Historical Data Recovery at the US Census Bureau: 1980 Census Address Data

John Sullivan, Census Bureau
Katie Genadek, University of Colorado
Randy Becker, U.S. Census Bureau

New data about the past must often be recovered for use in modern research, yet differences between the record keeping technologies prevailing during the period of study and those of today present challenges for the recovery of some historical data. The US Census Bureau maintains a large and nearly untouched collection of historical data. The data were created by the Census Bureau from the 1950s to the 1990s and contain survey and administrative data about the US population and economy, and much of this data was saved from scheduled destruction in a preservation effort and does not exist elsewhere. Files from tens of thousands of magnetic data tapes, designed to be read by mainframe computers, were transferred to a modern storage format. Though saved from loss, the data remain in their native format, creating challenges for use on modern computer systems. Furthermore, the metadata associated with each converted data tape existed only as printed and hand-written documents. This paper describes the conversion of historical data from magnetic tape to digital files, work conducted at the National Processing Center (NPC) to scan and key documents containing metadata, and the recovery and translation of geographic data used in the 1980 Decennial Census for use in the modern Decennial Census and Digitization (DCDL) project.

See paper

 Presented in Session 30. History of Data and Statistics