Differences in ACS and CPS/ATUS sampling frame


I am trying to examine some measures from the ACS and ATUS which have known differences to highlight and explain those differences. I have noticed that the CPS/ATUS appear to sample from the set of all civilian, noninstitutional households. By contrast, the ACS includes institutions and the armed forces.

Would excluding group quarters from the ACS bring these sampling frames closest together? I could also apply information on ACS respondents employed in the armed forces, but I cannot tell if this would be an appropriate restriction.


I believe I can answer my own question! From the IPUMS-CPS sample design documentation:

“To achieve comparability with IPUMS-CPS, IPUMS-USA users should exclude persons coded as 1, 2, 3, 4, and 6 on the IPUMS-USA variable GQTYPE.”

I would like to verify that this holds for the ATUS-X as well, but I believe it does.


You are correct about excluding institutional group quarters and military group quarters in IPUMS-USA (GQTYPE=1, 2, 3, 4, or 6) to achieve a comparable sampling frame with IPUMS-CPS. Since the ATUS draws a subsample from the CPS, you can use the same method of excluding institutional group quarters and military group quarters from your ACS sample.

Hope this helps.