Peer Effects in the American Civil War

Abstract

Click to increase image sizeClick to decrease image size Acknowledgments Dora L. Costa and Matthew E. Kahn. Heroes and Cowards: The Social Face of War. Princeton, NJ: Princeton University Press, 2008. Notes 1. Peter Viechnicki (2003, 326) reported that only 41 percent of eligible soldiers—those believed to be living in the United States at the time of the 1860 census enumeration—were successfully linked to the census. New electronic indexes for the entire population by Web sites such as ancestry.com no doubt helped the research team to increase the number of successful links and likely explain the discrepancies. 2. The data set, which is funded by the National Institutes of Health and the National Institute on Aging (P01 AG10120), has been the basis for dozens of articles and several books on nineteenth-century economic, demographic, and medical history and is available for use by qualified researchers. A full bibliography and access information can be found at the Center for Population Economics. A few other small problems with the current version of the database are worth mentioning. Although the database is well-documented and can be accessed with a useful extract system, potential users will discover a lack of some user-friendly features common to other public-use datasets. Dates are given as ASCII character strings, with no inference of missing data. A recruit reported to have died on an unknown day in April 1864, for example, is given a death date of “186504–”. Many variables are not coded, some are coded with abbreviations, and a few show signs of poor data cleaning and coding (e.g., about 10 percent of the recruits in the 1860 census data appear to be included twice in their households, and 84 recruits were reported to have been born in “BA,” the abbreviation, according to the documentation, for Bangladesh). The data set is still under construction, however, and enhancements—such as a new variable coding, such as cause of death and contributing cause of death, using new international standards—are regularly added. 3. Costa and Kahn infer marital status from the 1860 census. The variable is necessarily treated as time invariant. Given that peak war participation ages overlapped with peak marriage ages, however, it is likely that many men whom the authors inferred as being single in 1860 married before or during the war.