ACS Subjects not available for block groups?

Hi Ian,

Indeed, there are a number of tables for which block group data is not available. The reason for this (as far as I know) is that there's not a large enough sample at that level. In other words, the margin of error would be unacceptable.

In the past, I've found FactFinder as a good resource to see which tables do and don't contain block group data, but when I looked last week, it was not working correctly.

A search for "Population in Group Quarters" only shows it available for 2010. See, for example, 

https://factfinder.census.gov/bkmk/table/1.0/en/DEC/10_SF1/P29/0500000US01001.15000

That's the 2010 Decennial Census (not ACS). That may be a good alternative if you need block group data. I imagine they were able to include this data at the block group given that this is full census data, not sample data.

Within FactFinder, if you filter by block group FIRST, searching for "group quarters" won't bring up the ACS data, since none exists at the block group level.

Hmm yeah that's what I was thinking. The sample size explanation makes sense. If I look at different tables though, like B09019 "HOUSEHOLD TYPE (INCLUDING LIVING ALONE) BY RELATIONSHIP", I see there is a total for "In Group Quarters".

factfinder.census.gov/.../productview.xhtml

In my project, I'm trying to recreate a Social Vulnerability Index (originally created with tracts) for block groups. Documentation if interested: svi.cdc.gov/.../SVI2014Documentation.pdf.

The SVI documents what ACS tables it used, so I'm thinking that while I may not be able to use those exact tables, the same estimates might exist but housed in different tables.

if you look at the summary file documentation - the Appendices excel file found here
www.census.gov/.../summary-file-documentation.html

You will see the tables listed along with geography restrictions. Those that are not available by blockgroup have "No Blockgroups" listed as the geography restriction.

And yes, I believe the reason some are not available by block group would be due to too small of a sample size.

Thanks JamiRae and Bernie , very helpful. I was able to get (most of) the estimates I wanted through other tables. For example, there was no direct count of persons with disability for some reason, but there was Table B21007 "Age by Veteran Status by Poverty Status in the Past 12 Months by Disability Status for the Civilian Population 18 Years and Over" which you can sum to get the number of persons with a disability.

This is kind of hacky, and makes me wonder why? If sample size was the issue, having a much more specific count available like B21007 but not the broader "Persons with a disability" set doesn't make much sense. You would think only populations with the broadest pools would be counted for block groups.