How representative are ACS and CPS data for Immigrants by skill/race/ethnicity?

Dear IPUMS staff,

Hi, I have two questions about the extent of representativeness of ACS and CPS.

My understanding is that CPS is representative at the national level and it is representative at least at the state level when it comes to the sub-national level. Is this correct?

Another question is about representing the minority group I am working on: immigrants by skill/race/ethnicity. I was wondering if CPS data is as representative as ACS for immigrants by skill/race/ethnicity at the state level. Are there any documents published by the IPUMS team that I can check?

Always, thank you for your help, IPUMS team!


You are correct that CPS is representative at the state-level, though once you start to look at highly specific sub-populations, you should always look at your unweighted sample size to ensure it isn’t too small. In practice, what will happen is the sampling error around estimated statistics will be relatively large and will, therefore, limit any informative interpretation from the data. Note that even in these cases (particularly in the CPS basic monthly data where there are no replicate weights or design variables), the weights alone do not account for the increased sampling error.

One way to increase the sample size of your estimates is to pool multiple samples together (e.g., across various months if you are using the basic monthly samples or various years if you are using the ASEC samples in IPUMS CPS). The ACS data in IPUMS USA provide the multi-year files for this purpose. 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 (note that this is not necessary in the multi-year ACS files as this has already been done).

The ACS PUMS are a larger sample, so I would generally assume they will be more representative for smaller subpopulations. However, the CPS is a labor force survey and the interview is conducted by a trained enumerator rather than a form completed online; depending on your question, these differences may be important to consider. I am linking to a Census Bureau page summarizing differences between the CPS ASEC and the ACS that may be helpful.

Dear Kari,

Thank you so much for your detailed and comprehensive answer!
Your answer helped me a lot in understanding CPS’ extent of representation when I look at (highly specific) sub-populations and in exploring possible ways to navigate the next step in my data analysis!

Thank you very much!


1 Like