Visitors to our Open Data & Mapping Hub geared toward GIS professionals and its more layperson-friendly cousin, DataNexus, have been able to access data for “Neighborhood Statistical Areas” for quite some time. And observant users may have noticed the addition of “High School Statistical Areas” to the list of available geographies. You may have wondered, what is a Neighborhood Statistical Area and how does it compare to a plain-old neighborhood? And what do High School Statistical Areas have to do with high schools? And why create these statistical areas in the first place? The answers are long overdue, and the subject of today’s blog post.
This entry represents the fourth in a series about geographic units, and the explanation of “statistical areas” builds on issues discussed in relation to Census blocks and the relationship between ZIP Codes and ZIP Code Tabulation Areas. [1] The “why” for these statistical areas stems from the final point made in the chapter on ZIP codes: Census tracts are great from a statistical standpoint, but they do not carry much in the way of spatial meaning for people. Most just don’t know what Census tract they reside in. But neighborhoods, at least for residents of the city of Atlanta,[2] are quite meaningful. Similarly, people typically know what high school their home is zoned for: real estate agents assure you of the quality of the schools when you buy a home; school zones related to an address are reported on sites like Zillow; school quality is baked into rents and home prices; and high schools are important foci of American life (e.g., ‘Friday night lights’).
But there’s a fundamental problem with these more meaningful geographies, namely that they do not mesh well with geographic units used by the Census and other government agencies for their data reporting. By way of illustration, Figure 1 below depicts part of part of the boundary between the Midtown HS and Washington HS attendance zones in the city of Atlanta:
Figure 1: Comparing High School Attendance Zones with Street Centerlines
The way that the school attendance zone boundary jumps on and off of streets seems bizarre until you consider how the purpose of each geographic unit (in this case, data collection vs. school attendance) might drive how the concept of “block” gets handled differently in each of them. Recall from the discussion of the difference between “city blocks” and “Census blocks” that (unless you live in a cul-de-sac) your neighbors across the street live on the same city block as you do, but that they live in a different Census block. And while your backyard neighbors live on an entirely different street, you and they reside in the same Census block. School attendance zones operate more like city blocks than like Census blocks. When a street serves as the dividing line between two school attendance zones, typically both sides of that street are zoned to the same school. This is quite practical: there is no sense in running buses from two different schools down the same street for pickups. Figure 2 that follows adds tax parcels to the previous map to illustrate this.
Figure 2: Mystery Solved: Why High School Attendance Zones Diverge from Street Centerlines
As this map shows, the “true” dividing line between two attendance zones is often the property line behind properties along the boundary street rather than the street itself, meaning that school attendance zones are not directly compatible with Census geography.
There are a number of ways to address this problem. The one that we opt for in this case is inspired by the solution that the Census Bureau uses when translating ZIP codes to Census geographies. Recall from our previous discussion that their approach is to treat Census blocks as ‘atomic’ units and form “ZIP Code Tabulation Areas” that conform to true ZIP codes as closely as possible within that constraint. We do the same for the high school attendance zones in our 11-county region; the term “High School Statistical Area” denotes that these are statistical areas built from Census blocks and not the “true” attendance zones.
Neighborhood Statistical Areas (NSAs) apply the same concept to neighborhoods within the city of Atlanta, only with some additional constraints: some Atlanta neighborhoods are very small [3]. To ensure that we can present statistically valid data for each area, we require that a neighborhood meet a minimum population threshold of 2,000 to have its own NSA. If a neighborhood fails to meet that threshold on its own, it is paired with one or more contiguous neighbors within the same Neighborhood Planning Unit (NPU) until the threshold is met.[4] Parts of the city that do not lie within the boundaries of any neighborhood [5] are joined to contiguous areas (neighborhoods or otherwise) within the same NPU to ensure that every part of the city falls within one, and only one NSA.
In short, High School Statistical Areas and Neighborhood Statistical Areas are close approximations of high school attendance zones and Atlanta neighborhoods, respectively. They allow us to present government data in terms of geographies more relevant to our experiences than Census tracts.
Notes:
[1] The other post in this series on geography delves into the history and use of Census tracts.
[2] Neighborhoods in Atlanta are officially recognized by the city government and play an important role in planning through the Neighborhood Planning Unit system. The NPU system has its origins in the 1974 Citizen Involvement Ordinance, which created these bodies “for engaging in comprehensive planning matters affecting the livability of neighborhoods.” Atlanta is divided into 25 NPUs, each of which is comprised of a set of contiguous neighborhoods. Each NPU holds monthly meetings at which residents have the opportunity to provide input on matter such as variances, zoning issues, and long-term planning. To learn more about the NPU system, visit https://www.npuatlanta.org/.
[3] For example, there is a neighborhood called “Harvel Homes Community” in NPU J that measures just 0.02 square miles in area and has only 55 parcels. Another neighborhood called “Just Us” in NPU T has 57 parcels and covers 0.03 square miles.
[4] NPU Q violates the minimum population threshold, but we chose to keep it as its own unit rather than join it with another NPU. NPU Q was created in 2009 when the city annexed unincorporated territory near its southwestern edge. The original plan was for NPU Q to expand into areas that are now part of the City of South Fulton, which incorporated in 2017. The cities of Atlanta and South Fulton were locked in a dispute over the Loch Lomond area, which was eventually adjudicated in court– see this Atlanta Journal-Constitution article for details.
[5] Neighborhood boundaries were set through a self-identification process. As a result, the classification of city territory into neighborhoods is neither exhaustive (there is territory not claimed by any neighborhood) not mutually exclusive (Atkins Park is considered its own neighborhood in some contexts and as part of Virginia-Highland in others). There are also two neighborhoods called Wildwood; thankfully they are found in different NPUs (C and H).