|
Activity Number:
|
293
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #308908 |
|
Title:
|
Using Callback Models To Adjust for Nonignorable Nonresponse in Face-to-Face Surveys
|
|
Author(s):
|
Paul Biemer*+ and Kevin Wang
|
|
Companies:
|
RTI International/The University of North Carolina at Chapel Hill and RTI International
|
|
Address:
|
PO Box 12194, Research Triangle Park, NC, 27709,
|
|
Keywords:
|
paradata ; latent class analysis ; drug survey ; weighting class adjustment
|
|
Abstract:
|
Traditional methods of adjusting for survey nonresponse bias assume nonresponse is "ignorable." Biemer and Link (2006) propose an alternative approach that relaxes this assumption by using so-called level of effort (LOE) variables derived from call attempts to model the response propensity. Using a latent indicator variable, the model distinguishes between sample members who will eventually respond to a survey with sufficient call attempts and those that will never respond regardless of the number of call attempts (i.e., the "hard core" nonrespondents). They applied their callback models to a large RDD survey. The present paper applies a similar model to a face to face survey, viz., the National Survey of Drug Use and Health (NSDUH). The nonresponse bias reduction of this approach is assessed and compared with that of the traditional approach as implemented in the NSDUH.
|
