Neighborhood boundaries on the EH Data Portal
New York City has hundreds of neighborhoods and nearly as many ways of drawing neighborhood boundaries. When you visit the Environment & Health Data Portal, you might notice that data is available in several different neighborhood schemes.
For example, dig around in the asthma data pages and you’ll find Adults with Asthma presented by UHF34 neighborhoods, Public School Children with Asthma presented by UHF42 neighborhoods, and Asthma emergency department visits are offered by NTAs. Other data on the portal are offered by CD, ZIP code, PUMA, and occasionally even police precinct.
What are all these neighborhood schemes, why do we use one and not the other, and why isn’t the whole system consistent?
Most common neighborhood boundaries
The most common neighborhood boundary schemes on the EH Data Portal are United Hospital Fund neighborhoods, Community Districts, and Public Use Microdata Areas.
United Hospital Fund neighborhoods
United Hospital Fund neighborhoods (UHFs) are a neighborhood scheme created by the Health Department, the United Hospital Fund, and other city agencies in the 1980s. They were designed for health research, and to be roughly similar to NYC’s Community Districts. There are two versions of this scheme: UHF42 (with 42 neighborhoods), and UHF34, with 34 neighborhoods (where several neighborhoods from UHF42 are combined into one). But since there are 59 community districts, UHF42 isn’t that close to CDs.
Community districts
There are 59 Community Districts (CDs) in NYC, each overseen by a Community Board that advises on land use, zoning, city budgets, and more. As a political boundary, CDs are useful geographic units for breaking down city operations. Learn more about Community Boards.
Public Use Microdata Areas
There are 55 PUMAs in NYC. PUMAs have similar boundaries to Community Districts, but there are four PUMAs that are made up of two CDs combined into one PUMA. Because they are so similar, PUMAS and CDs can be used as a proxy for each other. PUMAs are also called “Subboro.”
While these geographies have similar boundaries between neighborhood areas, they are not identical. For example, areas in Mott Haven/Port Morris, Melrose South/Mott Haven-North, Longwood, and Hunts Point are all in separate Neighborhood Tabulation Areas. But in Community Districts, Mott Haven/Melrose is one Community District, and Hunts Point/Longwood is another. And in PUMAs, Mott Haven and Hunts point are both in one PUMA. These divisions don't nest neatly within one another - familiar neighborhoods can be broken up or aggregated into smaller and larger geographic schemes.
Continuing from the previous example, United Hospital Fund neighborhoods have similar boundaries, but they are not identical. Most UH42s break UH34s into smaller geographies; note how the South Bronx UH34 combines three UH42 neighborhoods (Highbridge - Morrisania, Crotona - Tremont and Hunts Point - Mott Haven).
Nesting: how neighborhood schemes have different root units
These neighborhood schemes have different building blocks. Let’s explore these.
United Hospital Fund neighborhoods
United Hospital Fund neighborhoods (UHFs) have boundaries based on ZIP codes. This geography was created by the Health Department, the United Hospital Fund, and other city agencies in the 1980s. They were designed for health research, and to be similar to NYC’s Community Districts.
Health data—like somebody’s hospitalization record, for example, or a response to a survey—often includes a person’s ZIP code. It’s the most readily available piece of geographic information in administrative data. It’s also the neighborhood designation that most people know and can provide when responding to a survey.
To protect privacy, we often bundle (or aggregate) data from a larger area, so we need a scheme of neighborhoods that are made up of a collection of ZIP codes: UHFs. Collecting data by ZIP code and then “rolling up” into UHF neighborhoods has been used in health research for decades. The methods for our surveys (like the Community Health Survey) are designed to include enough people from each UHF neighborhood so that there’s a “representative sample” of all New Yorkers, and so that we can compare neighborhoods with high statistical confidence. Usually, we use UHF42 neighborhoods, which breaks the city down into 42 neighborhoods. Sometimes, though, we use UHF34 neighborhoods—by grouping together some of the neighborhoods, we can increase the statistical power of a survey.
In the map below, notice how three UHF42 neighborhoods in the South Bronx are combined into one UHF34 neighborhood—and how the UHF neighborhoods have ZIP codes (or, more precisely, ZIP code tabulation areas) as their root unit.
Public Use Microdata Areas
Public Use Microdata Areas (PUMAs) have boundaries defined by the US Census. They are made up of groups of census tracts.
There are 55 PUMAs in NYC. PUMAs have similar boundaries to Community Districts, which means that often, one can be used as a proxy for the other. In the CD/PUMA map above, notice how Brooklyn CD 1, in Greenpoint/Williamsburg, is almost identical to the PUMA. And, there are four PUMAs that are made up of two CDs combined into one. Notice how two CDs in the South Bronx combine to form one PUMA.
Each PUMA breaks down into Neighborhood Tabulation Areas (NTAs), and each NTA breaks down even further into census tracts.
