The U. S. Census Bureau has provided public use microdata as a component of its decennial census data products since 1963 when we released a one-in-one-thousand sample file for the 1960 Decennial Census. Since then, microdata files have become an integral part of our decennial census and demographic surveys programs. As a result, researchers in other Government agencies and research institutes have been able to conduct important policy and planning studies that could not be answered through the use of published tabulations, Were it not for public use microdata files, the only way these studies could be done, if at all, would be by contracting with the Census Bureau for special tabulations. This is not the preferred solution for several reasons. First, these special requests are totally dependent on programming and computer support that is committed to routine Census work. Therefore, the time required to complete the work does not always satisfy user needs, Second, statistical analyses do not always turn out the way researchers intended. They may want to change the variables or the analytical methods after they see the initial results. Finally, in contrast to the costs of using available staff and micro-computers, the costs of using Census Bureau main-frame computers and programmers may exceed the available resources for the project.

The advent of public use files has eliminated many of these problems but has introduced some new ones both for the researcher and for the Census Bureau. Because of the flexibility available when using microdata files, the broad access to high speed computers, and the increased sophistication of data users, there has been an increased use of this medium in the 1970’s and, particularly in the 1980’s. With this increased use has come increased demand for detailed information that was excluded from or curtailed on public use files to protect the identity of survey and census respondents. The statute (Title 13) under which the Census Bureau operates requires that when we collect and publish data under this authority we not publish results that can be used to identify a particular respondent. Realizing that it may not be possible to release data from which it is absolutely impossible to identify an individual, we strive to ensure that the risk that the data will be used to identify someone on the file is extremely small. For example, current microdata disclosure protection criteria prevent the release of geographic identifiers for areas with small populations, extreme values for continuous variables, and information that is obtained from or matchable to administrative records systems. These restrictions prevent survey sponsors from conducting some analyses, such as certain microsimulations; reprocessing the individual responses; or having their own administrative data appended to the survey results. They also inhibit the potential of surveys we sponsor (for example, the Survey of Income and Program Participation (SIPP)) for program policy research by other Government agencies.