for a more detailed description.
9. Additional Notes
The following are miscellaneous additional notes regarding the 2022 HRS Core
(Early, Version 1.0). If we become aware of additional issues, they will be
9A. Households with No Family or Financial Respondents
As noted earlier in this document, the data collection design was to have asked
most questions of all respondents and some questions of just a designated
coverscreen, or family, or financial respondent on behalf of the household.
However, occasionally that is not what happened. For some households we did not
obtain an interview from a family or financial respondent. There were 380
households that had no family respondent; 189 households had no financial
respondent. There are 4 missing coverscreen respondents in this wave. The
household records for these households contain null values for the missing
information. Households missing a family or financial respondent can be
identified, respectively, by values of “Blank. No family/financial respondent”
(in the household record) in the following variables:
SPN_FAM – 2022 FAMILY RESP PERSON NUMBER
SPN_FIN – 2022 FINANCIAL RESP PERSON NUMBER
9B. Unfolding Bracket Variables and Imputations
Typically, a series of unfolding bracket questions followed a lead-in question
asking for an amount. If an actual amount was not given, a series of
“unfolding” questions were asked. The manner in which the unfolding questions
were programmed (Blaise) is different for the 2002 through 2010 data compared to
the CAI (SurveyCraft) software used for 1993 through 2000. This change was
transparent to the respondents, since exactly the same questions were asked with
the new software as would have been asked with the old software; but it did have
an implication for the data that were actually stored and also for the data that
are released.
Instead of storing the response to each unfolding question, three summary
variables were generated: the minimum and maximum values for the amount, given
the answers to the unfolding questions, and if the last answer a respondent gave
in an unfolding sequence was either “Don’t Know” or “Refused,” what that answer
was. In 2002, if the Respondent said “more than” to the unfolding question with
the highest value, then the maximum value was stored as ten times that value.
However, in 2004 and 2010, if the Respondent said “more than” to the unfolding
question with the highest value, then the maximum value was stored as 99999996.
For most analysts, those three variables (and in particular, the minimum and
maximum of the possible range) will be sufficient for analyses. For any analyst
who needs the more detailed information, it should be noted that the three
variables, combined with the information about the unfolding questions provided
in the box-and-arrow and codebook, are sufficient to allow the analyst to
reconstruct the sequence of questions asked of any respondent, and the answers
to each of those questions in many of the unfolding sequences.
For other sequences -- those in which respondents were randomly assigned to one
of three "entry" points for the first unfolding question -- the analyst will
also need to take into account a fourth variable (located in the preload
sections) that specifies the entry point for each respondent. The following
example shows the preload variable (PZ041) and the unfolding sequence that uses
the random entry point from RZ041.
16
February 2024, Version 1.0