An Empirical Study on Using ACS Supplementary Survey Data in SAIPE State Poverty Models

• Activity Number: 369
• Type: Topic Contributed
• Date/Time: Wednesday, August 11, 2004 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Elizabeth T. Huang*+
• Companies: U.S. Census Bureau
• Keywords: small-area estimation ; Fay-Herriott model ; American Community Survey

Abstract

The Census Bureau's Small Area Income and Poverty Estimates (SAIPE) program produces state poverty estimates from a Fay-Herriot model applied to direct state estimates from the Current Population Survey (CPS). In recent years supplementary surveys for the American Community Survey (ACS) have also produced state poverty estimates. While their different data collection procedures mean that these estimates should have different nonsampling errors than the CPS, they have the advantage of being based on much larger samples. We take data from the 2001 supplementary survey of ACS (SS01) and examine alternative models that use the SS01 data in different ways reflecting different assumptions about the relative biases between the SS01 and CPS estimates. We then examine the prediction error variances from the various models including the SAIPE production models without the SS01 data. The overall conclusions are that there is significant potential for improving the state poverty estimates by using the supplementary survey data in the models, and that the amount of improvement depends on what one assumes about relative biases in the estimates.