Josiah Grover, University of Minnesota
José Pacas, University of Minnesota
Steven Ruggles, University of Minnesota
The U.S. 1850, 1860, and 1870 Decennial Censuses did not include a question about relationship to head of household which is a major limitation for the study of historical family structures. The IPUMS project overcomes this lack of data by exploiting the 1880 census microdata as a donor pool for relationship status in these earlier years, making IPUMS USA the only source of family interrelationship data for the time period. Although the imputed data have been available since 1995, there has been little documentation on the procedure. In this paper we have three main goals. First, we outline the methods used to impute relationship status for the microdata samples from 1850 through 1870. We then highlight specific complications of working with these early datasets. Lastly, we document the reliability of these imputed relationships and introduce new ways of testing their reliability.
No extended abstract or paper available
Presented in Session 78. Big Data in Historical Research