From Millions to Meaning: Principles and Best Practices for Research with IPUMS Full Count Census Data

Matt A. Nelson, University of Minnesota

The IPUMS Full Count Census Data and Multigenerational Longitudinal Panel has revolutionized historical demographic, public health, and economic research in the United States. Containing 816 million unique person records between 1850 and 1950 and linking over 175 million persons between censuses, these data have encouraged both large-scale national studies and detailed local investigations. Because of the outward “cleanliness” of the data and likely a lack of knowledge on the data production, many social scientists do not consider and critique the data source itself. This leads to analyses where scholars apply the data inappropriately without asking whether the data is sufficiently capable of answering their research question. Scholars need to consider how far removed we are from the original source of information, and critically assess results not in line with previous facts and theory. This paper summarizes the journey from the original respondent to modern day analyses, pitfalls to consider and avoid, and finally recommendations on principles and best practices for working with these historical data. While this paper focuses explicitly on the IPUMS Full Count Data, the reflections and principles designated here could easily be applied to any other historical dataset.

No extended abstract or paper available

 Presented in Session 159. Building and Interpreting Censuses I