ9 as rural. A ZIP code takes the label of its county.
- ZIP codes and their counties: U.S. Census Bureau, 2020 ZCTA-to-county relationship file. Each ZIP code is assigned to the county holding most of its land.
- Boundaries (used to find neighboring places, not drawn on this site): U.S. Census Bureau cartographic boundary files for counties, states, and 2020 ZIP Code Tabulation Areas.
- Town names for ZIP codes: the most common mailing city among the organizations in each ZIP code, from the IRS records in MapAgora.
Colors use the "Smith" palette from MoMAColors by Blake R. Mills (MIT License).
More questions about the data?
The data, methods, and validation are documented in full elsewhere:
- Data paper: Kim, J.Y., de Vries, M., & Han, H. (2025). MapAgora, civic opportunity datasets for the study of American local politics and public policy. Scientific Data, 12, 1162.
- Harvard Dataverse and GitHub: the datasets and code.
- Mapping the Modern Agora: the research project, its findings, and papers.
- de Vries, Kim & Han (2024), "The unequal landscape of civic opportunity in America," Nature Human Behaviour.
How to cite
Kim, J.Y., de Vries, M., & Han, H. (2025). MapAgora, civic opportunity datasets for the study of American local politics and public policy. Scientific Data, 12, 1162. https://doi.org/10.1038/s41597-025-05353-6. Data: https://doi.org/10.7910/DVN/IRCA7C.
Who made this
Jae Yeon Kim (UNC Chapel Hill), Milan de Vries (SNF Agora Institute, Johns Hopkins University), and Hahrie Han (SNF Agora Institute, Johns Hopkins University). The MapAgora data was built at the SNF Agora Institute. Contact: jaekim@unc.edu.