Error in sequence file or table shells file for 2013?

Dear All:

In preparing the data for use in Social Explorer we found this error, as well we were going to contact the Census Bureau about this this morning.

Any response from the Bureau.

Andrew Beveridge

From the Bureau--No real problem, just a documentation issue.

Hi Andy,

We appreciate you bringing this issue to us. Our staff looked into the issue that you reported. We publish table shells regardless of data, so they will not match exactly back to the 1,3, and 5-year summary file look-up files. We need to add something to the SF tech doc to explain this to data users. As for the FTP site, there is not anything wrong with the table shells posted. The three tables that you noted are published in the 2013 5-year, so you are fine using the 2013 1-year summary file sequence files and appendices.

Please let me know if you have any additional questions.

Thanks,

Nick

For what it's worth, those tables are not showing in FactFinder.

If the table shells are not specific to individual ACS data releases, why are they named as if they were?

Nick, are you saying that a table shell file released with a specific ACS data release may not apply to the data it appears to be associated with? If not, where can we obtain a definitive table shells file for a specific release? The table shell files are critical for our ACS data processing.

Looking back at the three 2012 ACS table shell files, they are all different. If what you're saying is correct, why would they be different? Why doesn't the Bureau publish a single table shells file for all three annual releases? It would be much less confusing.

The 1-year table shells for 2012 do NOT include B02005, B02006, or B02007, while the 2012 3- and 5-year table shells do.

Glenn

www.census.gov/.../ -- every ACS year listed on this page includes separate table shells files for each of the 1- 3- and 5-year releases.

Could we convene a working group for better US Census metadata publishing? The participants in this thread are a perfect seed for such a group, and our Census Reporter project also has a vested interest.

FYI, after some discussion on the SDC email list, we decided that B02005, B02006, and B02007 have been renamed (to B02014 and C02014, B02015 and C02015, and B02016 and C02016, respectively).

The continued existence of B02005, B02006, and B02007 in the 2013-1 table shells file is evidently an oversight. Delete these from your table shells file (rows 650-738) and you should be good to go.

Glenn

The Census Bureau recommends using the 1-year Sequence Number/Table Number Lookup File (www2.census.gov/.../Sequence_Number_and_Table_Number_Lookup.xls) or 1-year Appendices (www2.census.gov/.../ACS_2013_SF_1YR_Appendices.xls) as resources to see the list of tables published for the 2013 ACS 1-year data release.

The 1-year table shells include some table shells (B02005, B02006, and B02007) that are no longer published for the 1-year as noted in the 2011 ACS Data Product Changes (www.census.gov/.../). We are looking into the discrepancy between these resources and appreciate data users bringing the issue to our attention.

FWIW, the "Subject Areas" for the new B28 (Computer) series are incorrectly listed as "Health Insurance" in www2.census.gov/.../Sequence_Number_and_Table_Number_Lookup.xls

Authoritative metadata from the Census Bureau is an important component of making this data useable for application builders as well as data analysts. I'm serious about the idea of a working group, and would volunteer time to help specify what this class of data users needs.

[Updated on 9/20/2014 5:52 PM]

Joe, I'd be on board for a working group on metadata accuracy.

Another "gotcha" now that the 3-year data is out: The ACS2013_TableShells.xlsx file contains rows for the new tables B2800x, but these tables don't exist in the 3-year summary file.

Aaaand... yet another error on the Bureau website now that the 5-year data is (almost) out: The sequence/table number Excel file for the 5-year release is truncated. It's missing some 8,000 rows. (The SAS and text versions are correct.) Here's a fixed version I made:

Glenn, thanks for creating that.

I've had a pretty busy fall, and it seemed to make sense to wait until everyone has had a chance to deal with all of this year's data, so I was thinking that I'd start connecting people to talk in January.

Again, if anyone reading this is interested in gathering ideas and use cases to help promote better metadata for the ACS, send me your email address and I'll include you on whatever groupware kinda thing we set up then.

For use with the ACS 2013 5-year data release, you'll need to delete rows 34884-34907 (table C27010) from the Bureau's Table Shells file (www2.census.gov/.../ACS2013_TableShells.xlsx).