Administrative Records Experiment in 2000: Outcomes Evaluation

This paper presents selected results from the Administrative Records Experiment in 2000 (AREX 2000). AREX 2000 used administrative records to enumerate the population in two test sites and compared the results to Census 2000. The test sites included two Maryland and three Colorado counties that offered distinct challenges to the enumeration process. The Outcomes Evaluation assessed two enumeration methods, compared county and sub-county population counts to Census results, and examined the impact of race imputation and other processing issues. The results confirm that administrative records provide good estimates of Census household population counts at larger geographies with greater accuracy using the Bottom-Up enumeration method. Both Top-Down and Bottom-Up methods undercounted Census in most counties, with Bottom-Up population counts ranging from 97-102% of Census results. Age and sex differences indicated problems with source administrative files. Most of the AREX race distributions did not accurately replicate Census results, which was attributed to weaknesses in the race imputation methodology. Imputation rates and type of imputation, housing unit characteristics, and presence of non-relative household members were all associated with AREX-Census differences.