Differences between UHRSWORKORG and UHRSWORK1 in the outgoing rotation groups


I read through the very informative document “Hours Worked Variables Notes” listing all Hours of work related variables and detailing nuanced differences among the array of variables available. I also read the description pages of the different variables. If I understand correctly there are two main distinctions between UHRSWORKORG and UHRSWORK1: First, UHRSWORKORG is only asked within the outgoing rotation group framework while UHRSWORK1 is part of the regular set of questions directed to the whole monthly sample. Second, UHRSWORKORG is asked using the stipulation “at the pay rate specified” as the respondent’s hourly wage.

Assuming these are correct I would expect UHRSWORK1 to always be equal are bigger than UHRSWORKORG because UHRSWORK1 should report hours worked at rates other than the one specified in “Hourwage”. And yet, in years following the 1994 redesign, after selecting only the ORG sample, looking at the differences in values between these two variables it becomes apparent that often UHRSWORKORG is bigger than UHRSWORK1 (in years prior to 1994 values are equal).

Did I misunderstand something? Are my assumptions wrong? If not, how can this be and which of the variables better measure hours of work in main job with the time frame is unspecified? Which should I use as the primary choice for measuring hours of work for the whole sample (not only those paid by the hour)?

Your answer will be much appreciated and thanks a lot for time and effort!


You are correct that we would expect UHRSWORK1 to be greater than or equal to UHRSWORKORG. Looking at a random sample of post-1994 IPUMS-CPS data, this appears to be true for approximately 98% of cases. In the small number of cases that do not match our expectations, I do not see an explanation in the CPS documentation. This leads me to believe that these cases can be explained by inconsistent information provided by the respondent or by a data recording error. These questions do not specify a timeframe for “usual”, which could have also led to some of the inconsistency we see between these variables. We generally report the CPS data as it appears in the original data source; therefore, any data issues present in the Census Bureau’s data will also be present in the IPUMS data. Thus, you might consider contacting the Census Bureau directly for further insight.

Sorry that I am not able to provide a more direct answer.