Using 2010–2020 Crosswalk Files for Intercensal ACS?

I am interested in harmonizing data at the census-tract level for the 5-year ACS data (ACS 2006–2010, ACS 2007–2011, …, ACS 2017–2021).

I understand IPUMS recommends using census block group for ACS crosswalks, but I wanted to ensure I am using the correct years. It is my understanding that changes in the Census Bureau’s geographical designations were negligible from 2010 to 2019, so I can assume ACS 2006–2010, …, ACS 2015–2019 census block groups correspond to those from the 2010 census. Moreover, I understand ACS 2016–2020 and ACS 2017–2021 census block groups correspond to those from the 2020 census. For these reasons, I should use the census block group crosswalk files from 2010 to 2020.

Is my understanding accurate?

Thanks so much for your help! I greatly appreciate it.

Yes, that’s all correct. The NHGIS crosswalks page provides more information here about the small number of cases where block group definitions changed between 2010 and 2019.

1 Like

Thanks for your response. I have one additional question.

In the 5-year ACS data, I see unique IPUMS names are used for each column, rather than the table names. It seems quite tedious to manually rename each column for consistency across ACS samples.

Is there any easier way to combine ACS samples using Stata, Python, or R? (I know about the integrated time series, but they lack some of the samples I need.)

Again, thanks so much for your help!

Unfortunately, I don’t know of a simple existing short-hand approach. To help, I’ve created this lookup file listing the NHGIS table code that corresponds to each ACS table code for each NHGIS 5-year ACS dataset.

FWIW, one thing we’re working on right now–hoping to release in the next few months–is extending our time series collection to include data from all 5-year ACS periods since 2006-2010 and to include ACS data in more time series tables, particularly those that are currently restricted to decennial data sources (e.g., Persons by Race). We also have longer term plans to provide the original ACS table codes in our files, but there’s no timeline for that yet. I.e., we’ve heard of this concern before, and we’re working on it!

1 Like