Under-enumeration, inconsistency, and bias in the U.S. Manufacturing Census 1860--1880: Case studies from the American manufacturing belt

A new typology of enumeration and tabulation errors found in the manuscript schedules and the published reports of the U.S. Manufacturing Census 1860--1880 is proposed. This is based on a review of historical and contemporary assessments of census accuracy and additional new findings. Detailed case studies of different manufacturing sectors in the rapidly growing city of Cleveland, Ohio, including railroad shops, transportation-related manufacturing, and oil refining are undertaken using non-census sources, as worked examples of census reporting problems, and comparisons are provided with cities in surrounding states. Significant under-enumeration, bias toward small companies, and inconsistent reporting (both within and between census years) emerge as the norm rather than the worst case scenario. The manufacturing census is found to be unsuitable for use in quantitative studies without extensive corroborating evidence of accuracy.