A Monte Carlo Study Comparing CensusPES and CensusPlus when there is the Possibility of Undercounting

SYNOPTIC ABSTRACT

This paper presents the results of a Monte Carlo Study comparing a version of the capture-recapture estimation methodology called CensusPES with a ratio estimation methodology call CensusPlus. CensusPES and CensusPlus are similar methods for integrating sample evaluation results with mass enumeration results in an effort to provide one improved set of census numbers. The Monte Carlo Study makes use of 1990 official census block level counts by person type for the state of Alabama. When there is the possibility of missing persons in the mass enumeration as well as in the sample evaluation, it is demonstrated that statistical methods such as CensusPES or CensusPlus with appropriate data can successfully yield a high quality census count at all levels of geography.

Key Words and Phrases:

• Capture-recapture estimation
• CensusPES
• CensusPlus
• One number census
• Possibility of undercounting, and Ratio estimation