ACS Data Users Group

Hello,

I wanted to download the census block group level data of table B08013: AGGREGATE TRAVEL TIME TO WORK (IN MINUTES) OF WORKERS BY SEX. According to 2019 ACS Detailed Table Shells: https://www2.census.gov/programs-surveys/acs/summary_file/2019/documentation/user_tools/ACS2019_Table_Shells.xlsx (You need to download it to see as preview only shows the first 5000 rows), the table should be available in both one-year and five-year estimates. I want the census block group level data, so I plan to use the five-year estimates. I tried both https://data.census.gov/cedsci/ and IPUMS NHGIS, but B08013: AGGREGATE TRAVEL TIME TO WORK (IN MINUTES) OF WORKERS BY SEX is only available at the census tract level in ACS 2015-2019 5-year estimates. I am really confused, as I thought ACS 2015-2019 5-year estimates should have tables at the census block group level.

Any thoughts would be greatly appreciated. Thank you.

Most of the detailed ACS tables are not available for block groups. The summary file Appendix A, which you can find through the Appendices links on the sequence-based summary file pages (here's 2019's page), indicates which tables have "No Blockgroups" [sic] and which are "Migration Only" or "Place of Work Only". The latter types (of which B08013 is one) also provide no block group data.

Some extra explanation for why this table has no block group data: the ACS asks workers where they work by place and (I think) minor civil division. The ACS tabulates "place of work" tables by counting the number of workers in their reported place of work. It's not possible to identify specific block groups this way... There's no direct relationship between the reported places and minor civil divisions and block groups, so it's impossible to tabulate by block group place of work. Similarly, these tables have no tract data.

NHGIS breaks up the 5-year ACS data into 4 datasets specifically to distinguish these 4 key types of tables: those with BG data, place of work, migration, and all others. If a table isn't in the NHGIS dataset "a", it doesn't have block-group-level data.

I misstated something in my last post because, unfortunately, the appendices are also not completely accurate!

Appendix A indicates that B08013 has a geographic restriction of "Place of Work Only", which I initially interpreted to mean only the levels for which the ACS tabulates counts by workplace. But B08013 is not a workplace-geography table, and therefore it does provide census-tract level data.

FWIW, to determine geographic availability by table for NHGIS datasets, we've stopped relying on the Appendix almost completely. Instead, we systematically examine the contents of the summary files to determine which levels are available for each table. Therefore you can rely on NHGIS (even more so than the appendix) to determine which tables have block group data and which do not.

Hello Jonathan,

Thank you so much for your detailed reply. I was wondering how to determine whether a table is residence-based or workplace-based. In ACS 2019 subject definitions page 95 (https://www2.census.gov/programs-surveys/acs/tech_docs/subject_definitions/2019_ACSSubjectDefinitions.pdf), it says that "If you are interested in the number and characteristics of workers living in a specific area, you should use the standard (residence-based) journey-to-work tables. If you are interested in the number and characteristics of workers who work in a specific area, you should use the workplace-based journey-to-work tables." I want to know how to determine whether a table is residence-based or workplace-based.

Initially I thought I should just check the appendix (https://www2.census.gov/programs-surveys/acs/summary_file/2019/documentation/tech_docs/ACS_2019_SF_5YR_Appendices.xlsx), and if the geography restriction is "Place of Work Only", then it's a workplace-based table. But your most recent reply just says that even Appendix A indicates that B08013 has a geographic restriction of "Place of Work Only", B08013 is not a workplace-geography table. Therefore, I'm lost as to how to determine whether a table is residence-based or workplace-based.

I was wondering if you could help me with this. Thank you so much for your patience.

"Therefore, I'm lost as to how to determine whether a table is residence-based or workplace-based."

In general, the tables that summarize by workplace have the phrase "for workplace geography" in their titles. Accordingly, the workplace version of B08013 is "B08536. Aggregate Travel Time to Work (in Minutes) of Workers by Means of Transportation to Work for Workplace Geography"

In NHGIS, these tables stand alone in the "d" dataset for 5-year ACS data. You could find all of the 2015-2019 5-year workplace tables by using the Dataset data filter and selecting 2015_2019_ACS5d: 5-Year Data [2015-2019, Summary by Place of Work]

Hello Jonathan,

Thank you so much for your detailed reply. I really appreciate your help.

I think you begged the question and it still remains. So why is there no BG data for this table since it is resident based?

"So why is there no BG data for this table since it is resident based?"

The migration and workplace tables aren't the only tables that have no BG data. Most others don't either. I believe the main reason is that BG data are simply not as reliable given the small samples by BG. The Bureau seems generally to limit BG data to those tables with relatively large categories and well represented subpopulations. E.g., the more general tables with codes that start with "C" have BG data more often than the detailed "B" tables, and you can get BG data for the general Sex by Age (B01001) table but not for the race iterations of that table (B01001A through B01001I).

Jonathan--I understand the idea of BGs having small samples but if the CB is going to suppress they have to have a reason. When there were 3 year tables there were statistical tests done to the tables to determine what got collapsed and suppressed. How is that determined here and now? We shouldn't be guessing but finding it in documentation somewhere.

Ed, I agree. I don't know if such documentation exists. It might; I didn't go looking. If others know of any official explanation of how tables are selected for BG data, I hope they'll reply here.