|
Activity Number:
|
191
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #305852 |
|
Title:
|
2010 Census Coverage Measurement: Initial Results of Net Error Empirical Research Using Logistic Regression
|
|
Author(s):
|
Richard Griffin*+ and Thomas Mule and Doug Olson
|
|
Companies:
|
U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau
|
|
Address:
|
104 Arrowhead Circle, Lansdale, PA, 19446,
|
|
Keywords:
|
dual system estimates ; logistic regression models ; cross-validation
|
|
Abstract:
|
For the dual-system estimates in past censuses, a post-stratification approach has been used. The post-stratification approach has significant limitations as it limits the number of factors that can be included due to sample size constraints. Statistical modeling techniques, such as logistic regression, offer more flexibility and possibilities for reducing sampling error and bias. The March 2001 post-stratification variables are used. The focus is on investigating the trade-off between bias and variance of estimates obtained by the elimination of higher-order interaction terms in the models and examining measures to evaluate and compare the fit of alternative models. Alternative estimates and their standard errors for race/origin/tenure groups are presented and contrasted. A logarithmic penalty function is suggested for model selection, and cross validation is used.
|
