Evaluating the Consistency of Responses to the Hispanic Origin and Race Questions on Census 2000, Using the 2000 Content Reinterview Survey

• Activity Number: 408
• Type: Topic Contributed
• Date/Time: Wednesday, August 6, 2003 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Phyllis E. Singer*+
• Companies: U.S. Census Bureau
• Keywords: Hispanic origin ; data quality ; content reinteview ; response variance ; race

Abstract

The purpose of the Content Reinterview Survey was to evaluate the consistency of responses to Census 2000. Knowledge of the accuracy and reliability of census data aids both data users and census planners. Data users use such information to determine how errors in the data might affect the conclusions they draw from analyzing census data. Census planners use such information to develop and test methods to improve the quality of future censuses. Prior to Census 2000, we randomly selected 30,000 households that were initially selected to receive the census long form. Experienced census field representatives telephoned the selected households to re-ask most of the census long-form questions. Personal visit interviews were allowed if the households couldn't be reached by phone. We were able to analyze data from about 20,000 of the preselected households. Analysts computed the index of inconsistency--a measure of relative response variance--and evaluated the consistency of each item at the national level. This paper documents the study, our key findings, and limitations. It will focus on the consistency of the Hispanic origin and race questions.