I’m only keeping people who worked in civilian labor force last year.
in the 2019 1 year file WKSWORK1 is all in 1/52 as you would expect. But in the 2019 5-year file around 80% of these cases have WKSWORK1 equal to 0.
I’m assuming this is because the question was changed from intervals in the 2015-2018 ACS to a direct count of weeks in 2019. But then why make the variable available on the 5 year file? When working with 5 year files I guess it’s on the user to confirm the question is unchanged over the period? Maybe everyone else knows this, but seems like it’d be easy enough to just not make available on the 5 year files.
When downloading directly from the Census Bureau the 5 year PUMS file does provide WKWN correctly in values 1/52 through the 5 year file.
Since the universe for WKSWORK1 is persons aged 16 and older who worked last year, there are no reported values of zero hours in the data; the range of valid responses for this variable is 1-52. As noted on the codes tab for WKSWORK1, all 0 codes can be interpreted as N/A. I confirmed that the frequency of N/A values is the same for WKSWORK1 in the IPUMS 2019 5 year sample and WKWN in the Census Bureau 5 year PUMS file (you can include the unharmonized version of the variable, US2019C_WKWN, in your data extract; you can identify unharmonized versions of variables are listed on source variables tab (see WKSWORK1 as an example).