The Census bureau finally released the repeatedly delayed Demographic Profile and Demographic and Housing Characteristics File (DHC). Within hours of the release, our LinkedIn feeds were littered with maps and short commentaries, most without any acknowledgement that there may be issues with the data. Regular readers of our posts will know that the original PL-94 release was, to say the least, a disappointment to users interested in detailed geographic analysis.

The DHC boasts more detailed base demographic information – age, sex, race, and Hispanic origin – but also includes key information on household structure and housing. This release was subject to the same non-disclosure methodologies which resulted in such anomalies as ghost blocks, mermaid blocks, and ‘Lord of the Flies’ blocks in the redistricting file. Readers should be reminded that the Census Bureau themselves advised that you should not compute average household size by dividing the population not in group quarters by households for small geographic areas.

Over the coming weeks, rather than blindly charging forward with integration of the data into our models, we will take a more restrained approach and analyze the data every way we can. Our findings will be reported in some detail over the coming weeks, and since we expect to find significant issues with the data, will discuss our plans to improve the usability of the census at the block and block group levels. We would advise users to exercise caution given the inconsistencies in the redistricting release which can only be exacerbated in these more detailed tables.

It is our position that without the application of sophisticated spatial and matrix analytics tools to clean and normalize the data, it is not suitable for use for small area geographies such as block groups, blocks, or even within radius or drivetime trade area studies.

In the interim, we do note that the DHC release provides critical and accurate data at the county level which provides a superior baseline than does the sampled ACS product. As useful as the ACS is, it must be remembered that it is based on a spatially stratified sample that may not include respondents in many block groups in any given year. The integration of the DHC with the ACS data is a significant undertaking, but one which provides a baseline for the remainder of the decade.

The collection of tables of differing content detail, released at different geographic scales, will make it possible to create a set of block and block group tables that will anchor the ACS over the coming years. Over the coming weeks, we will begin to balance and spatially restructure the data in preparation for integration into the 2023B release coming this fall.