|
Activity Number:
|
109
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, August 4, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Memorial
|
| Abstract - #301398 |
|
Title:
|
A Closer Look at the Quality of Estimates from the American Community Survey
|
|
Author(s):
|
Susan Love*+ and Deborah H. Griffin
|
|
Companies:
|
U.S. Census Bureau and U.S. Census Bureau
|
|
Address:
|
Housing & Household Economic Statistics, Washington, DC, 20233-0001,
|
|
Keywords:
|
unit nonresponse ; item nonresponse ; nonsampling error
|
|
Abstract:
|
The Census Bureau has stated that the quality of the estimates from the American Community Survey will compare favorably to the estimates produced from the decennial long form sample. Any assessment of the quality of a survey, such as the ACS or the long form sample, must take both sampling and nonsampling errors into account. Previous papers have compared the level of sampling error between the ACS and the decennial census, and demonstrated that the increase in the margin of error for 5-year averages from the ACS is modest when compared to the expected census sample sizes. This paper focuses on nonsampling errors and addresses the proposition that the ACS will be better able to control for nonresponse-related error than the decennial census sample. Two key measures of nonresponse are analyzed in this paper - unit nonresponse and item nonresponse. Unit nonresponse refers to the sample cases for which an acceptable interview could not be collected. In the ACS these are survey noninterviews. In Census 2000 these sample noninterviews refer to the instances where insufficient data were collected on census long forms for them to qualify as sample interviews.
|
