Putting People in Places: A Semi-Automated Approach to Creating Record Linkages and High Resolution Geocoding across Historical Datasets.

Garand Spikberg, Michigan Technological University
Robert Pastel, Michigan Technological University
Don Lafreniere, Michigan Technological University

The Copper Country Historical Data Infrastructure (CCHSDI) contains a collection of many distinct historical datasets for the period 1880-1950, that have all been record linked and geocoded to digitized building footprints derived from Sanborn Fire Insurance Plans. This paper will outline the development and methodologies used to create a model of the social environment of the region through record linking Polk City Directories to the 100% count US census files from the IPUMS NHGIS project. We will outline our workflow from paper copies of city directories, to optical character recognition, text parsing with relational algebra, historical geocoding, and using probabilistic record linkage to the 100% count census files. The paper concludes with a discussion of the resulting representativeness of the population sample, and limitations of the process to studying the spatial dynamics of a population during the peak of US industrialization.

No extended abstract or paper available

 Presented in Session 31. Emerging Methods: Computation/Spatial Econometrics