ACS Industry Totals vs CPS Industry Totals


I know this is the ACS forums, but we have an ACS vs. CPS question. So I wasn’t sure where to post it! :slight_smile:

Without going into too much detail, I’m looking at the number workers in a specific industry in ACS 2019 vs the same industry in CPS 2019 - all in 1 state.

I’m getting very different numbers in each, by a large margin. For example:

ACS 2019 [Industry A]: 12,000 workers
CPS 2019 [Industry A]: 7,000 workers

I understand that ACS will inherently have more observations - but why would we expect that the projected total numbers (accounting for PERWT) would be so different across the two surveys?

If there any good links/readings on why this may be happening - I’d appreciate it!

Hi Cela,

If you could tell me which industry and state you’re looking at, I might be able to provide more targeted support on documentation and potential data sources. Generally speaking however, estimates of a small subpopulation using the CPS & ACS will have a large variance due to fewer observations being drawn from it. In your case, it seems the issue is coming from trying to produce statistics on either too small of an industry or state. There are also a few differences between the surveys, such as the ACS including the institutionalized population, which are discussed in this fact sheet.

One way to increase the sample size of your estimates is to pool multiple samples together (i.e. across multiple years of the ACS and CPS ASEC). This will increase the number of observations in your data and the statistical precision, but will limit the temporal precision of your analysis. Note that if you do pool together multiple samples you will need to adjust the sampling weights so that they properly account for the combined samples. An approximate way to do this is to divide the sampling weight by the number of samples you are pooling together. Another option to increase statistical precision is to aggregate multiple similar industries together within a given year.

You might also be interested in using estimates from the BLS’s Current Employment Statistics surveys rather than the ACS or CPS. Though technically smaller than the ACS, this survey of employers may provide a wider and more representative picture of employment on the industry level.