A Comparison Study of ACS If-Then-Else, NIM, and DISCRETE Edit and Imputation Systems Using ACS Data

• Activity Number: 305
• Type: Topic Contributed
• Date/Time: Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods*
• Author(s): Bor-Chung Chen*+
• Affiliation(s): U.S. Census Bureau
• Keywords: integer programming ; set covering algorithm ; hot deck ; decision logic table

Abstract

In this paper, we compare the ACS If-Then-Else rules with alternative approaches that have potential to improve the data quality of survey data. The alternative approaches are the Fellegi-Holt model-based DISCRETE edit and model-based imputation system of the U.S. Bureau of the Census and the Nearest-Neighbor Imputation Method (NIM) of Statistics Canada. The Fellegi-Holt method has the virtues that the logical consistency of the entire set of edit rules can be checked and that, in one pass through the data, an edit-failed and imputed record can be assured to satisfy all edits. The NIM is a successful editing and donor imputation method that has been used for the 1996 and 2001 Canadian Censuses and will be used for the 2006 Canadian Census. We use the 1999 American Community Aurvey (ACS) data of 26 states for the comparisons. The If-Then-Else rules used are described in the 1999 ACS Edit and Allocation Specifications for Basic Population Variables, which include sex, age, householder relationship, marital status, race, and Hispanic origin.