Survey Methodology #2006-02 - Use Of Dependent Interviewing Procedures To Improve Data Quality In The Measurement Of Change - U.s. Census Bureau Page 5
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
253.1. SIPP’s pre-2004 use of DI
Throughout its twenty-year history prior to 2004, SIPP always relied heavily on dependent
questions for its “control card” material – i.e., the household roster, and the demographic
characteristics of household members – but such procedures were never prominent in the main
body of the SIPP questionnaire. In the survey’s early panels this was in part a function of its
paper-and-pencil interview mode. Computer-assisted interviewing (CAI), which was introduced
in the 1996 SIPP panel, is much more conducive to a smooth and accurate administration of
dependent questions than is the case with a paper instrument (Brown, Hale, and Michaud, 1998;
Corti and Campanelli, 1992). Nevertheless, despite the expanded potential for the use of
dependent techniques with CAI, neither the 1996 SIPP panel nor those that followed made much
more use of such questions than did their paper-and-pencil predecessors. In the 2001 SIPP panel,
for example (the most recent SIPP panel before 2004), some key subject-matter areas, such as
health insurance coverage, did not use any dependent procedures; each wave of the survey asked
about health insurance coverage without any reference to past reports. In fact, the survey’s
approach to asking about jobs and businesses after the initial interview wave comprised almost
the complete set of truly dependent questions used in the 2001 panel: “Last time I recorded that
you [had a job with/owned a business named] [employer name/business name]. Do you still
[have that job/own that business]?”
Other areas of the 2001 questionnaire employed dependent-like questions, reminding respondents
of their prior reports, but then simply repeating a wave-1-style question regarding the current
wave: “Last time I recorded that you received Supplemental Security Income (SSI) payments.
st
Did you receive any SSI payments at any time between [MONTH 1] 1
and today?” These
dependent-like questions offer one clear advantage over the more standard form of dependent
questioning: they are simple to implement, because they do not require any restructuring of the
initial questionnaire beyond the simple addition of the “Last time I recorded...” introduction. A
major drawback of the form, however, is that it only weakly anchors the respondent’s current
report in the past, and does little to actually invite consideration of whether a characteristic has
continued in that past state or changed. Simply putting the information from the prior interview
“out there,” without tying it to the response task, still leaves the respondent focused primarily on
the immediate reporting period, probably to about the same extent as a non-dependent question.
3.2. Development of new DI procedures
Motivated primarily by concerns about increasing nonresponse/attrition, and by a desire to bring
known data quality problems under better control, in the mid-1990's the U.S. Census Bureau
launched the SIPP Methods Panel project, a research and development program to create an
improved SIPP questionnaire for implementation in the new SIPP panel to begin in 2004. For the
most part, the Methods Panel’s efforts focused on “interview process” improvements that would
yield a less burdensome interview. Another goal, however, was to improve data quality, one
specific aspect of which was a reduction in seam bias. This became an important focus of the
Methods Panel when a comparison of seam bias in the 1993 and 1996 panels confirmed that the
CAI procedures introduced in 1996 had not had any impact on the problem (Moore et al., 2004).
-4-
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