Interpretation of "N/A" (00) value for in-universe cases for place of work (PWSTATE2)?

USA
1

We are working with 2005-2019 ACS samples to estimate the number of workers in California covered by the state’s workers’ compensation system, and we are considering using the place-of-work variable (PWSTATE2) to identify out-of-state residents who work in California and therefore would be covered by workers’ compensation.

We were surprised to find that a non-trivial number of people who should be in-universe for this variable (PWSTATE2) have a value of 00 (“N/A”). For example, if we look at California residents aged 16-79 who were employed last week, we estimate that 2.9 percent of respondents over 2005-2019 had “N/A” as the state of work.

Here’s the Stata output (sorry for alignment issues, can’t find monospace font). Note that the extract was limited to ages 16-79, so this condition is not in the stata command:

. tab pwstate2 if empstat==1 & statefip==6 [iw = perwt], sort

               Place of work: state |      Freq.     Percent        Cum.

----------------------------------------±----------------------------------
California |253,107,729 96.59 96.59
N/A | 7,693,328 2.94 99.52
Nevada | 177,629 0.07 99.59
Arizona | 91,879 0.04 99.63
Texas | 81,291 0.03 99.66
…

The universe for PWSTATE2 should include everyone in this tabulation, I think:

ACS, PRCS: Persons age 16+ who worked last week.

I was unable to find any clarification of what “N/A” means here and why it would be so widespread for people in the universe for this question.

Can you provide additional detail on what situations (besides out-of-universe) lead to a value of “N/A” for PWSTATE2, and why there are so many responses with this value among people who should be in-universe?

Is it possible that people who did not respond were recoded as “N/A?” I ask this because I noticed that there are no cases in this extract with a value of 99 (“Not Reported”).

2

From the comparability tab of PWSTATE2, “In some cases, PWSTATE2 is not specified, due to sampling design and/or confidentiality requirements. For example, the state is not identified when the PUMA or county group containing the workplace crossed state lines.” In cases where PWSTATE2 is N/A, PWCOUNTY also N/A (and equals zero). However, there are also cases in which PWCOUNTY is not identified but PWSTATE2 is identified. You can read more about requirements for identifying PWCOUNTY in the variable’s description. The IPUMS USA team will look into differentiating in the PWSTATE2 codes between NIU respondents and those whose state of work was not identified in the public use data.

3

Thank you! This answers my question & totally makes sense.