Skip to content

Response to Review no. 811

We appreciate the opportunity to respond to Brian H. Reid’s thoughtful review of Heroes and Cowards: The Social Face of War. Given book review publication lags, we have eagerly awaited the first set of reviews written by historians to appear. We have wondered how a book that is set in the Civil War but is not about the Civil War would strike historians.

Our book differs from the hundreds of Civil War books published every year in two ways. One is methodological. The second is that our book has implications beyond those four years of bloody combat and its immediate aftermath. We examine four main questions that are basic to social science research. First, when are men willing to sacrifice for the common good? We answer this by looking at why men deserted during the Civil War and find that men in more homogeneous companies, companies that were more similar in occupation, ethnicity, and age, were more loyal. Our second question is what are the benefits to men of friendship and of community? Loyalty to comrades raised men’s chances of dying during the Civil War. We find that the stronger the ties between men (and ties were stronger when men had more in common – whether because they were related or because they were of the same ethnicity), the more likely men were to survive the hell holes that were Civil War POW camps. Our third question is how do communities deal with betrayal? We found that during the Civil War men who had betrayed their communities by deserting moved away, driven out by shame and ostracism. We conclude that community codes of conduct are reinforced not just by loyalty but also by punishments. Our work shows that more diverse communities are less cohesive. Their members are less willing to sacrifice and derive fewer benefits from being part of the community. Our fourth and final question is whether there are any benefits to living and working in a diverse community. When we look at the lives of black soldiers after the Civil War, the tensions between the short-run costs of diversity and its long-run benefits become apparent. Men did not like to serve with those who were different from them, so much so that they were more likely to desert, but in the long run the ex-slaves who joined the Union Army learned the most from being in units with men who were different from themselves.

Our book is about the costs and benefits of diversity in any organization, not just in Civil War companies.  The same factors play out in terrorist organizations, in United States cities, on college campuses, and in towns in developing countries. Men may measure similarity in different ways across time and space. The distance between us may now be more likely to be education rather than ethnicity, but the basic lesson remains the same. A community of similar people is likely to be cohesive and its members are likely to sacrifice time, effort, and even their lives for each other. But in a diverse community members can learn from one another.

Much of history is written as if each case were unique. While each case does have unique features, we approach history as allowing us to uncover basic facts about human behaviour.

Our work is a quantitative history. Unlike many historians, we avoid using diaries and letters as our primary sources of evidence. Not only is ‘talk is cheap,’ but it is also the talk of the literate. Many black soldiers’ stories cannot be told if one does not ‘crunch the data’. Second, sophisticated men may ‘spin’ their diaries with an eye to history as they re-write events to play up bravery and downplay cowardice.  We therefore focus solely on outcomes we can observe in our data. Who deserts? Who survives? Who migrates where after the war? Who deserts? Who survives? Who migrates where after the war? These are well defined, costly choices that are easily amenable to statistical analysis.

We are not arguing that statistical analyses do not come without their drawbacks. (These are not, however, the drawbacks pointed out by Brian Reid. For example, when we say that ‘A good soldier was older’ we are not drawing any inferences beyond unit loyalty. The physical demands of the job are a completely different matter.) We acknowledge that one of the limitations of our study is that we have very little to say about the importance of leadership. We do find that having one of your own officers (and most of these officers were non-commissioned officers) with you in a POW camp, improved your survival probability. We can say that for black soldiers having an abolitionist officer decreased desertion rates but that the effect was relatively small. We agree that we could have done more with the non-commissioned officers for the white troops. The empirical challenge that a researcher faces in measuring the effects of ‘good leadership’ is to identify before combat begins who are the ‘good leaders’ and who are the ‘bad leaders’.

Finally, we want to emphasize that we do not just ‘exploit the 41,000 computerized soldiers’ records in the National Archives.’  Dora Costa has spent almost 20 years working on the digitalization and analysis of these records. The Nobel Laureate, Robert Fogel, has spent more years on this project. The final product is a unique data base that provides the life histories of men before the war, during the war, and after the war. A unique feature of our study is our ability to recreate each soldier’s peer network. We can do this because of the way the sample was drawn. The Union Army was built from companies of roughly 100 men. Our sample is based on randomly sampling companies and then conducting a 100 per cent census of all men in a randomly selected company. Without jargon, this means that for any soldier in our sample we also know information about all the other soldiers in his war company. These men lived together 24 hours a day and were forced to interact with each other continuously.

Our book does not reveal the full power of the Union Army data base. The data base has been used in studies of health and of retirement. Work is underway to continue to expand the data base to examine specific subpopulations. The data base is currently available at . We encourage other researchers to use this unique resource.