|
Activity Number:
|
323
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305690 |
|
Title:
|
Estimation of Median Household and Family Incomes for Small Areas: A Bayesian Semiparametric Approach
|
|
Author(s):
|
Dhiman Bhadra*+ and Malay Ghosh
|
|
Companies:
|
University of Florida and University of Florida
|
|
Address:
|
102 Griffin Floyd Hall, Department of Statistics, Gainsville, FL, 32611,
|
|
Keywords:
|
Current Population Survey ; Hierarchical Bayes ; Penalized splines ; Semiparametric Regression ; Small area estimation
|
|
Abstract:
|
Median income estimates at state or county levels are important for the formulation of Government decisions and policies. For estimating these, the Bureau of Census uses the yearly estimates from Current Population Survey (CPS) along with covariates like Adjusted Census or the IRS median incomes for each state. These variables may have a non-linear relationship over time. Here we apply semiparametric regression tools to accurately estimate the yearly median household incomes for each state. We model the underlying true trajectory of the state-wise and overall median incomes using P-splines. Different basis function expansions are used for comparison. Univariate and bivariate Hierarchical Bayesian analysis is performed using families of different sizes. Model parameters are estimated using Gibbs sampling. We validate our model by comparing our estimates with the available census values.
|
