Summary: Pursuant to a congressional request, GAO reviewed Census Bureau procedures to estimate the accuracy of the census counts in the 1988 final precensus test, focusing on: (1) the post-enumeration survey (PES); and (2) major challenges confronting the 1990 PES.
GAO found that: (1) successfully matching PES and census records was extremely difficult; (2) even a small percentage of matching errors could significantly distort undercount estimates; (3) thorough assessments of the PES data quality were vital, due to difficulties in successfully completing matching and other operations; (4) PES or census data might be incomplete, inaccurate, or conflicting; (5) even when data were accurate, PES and census data might be improperly matched; (6) the Bureau planned to conduct 18 quality assessments of the 1990 PES to identify errors in the undercount estimate; (7) accurate matching would be an even greater challenge for the 1990 PES than in the dress rehearsal, since the Bureau must implement some new and untested procedures; (8) late additions to 1990 census counts complicated matching and follow-up efforts; (9) unlike the dress rehearsal, the 1990 PES would be nationwide and decentralized; and (10) the July 15, 1991 deadline forced the Bureau to compress its original PES schedule, putting PES quality at risk.
