I am new to CPS and attempting to understand a bit more about how the county variable may be used. I’m interested in looking at a large county that is part of a large metro area. Given that estimates are generally not intended to be reliable at smaller geographies than the state, I am wondering 1) if samples or supplements may be pooled across years for a larger N at the county level and if not, 2) in what case it may be appropriate to use estimates at the county level.
Welcome to IPUMS CPS! To address your specific concerns in order:
-
Samples can be pooled across years for a larger sample size at the county level. You will need to adjust your sampling weights to account for the pooled datasets due to the fact that sample weights are different each month (see this forum post for more details). It is also worth confirming that the county boundaries did not change among the years you are pooling for the county you are looking at, as that would create a situation where pooling at the county level is not appropriate.
-
The CPS is stratified at the state level (you can read more about CPS sample design in the Census Technical Paper 66, starting on page 20). Therefore, the CPS is not designed to generate sub-state level estimates. That being said, the March 2018 ASEC codebook (page 2-8) provides insight into geographic limitations of the CPS and cautions against estimates for smaller metropolitan areas but notes that large metropolitan areas like NYC or LA should generally be fine. So, in answer to your question, analysts may determine whether it is appropriate to use estimates at the county level for their specific use case or if they pool enough adjacent months or years of data together.