I’m working with a dataset that crosses 2020 with data from 2015-2022. What is the best practice for geocoding within a dataset that spans two census' and multiple ACS years.
Should I be benchmarking to look at a single year snapshot for each ACS product/year or should I be benchmarking on the decennial census alone for Pre 2020 (2010) and post 2020?
I know that boundaries change between decennial census’ and when going to connect to tract level ACS data I will want to have the correct tract for correct 5 year estimates I’m going to get.
For those who are curious my code for 2021 results is here though likely not required for answering the question.
Thanks, Rez
US Census got back with me after two weeks:
"American Community Survey (ACS) data uses boundaries from the last year of the 5-year estimate. Therefore, the 2015-2022 dataset will use 2020 tract boundaries (tracts are only updated every 10 years). It will therefore be consistent with your 2020 Decennial Census data."