Full Citation

Title: Representativeness and False Links in the 1850-1930 IPUMS Linked Representative Historical Samples

Citation Type: Miscellaneous

Publication Year: 2017

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: The Integrated Public Use Microdata Series Linked Representative Samples (IPUMS-LRS) have transformed historical social, demographic, economic, and health research. These public-use samples linking the 1850-1870 and 1900-1930 Censuses to the full-count 1880 Census contain nearly 500,000 individuals observed at multiple points in time and cover many races and subpopulations, including migrants and women. This paper describes the representativeness of the IPUMS-LRS and uses an independent metric to quantify the incidence of incorrect matches. We find weighted IPUMS-LRS data fail to produce representative samples with respect to some variables characteristics, and we outline a simple procedure that—under certain assumptions— allows researchers to create weights customized to specific samples and research questions. We also find that suggestive evidence that errors in linking are very low in the 1880-1900, 1880-1910, 1880-1920, and 1880-1930 IPUMS-LRS, hovering around 1 percent. Although lower than other automated methods, the rate of false links in the pre-1880 IPUMS-LRS may range from 5 to 10 percent. We conclude with simple recommendations of ways to improve inference with these data.

Url: http://www-personal.umich.edu/~baileymj/Bailey_Cole_Massey.pdf

User Submitted?: No

Authors: Bailey, Martha; Cole, Connor; Massey, Catherine

Publisher: University of Michigan

Data Collections: IPUMS USA

Topics: Methodology and Data Collection

Countries: