Skip to Highlights

To aid Congress and the Census Bureau in planning for the next census, GAO reviewed portions of the 1980 census program concerning mailing list development, follow-up on nonrespondents, and activities to reduce the number of persons missed.

Skip to Recommendations


Matter for Congressional Consideration

Matter Status Comments
Congress should enact legislation, if the Secretary of Commerce decides to purchase address information for the 1990 census from USPS, that: (1) specifically authorizes USPS to provide the Census Bureau address information; and (2) protects the confidentiality of address information provided to the Census Bureau by USPS.
Closed - Not Implemented
Legislation was enacted, P.L. 98-166, November 28, 1983, to authorize USPS to provide the Census Bureau with address information to test the usefulness of the data for the 1990 census. Test results were inconclusive as to the advantage of USPS-provided lists. Therefore, the lists will not be obtained from USPS and legislation will not be requested.

Recommendations for Executive Action

Agency Affected Recommendation Status
Department of Commerce 1. The Secretary of Commerce should, in cooperation with the Postmaster General, comprehensively evaluate alternatives for developing address data for the 1990 census.
Closed - Implemented
Alternative methodologies were explored by the Census Bureau, in conjunction with the Postmaster General, during the 1990 Decennial Census Testing Program. The GAO analysis of these tests showed that there could be cost advantages for USPS lists. However, the tests were limited because of cost limitations.
Department of Commerce 2. The Secretary of Commerce should test the feasibility of using mail reminder cards and follow-up mailings. If one or both of the techniques prove to be adequate to meet Commerce's needs, they should be used as alternatives to reduce the need for personal visit interviews for the 1990 census.
Closed - Implemented
The Census Bureau tested the use of reminder cards in its 1985 and 1986 pretests. In September 1986, the Bureau decided to use reminder cards, but because of the processing methods, the cards were sent to everyone not targeted to nonrespondents. Mailing a second questionnaire is not being considered.
Department of Commerce 3. The Secretary of Commerce should extend the time between census day and the start of follow-up operations to allow field staffs enough time to sort out duplicate and inappropriately mailed questionnaires and to allow them time to check in late mail returns.
Closed - Implemented
In its 1986 pretest, the Census Bureau experimented with extending the time between census day and the start of follow-up operations. The Bureau has decided to delay the follow-up activities.
Department of Commerce 4. The Secretary of Commerce should evaluate the feasibility of increased use of imputation, where legally permissible, as a method for developing census information on difficult-to-enumerate households.
Closed - Implemented
An important case in Indiana, in which the Census Bureau's 1980 statistical methods were challenged, was decided in favor of the Bureau. However, the Bureau decided against increased use of imputation because it is not consistent with the goal to conduct a complete count. This decision was reached despite the cost savings possible by imputing.
Department of Commerce 5. The Secretary of Commerce should evaluate the cost and effectiveness of 1980 census coverage improvement programs to determine if they should be used in the 1990 census. When practical, the evaluation should: (1) identify the cost and result of each 1980 coverage improvement program for various geographical areas and target groups; (2) test the sensitivity of program costs and results to changes in the assumptions upon which the programs are based, such as increasing and decreasing the levels of program activity on target groups and in geographic areas; and (3) express 1990 estimates of cost and results for coverage improvement programs in ranges of values by target groups and geographic areas rather than just a single national value.
Closed - Implemented
The Bureau completed an evaluation of the 1980 coverage improvement program. It issued a report on the subject in January 1987.
Department of Commerce 6. The Secretary of Commerce should evaluate coverage improvement programs used in future censuses by compiling aggregate cost and results data on the operations. The data to be gathered should track the results of coverage improvement programs at the state and sub-state levels, also by target groups.
Closed - Implemented
Action cannot be initiated until after the 1990 census. A joint Census Bureau, Office of Inspector General, and Department of Commerce cost benefit analysis was prepared on improvement procedures. However, because of accounting procedures and technical problems, the analysis could not be completed on about 25 percent of the improvements.
Department of Commerce 7. The Secretary of Commerce should, by 1984: (1) have the Census Bureau conduct pilot tests to develop better cost and effectiveness information on updating the 1980 mailing lists and for purchasing lists from USPS to ascertain the quality and cost of mailing lists produced by these alternatives; and (2) compare the results of the pilot tests with comparable information compiled on the 1980 census and any other alternative the Census Bureau may identify and, after considering the quality and cost of the mailing lists produced, select the best method.
Closed - Implemented
In September 1986, the Bureau made its decision on address compilation and list updating. Although GAO had some reservations about the methodology, the Bureau did conduct pilot studies. GAO did not have any question as to the Bureau's decision.

Full Report