2020 Census Tract Relationship Files

Hi everyone,

Apologies for this simple question, but when will the 2020 census tract relationship file be released? Specifically, I'm looking for the proportion of the 2010 census tract population that falls within 2020 census tract boundaries.



  • I don't know when the tract files will be released (and it seems no one else on this forum does either!), but I can offer a couple other suggestions.

    First, you might like to try posing your question on the U.S. Census Bureau's Slack workspace, accessible through their Developers page. Although it's framed as a "Developers' Forum," there are all kinds of Census-related questions asked there, and Census staff respond pretty regularly.

    Second, you might be interested in using the block-to-block crosswalks from IPUMS NHGIS. Because blocks nest in tracts, and tract IDs even form the first part of block IDs, it's relatively straightforward to generate tract crosswalks from block crosswalks. There's also some guidance on the NHGIS page about how you could generate (e.g.) 2010 estimates for 2020 tracts directly using the block crosswalks, along with an explanation for why that's generally more accurate than directly allocating data from one year's tracts to another's.

    Full disclosure: I'm the lead developer for these crosswalks. As such, I can also say that we're working on extending NHGIS time series to include 2020 data for 2010 tracts, which we hope to release in a month or two. Sometime after that, we'll also work on creating time series including 2010 data for 2020 tracts.

  • Just a quick note, Social Explorer www.socialexplorer.com has Census files from the 2010 and 2020 Censuses by summary level that have results of both Censuses in a form that makes it possible to map or get reports or download data, that has combined both sets of data and allocated the 2010 data to 2020 geographies using the Census Bureaus block relationship files.  Currently we have a selection of Tables 2 and 4, Total Population with Hispanic, plus Non-Hispanic Single race, plus all multiple race for a selection of summary levels down to Census tract.  These data were used on release day for coverage of the Census in the NY Times see:   https://www.nytimes.com/interactive/2021/08/12/us/2020-census-race-ethnicity.html?searchResultPosition=2

    Early next week, we plan to release a wider array of 2010 and 2020 combined Census data  down to the Census Block Group, and include all the tables in the PL94 release, except we have combined the 3 or more race into on category.  Seven day trials are freely available.

    We include the following note:

    The data reported  here for the 2010 Census may not exactly match the data as originally reported in 2010. To make the 2010 data comparable and place it on the 2020 boundaries the 2010 block population and characteristics were allocated to the 2020 blocks.  The   2010 data  were then aggregated to the 2020 various geographies. Because of this the population and characteristics may be different than those reported officially by the Census Bureau in 2010.  There are several reasons for this 1) Some of the boundaries were redrawn due to annexation or in some case remapping and correction of the previous boundaries;  2) The data allocation was based upon the land area occupied so there may be some slight differences injected because of that. However, this method does make it possible to directly compare at various levels of geography the changes between 2010 and 2020 because it uses the same areas.  Among the areas that officially changed boundaries between the 2010 and 2020 Census were most of blocks, tracts, and block groups, as well as some counties, minor civil divisions and places.  Except for annexation, remapping and other changes in the actual geography of the areas, the other differences are minor. 

    You should also note, that the 2020 data has had noise injected into it because of the Census's new Disclosure Avoidance System (DAS).  The is can mean that population counts and characteristics, especially when they are particularly small, may not exactly correspond to the data as collected.  As such caution should be exercised when examining areas with small counts.  Ron Jarmin, acting director of the Census Bureau posted a discussion of the redistricting data, which outlines what to expect with the new DAS.  For more details on accuracy you can read  it here. https://www.census.gov/newsroom/blogs/director/2021/07/redistricting-data.html