|
Activity Number:
|
357
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #303976 |
|
Title:
|
Simultaneous Calibration and Nonresponse Adjustment
|
|
Author(s):
|
Eric V. Slud*+ and Yves Thibaudeau
|
|
Companies:
|
University of Maryland College park and U.S. Census Bureau
|
|
Address:
|
4600 Silver Hill Road, Washington, DC, 20233,
|
|
Keywords:
|
survey weights ; generalized raking ; weight adjustment ; population controls ; weight truncation ; linearization
|
|
Abstract:
|
Single and joint inclusion probabilities are generally known in complex surveys up to the point where survey weights are modified for nonresponse and population controls. Best practice by survey practitioners usually includes weight adjustment, first by calibration, ratio adjustment or raking to correct for nonresponse; next by further steps to impose population survey controls; and often also by final steps involving weight truncation or cell-collapsing to constrain the modified weights so that large and small weights do not differ by too much. These adjustments are made in successive stages, the order of which may differ from one survey to another. In this paper, generalized calibration is adapted to make all these weight adjustments in a single stage, after which a linearization-based large-sample variance formula is available. The method is illustrated on SIPP 1996 first wave data.
|
