ACS Data Users Group

Hi, we're working with a dataset that provides estimates for both 2010 and 2020 tracts with no indication of which boundary is being used. We've noted that there are some instances where a tract boundary changes between decennial censuses, but the tract ID is unchanged.

For an example - if you go to our website and look up tract ID 011000 in Grand Forks, ND (full FIPS = 38035011000) and compare it between years 2019 and 2020, you'll see a pretty significant boundary change: www.cityhealthdashboard.com/.../metric-detail

Can anyone point me to any resources about why this might be occurring? My understanding was that tract IDs and boundaries should be relatively stable, unless a split or merge occurred.

Taylor, this comes as no surprise to me. Whatever the stated intentions of the stability of tract definitions, there are in fact many, many adjustments to tract boundaries with every cycle that are neither splits nor merges but rather some form of "transfer", shifting part of one tract over to a neighbor. You can determine this from the Census Bureau's tract relationship files, where you'll find many cases of small "parts" associated with one tract in one year and a different tract in a different year. Most such parts are very small, but many are quite large. Large changes often also involve tract ID changes, but unfortunately, not always.