Review of Certain Costs Associated with Pretests for the 1980 Census

Cost data of the 1980 census pretests were examined to determine problems experienced, their cause, and their potential impact on the cost estimates of the 1980 census.

The pretests cost significantly more than anticipated, but an accurate projection of 1980 census costs was not possible because: data were not accumulated in a manner to facilitate projection; procedures were experimental; geographic areas chosen were not representative; and better public cooperation may be obtained in the actual census. Major problem areas revealed by the pretests which could have an impact on the 1980 census include: poor mail response to questionnaires, lower productivity achieved by enumerators than for the 1970 census, and higher personnel turnover than in the 1970 census. Cost overruns of 28.6 percent for Travis County, Texas, and 73.4 percent for Camden, New Jersey, reflected the impact of these problems. The Bureau of the Census estimate for the cost of the 1980 census was revised from $565 million to $874 million in August 1977. This estimate may need to be revised upward because no provision was made for cost escalation.