Abstract:
|
This AREX 2000 evaluation compares county, tract, and block tallies between administrative records and Census 2000 results. The AREX sample consists of all persons in two Maryland and three Colorado counties and was collected in two modes: those with city-style addresses (top-down) and those with city-style addresses matching Census master address files (bottom-up). The analyses compare top-down and bottom-up differences and examine how well AREX approximates Census by key demographic characteristics. AREX-Census differences are presumed to differ by race, Hispanicity, age, sex, and the structure and composition of neighborhoods. Thematic maps elaborate AREX-Census differences.
The second part of this evaluation emphasizes regression models to determine the key characteristics of block-level AREX-Census differences. Residuals are mapped to determine the pattern of unobserved/unmeasured heterogeneity in the AREX sites. The results confirm that administrative records provide good estimates of Census counts at larger geographies, with greater accuracy from BARCUF counts. AREX processing deficiencies are investigated and confirm known problems in identifying selected demographic groups.
|