Using 2010-2020 Block group crosswalk

NHGIS
1

Hello,

I’m trying to build tract information from 2022-2018 ACS, 2021-2017ACS, 2020-2016 and 2015-2019 ACS on the share of people who work from home.
I want the data to be in at the 2010 block group level. I have a couple of questions:

  1. The instructions in the crosswalk page show steps to go from old to new block groups. is it accurate to use the same weights to go from new to old block groups? I think the answer is no, could you please help me understand if there is any way to build an alternative crosswalk from 2020 to 2010?
  2. I tried using the crosswalk to transform 2015-2019 block group level data to 2020 block group level data but when I merge the 2015-2019 ACS data with the crosswalk I find 216 block groups that do not match (72 only in the crosswalk and 144 only in the ACS data- Notice that I’m using all USA data, not including Puerto Rico) Why does this happen?
  3. Why are there so many entries in the crosswalk that have zero as a weight?

Many thanks for the clarifications you can provide.

2

Answering each of your questions…

  1. NHGIS provides separate crosswalks for allocating data from 2010 to 2020 units and from 2020 to 2010 units. To allocate from newer (2020) block groups to older (2010) block groups, you can use the latter crosswalks. (You’re correct that it wouldn’t be valid to use the weights in the 2010-to-2020 crosswalk to allocate from 2020 to 2010 units.)
  2. Unfortunately, the block groups identified in ACS data releases between 2011 and 2019 don’t match completely with the original 2010 block group definitions. As explained on this section of the NHGIS crosswalks page, the Census Bureau made a few changes that affect the block group codes in a few counties. These changes likely explain the mismatches that you found. We don’t have a guide to how to handle these cases, but in the section I linked to, there are references to Census Bureau documentation about the relevant changes.
  3. The block group crosswalks include a record for every spatial intersection between one year’s block groups and another year’s, according to the Census Bureau’s official boundary definitions. The weights in the crosswalks are not based solely on the area of overlap, though. Rather, we use more advanced models, mainly based on the block-level population and housing density in each area of overlap. There are many intersections where the block-level housing and population counts are zero, which results in weights of zero in our crosswalks, even though the areas of the intersection are non-zero.