Starting with the 1950 Census of Population and Housing, the Census Bureau has followed a policy of carrying out a careful evaluation of each census's accuracy in terms of coverage, content, and procedures, together with research and experimentation.

The 1980 census had a research, evaluation, and experimental (REX) program consisting of over 40 separate projects that could be categorized in terms of their objectives as follows:

• Coverage evaluation and coverage measurement procedures—By far the largest group of projects, this category included the major coverage measurement studies. Population coverage was estimated through the post-enumeration program (PEP), administrative-record matching, and demographic analysis. Housing coverage, including estimates of overenumeration, was the topic of  three studies, while the forward- and retrospective-trace studies and the IRS [Internal Revenue Service]/census match dealt with alternative population-coverage measurement procedures.
• Experimental program—The research efforts in this category included tests of alternative data-collection methods and ways to recruit, train, and maintain the temporary work force.
• Coverage-improvement evaluations—The studies in this category evaluated the cost, results, and field procedures for various components of the 1980 census coverage-improvement program.
• Processing and quality-control evaluations—A number of evaluations to study various aspects of the data capture and processing system were included in this category. The staff was particularly interested in the effectiveness of quality-control operations on the census program.
• Content evaluations—These investigated the validity of the responses to various census questions, such as utility costs and education.
• "Other" studies—Two studies in this category, one of the components of variance and the other of total error, were designed to estimate nonsampling errors in the census data. The other evaluations addressed the effect of the publicity program and respondents' behavior regarding the census questionnaire.