Summary: Pursuant to a congressional request, GAO reviewed the Census Bureau's 1990 Decennial Census planning efforts, specifically its decision to exclude optical mark recognition (OMR) technology as an option for the 1990 census.
GAO found that: (1) converting data obtained from questionnaires to a computer-readable format at an early stage improves data accuracy by allowing computerized editing and review; (2) substituting automation for a number of manual, labor-intensive processing efforts can control costs, but optimizing the advantages of automation requires advance detailed planning and early decisions; (3) although the Bureau made an early commitment to increased automation, it started detailed planning late and progressed at a slow pace; (4) late planning efforts, the reluctance to revise the questionnaire form, and a slow procurement process influenced the Bureau's decision to exclude a potentially useful automation option without fully exploring it; (5) the Bureau did not plan the necessary evaluations required to fully test OMR technology limitations or potential; and (6) although the Bureau decided to discontinue its efforts to test OMR as a primary data methodology for the 1990 census because of the costs to develop the proposed system and the short time frame remaining to accomplish the modifications, it should have had the same concerns about one or both of the other data capture methodologies it was considering for the 1990 census.
