Survey Methodology #2006-02 - Use Of Dependent Interviewing Procedures To Improve Data Quality In The Measurement Of Change - U.s. Census Bureau Page 6
ADVERTISEMENT
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25Armed with evidence that the need to address this measurement error issue remained as strong as
ever, we designed and implemented new procedures to reduce seam bias, and tested them in the
Methods Panel’s experimental SIPP questionnaire. Because of the positive track record of DI
with regard to reducing spurious change, our main focus was on strengthening those procedures
in SIPP, which we did in the following ways:
(1) We expanded the survey’s reference period to include not only the four preceding
calendar months, but also the current month, up to the date of the interview. This had been a
feature of many of the questions in the 2001 SIPP panel; in the new questionnaire it became
universal when asking about the timing of events. The more fundamental change, however, was
to store the interview month information and carry it forward to the next interview wave, where it
is featured prominently in the logic used to decide whether to ask a dependent question, and, if so,
the specific form of that question. As noted above, the interview month (“month 5”) of one
interview wave is the first month of the next wave’s reference period. Before 2004 the SIPP
questionnaire had not exploited the fact that when an interview month event is reported, a basic
fact about the next wave’s four-month reference period is already known.
(2) We framed the new questionnaire’s dependent questions in truly dependent language,
explicitly linking the current wave report to what is known from the last interview, and focusing
the cognitive task on whether or not the prior circumstances did or did not continue on into the
current wave. The concentration on whether something continued from one interview’s reference
period to the next also led us to restrict the range of circumstances under which a dependent
question would be asked. In SIPP’s 2001 questionnaire, an event that occurred in any month of
the previous interview’s reference period was sufficient to trigger a dependent-like question in the
next wave – even if the event happened only early in the previous interview’s reference period,
1
and was no longer appropriate to the notion of “continuing” into the next reference period
. The
new instrument, in contrast, concentrates on what was happening as of the end of the previous
reference period. Thus, only the previous interview’s months 4 and 5 – “last month” and the
interview month – are used to determine whether to ask a dependent question. Events that
happened only before those months trigger a non-dependent question in the subsequent interview
wave, with no mention at all of pre-month-4 events or characteristics.
More specifically, we instituted the following new procedures, with some slight variations,
throughout the 2004 SIPP questionnaire:
– An event or circumstance reported in “month 5” of the prior wave (i.e., the first month of
the current wave’s reference period), triggers an initial question in the next interview
wave to confirm that prior event or circumstance: “Last time I recorded that you [received
X, were enrolled in school, had health insurance, etc.] in [MONTH 1]. Is that correct?” If
the respondent confirms the prior wave report, then his/her status for the current reference
period is confirmed; a later question fills in the details about the remaining months of the
reference period. If the respondent does not confirm the prior wave report, then the
1
The 2001 instrument’s dependent questions about prior-wave jobs and businesses were an exception to this rule,
and in fact closely mirrored the procedures implemented in the 2004 questionnaire for other subject-matter areas.
-5-
ADVERTISEMENT
0 votes
Related Articles
Related forms
Survey Methodology #2012-02 - New Relationship And Marital Questions: A Reflection Of Changes To The Social And Legal Recognition Of Same-sex Couples In The U.s. - U.s. Census Bureau
Legal
Early Childhood Program Participation, From The National Household Education Surveys Program Of 2012 - U.s. Department Of Education
Legal
Form Hud-9887 And 9887-a - Document Pachagefor Applicant's/tenant's Concent To The Release Of Information - U.s. Department Of Housing And Urban Development
Legal
Related Categories
Parent category: Legal