|
Activity Number:
|
387
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301314 |
|
Title:
|
Responsive Design for Household Surveys
|
|
Author(s):
|
Steven G. Heeringa*+ and Robert Groves
|
|
Companies:
|
University of Michigan and University of Michigan
|
|
Address:
|
Survey Research Center, Ann Arbor, MI, 48104,
|
|
Keywords:
|
survey error ; survey cost ; process data ; sample replication ; response propensity ; two-phase sampling
|
|
Abstract:
|
For more than 70 years, the development of sampling techniques and data collection methods for household surveys has focused on reducing costs and errors. Historically, survey designers have relied heavily on experiences in prior surveys to model the error and cost structures for a new design. Despite the advantage that existing information brings to the survey planning process, many survey designs do not incorporate features that anticipate the uncertainty inherent in the essential survey conditions. Very few are designed to respond to real-time data with the possible exception of a crisis response to major failures in estimating survey design parameters or data collection costs. The transition to computerized data collection and survey management systems has presented survey designers and practitioners with a wealth of new real-time data on the survey process. These data include not only traditional sample design and production indicators but detailed data on the work progress and associated costs for distinct sampling units, individual interviewers and even for individual sample lines. This paper presents a new model for survey design and management.
|