Community districts
Unlike PUMAs and UHFs, Community Districts don’t have a convenient root unit. So, NYC Planning created Community District Tabulation Areas (CDTAs) to approximate Community Districts using census tracts as their building blocks. Census tracts don’t always align perfectly with Community District boundaries—there are areas where tract lines and CD boundaries differ slightly. But, they’re close enough that CDTAs can serve as a “census-compatible” version of CDs. This makes them especially useful for linking Community District data with data from the US Census or American Community Survey (ACS).
Boundary updates in 2020
In 2020, the US Census updated the boundaries of census tracts—which means that schemes based on census tracts (NTAs, PUMAS, and CDTAs) also changed. These changes reflect population and housing changes and were made to more accurately represent the communities that live there. Our recent data generally uses the updated 2020 maps, but you may find older data on our website that uses 2010 map versions. The map changes are generally subtle, but they may affect trends in data for certain neighborhoods.
What do you do when you’re looking for data for one type of neighborhood, but the data is only available at a different scheme?
It can be difficult to work with several datasets when the data are for different types of neighborhoods. For example, it can be a challenge to look up health data for a Community Board or a City Council District when those data are only available at UHF42.
Beta NYC has a tool called Boundaries, which allows you to compare how NYC is divided into different districts. This tool may help you decide which neighborhood area to choose when presenting data in research papers or at board meetings.
You can also use our new Find your UHF neighborhood tool, where you can search by Community District or City Council District, and see what UHF42 neighborhoods overlap it.
Using these tools, you can find that sometimes, neighborhoods in different “schemes” overlap pretty well—meaning that data for one “scheme” can be used in another scheme. But more often, different neighborhood schemes have boundaries that conflict and don’t conveniently overlap each other. When this happens, you can use the Boundaries or the UHF neighborhood tools to:
- Look up the overlap between your desired area and the available neighborhood scheme.
- Get values and see how much overlap there is. For example, if you are looking up a certain Community District, but it’s only available at a larger neighborhood scheme like UHF42, maybe 70% of the CD is in one UHF42, and only 30% is in another. This can help you use the UHF42 data to estimate values for your CD.
What about when data is available at multiple geographies? How do you choose?
Here are some common scenarios that might help you think about which neighborhood scheme to choose for your needs.
- If you’re presenting at a Community Board meeting, you can use data by Community Districts, CDTAs (if census data), or estimate with PUMA/subboro.
- If you’re using ZIP code level data, you might need to aggregate up to UHF neighborhoods.
- If you’re conducting a research study with Census data, PUMAs or NTAs might be best.
Neighborhood boundaries can be imperfect representations of New York City’s communities. But even when data are only available at imperfect neighborhood schemes, they reveal important insights about health, housing, and the environment. When we analyze differences by neighborhood, the data allow us to explore questions and provide meaningful perspectives about inequities and opportunities across New York City.
Appendix: common uses for each scheme
| Boundaries | Based on | Number in NYC | Common use |
|---|---|---|---|
| UHF42 | ZIP codes | 42 | Health surveillance and public health reporting (like the Community Health Survey) |
| UHF34 | ZIP codes | 34 | More statistical power for public health reporting |
| CD | Political boundaries | 59 | Local governance through Community Boards |
| CDTA | Census tracts | 59 | Approximating CDs, for census-compatible statistical analysis |
| PUMA | Census tracts | 55 | Research, using statistically meaningful areas of ~100,000 people |
| NTA | Census tracts | 195 | Neighborhood identity, fine-grained data |
Matching datasets: a brief overview of GEOIDs
A GeoID (Geographic Identifier) is a unique code used to label a specific area on a map—like a neighborhood, ZIP code, or census tract—so that data about that area can be organized, matched, and analyzed. A GeoID is like a “name tag” for places. Every area, from a small city block to an entire borough, can have a code that identifies it in a dataset.
| Boundary scheme | GeoID Example | Format | Determined by |
|---|---|---|---|
| State | 36 | Numeric | Census FIPS code for NY |
| County | 36061 | Numeric | State + County |
| Census Tract | 36061000100 | Numeric | 11-digit Census tract |
| Block Group | 360610001001 | Numeric | 12-digit Census block group |
| NTA | MN0302 | Alphanumeric | NYC Planning-defined |
| PUMA | 03714 | Numeric | Census-defined |
| UHF42/UHF34 | 303 | Numeric | DOHMH-defined |
| CD | 203 | Numeric | BoroCD (Borough + District) |
| ZCTA (ZIP) | 10454 | Numeric | Census ZIP approximation |
A GeoID (Geographic Identifier) is a unique code used to label a specific area on a map, like a ZIP code or census tract, so that data about that area can be organized, matched, and analyzed.
A GeoID is like a “name tag” for places. Every area, from a small city block to an entire borough, can have a code that identifies it in a dataset. These codes help different datasets talk to each other by matching information. Census GeoIDs are numeric and follow a strict nesting structure—as shown in the previous table. Visit NYC OpenData for the full crosswalk of 2020 Census Tracts, to 2020 NTAs, to 2020 CDTAs. If you’d like to see how ZIP Codes (or ZCTA) form UHFs, visit the geographies folder in our data repository.
Edwin J. Torres/Mayoral Photography Office, City of New York
Published on:
August 1, 2020