ABSTRACT
Census takers started to use printed questionnaires in the 18th century. The first breakthrough was Hollerith’s electromechanical tabulator for the US 1890 census. Other rationalizations involve self-enumeration, individual forms and calculation gadgets. In 1951, Univac installed a mainframe computer in the Census Bureau, using FOSDIC microfilm to enter the data. Statistics Norway in 1964 built the Central Population Register, spelling the end of the questionnaire-based census. Eighteen European countries used register data to take the 2011 census; difficult to monitor are international students, homeless people and migrants.
Traditionally, statistical bureaus print their aggregates. Computers brought search engines to trace individuals and scanned or transcribed aggregates. Census bureaus, genealogists and population researchers bring widespread access to the census manuscripts as historic or anonymized contemporary individual-level microdata. Infrastructure undertakings encode the transcribed versions, construct variables for studying family structure and harmonize codes and record layouts in order to facilitate comparative research. The Minnesota Population Center (MPC) carries this work furthest with over a billion individual records. The IPUMS provides modern census samples of individuals in households from countries on all continents, and the North Atlantic Population Project distributes full-count censuses from North America, the Nordic countries and the UK from 1703 to 1910. Even small population groups can be analyzed online, in data enclaves or after downloading selections. In Norway, the censuses are transcribed for record linkage when building the historical population register.
Censuses based on central population registers rather than questionnaires entail transfer of control over census categories and classification to the state. Researchers’ limited access to recent microdata (even if anonymized) is an extension of this conflict between the central and local levels.
