The ZIP code is commonly used in business applications for two primary reasons – they are well recognized and utilized as the core geographic identifier in a wide range of both postal and non-postal sources. Because they are created by the post office for the purpose of delivery logistics, they have little regard for behaving well with census boundaries (block, block group, tract) and administrative boundaries (place, county, and even state).
Historically, licensing ZIP code boundaries has been an expensive proposition. When the Census Bureau first created its ZIP Code Tabulation Areas (ZCTA), many jumped on the bandwagon – mainly because they are free to use. Free, however, isn’t always your best choice.
The methodology used is to query addresses by ZIP code within each census block then assign that block to the ZIP code with the most addresses. Ties and blocks with no addresses are assigned to the ZCTA with the longest shared perimeter. Note that this should make it clear that ZIP codes do not follow block boundaries. In general, the ZCTA boundaries are decent representations of the actual boundaries in urban areas. In rural areas, the sheer size of census blocks can result in substantial error.
Free comes with known problems, but are they significant enough to warrant upgrading to a commercially developed boundary set, even one which is modestly priced?
For many years, AGS has been developing and using internally developed ZIP code boundaries. These boundaries are updated annually using the USPS address and ZIP records, TIGER street files, natural and landmark features (such as rivers and rail lines), and georeferenced ZIP+4 records. When Census Blocks are split between ZIP codes, we use street setback corridors and medial axes from Voronoi diagramming. The results are detailed representations that include some of the nuances of the boundaries in urban areas that cross census block boundaries, and in rural areas provide substantial improvements over the full block assignment method used by the Census Bureau.
The map below shows Jackson County in northeastern Alabama. While some of the ZIP code boundaries clearly follow the county and state boundaries, many are split by the county boundary. There is little correspondence between the commonly used block group level and the ZIP code.
In the comparison map below, the ZCTA boundaries are in blue and the AGS boundaries in purple. The table below shows the overlay:
Note that the AGS boundary is considerably more complex and does not include territory which includes addresses which have a different ZIP code than the drawn boundary would indicate.
Coupled with the more accurate representation of boundaries is the ability to properly allocated demographics to the boundaries using the AGS block allocation tables, which use a set of disaggregation factors which should be used for different data types. With access to the complete underlying data, AGS is able to provide both more accurate boundaries, but more importantly, more accurate demographics and business establishment data summarized by ZIP code.
In exchange for a very modest investment in better quality data, users of analytics platforms or custom internal applications and processes can eliminate at least one source of error – reducing uncertainty in business decision making, which is, after all, why we all are here in the first place.
Recent Comments