|
Activity Number:
|
6
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 2, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Government Statistics
|
| Abstract - #303161 |
|
Title:
|
Using the Fraction of Missing Information to Monitor the Quality of Survey Data
|
|
Author(s):
|
James Wagner*+
|
|
Companies:
|
University of Michigan
|
|
Address:
|
1248 Thompson G398 Perry, Ann Arbor, MI, 48106,
|
|
Keywords:
|
nonresponse ; paradata ; responsive design ; adaptive design
|
|
Abstract:
|
The response rate has been a key indicator in judging the risk of nonresponse bias. Recent investigations have shown that the response rate is not always a good indicator for nonresponse bias. This paper proposes indicators that rely on more of the available data than just the response indicator. Under an assumed model relating frame data and paradata to the observed survey data, two statistics can be developed that are useful monitoring tools. The first is the difference between the complete case mean and an adjusted mean (either based on imputations or nonresponse weights). The second is the fraction of missing information. This statistic quantifies our uncertainty about the values we would impute for the nonresponding units. These monitoring tools may lead to improved data collection practices. Examples from a personal visit survey are demonstrated.
|
