Correlates and consequences of American war casualties in World War I

• In Event: 2129 - Section on Comparative-Historical Sociology Refereed Roundtables (1 hour)
  In Refereed Roundtable 60min: 7. Violence, Non-violence, and Crime

Abstract

In this paper we address two questions. First, a demographic question: What were the risks of death in World War I for American men? With most records of United States soldiers from the era destroyed, there is limited microdata to estimate hazards of death, adjusting for individual characteristics. Thus, our assessment of social inequalities in casualties is ecological, examining whether casualty death rates were correlated with county-level measures of socioeconomic status. Second, we ask a sociological question about how casualty level affected different communities. The extent to which public opinion reacted to combat deaths sheds light on the strength of national ties at the local level. Were people willing to retrospectively approve of the local people lost for the pursuit of international political goals?

We find significant variation across states and counties in the level of casualties, with a distinctive geographic pattern. Casualties were higher in the Northern states, particularly in the Midwest and Mid-Atlantic. Initial analyses suggest, that similarly to recent conflicts, there was social inequality in military sacrifice. We examine inequality among counties by regressing county mortality ratios (for the total population) on a range of social indicators. In the North, counties that were less urban or had more farming households had higher casualties. We find a strong effect of German populations on mortality ratios. These results are significant as they may provide the basis for the strong reaction against the Democratic party in the 1920 election. Many states that had been competitive in the 1916 election suffered high casualties in the war, and swung strongly against the Democratic party. A one-standard deviation rise in the statewide casualty rate from WWI is associated with a one-third of a standard deviation decline in the Democratic Party vote share between the 1916 and 1920 election.

Authors

• Evan Roberts, University of Minnesota

• Alexandra Burda, University of Michigan