Chapter 4 presented the patterns of compliance graphically and in simple tables. This companion chapter provides the details of the statistical analysis for readers interested in seeing “what is going on underneath the hood” of these results. I discuss issues of research design and the coding rules for the data, and present the tables of the estimated coefficients and additional tests of the results reported in the prior companion chapter. I understand that some readers will not be interested in the material presented here and so may wish to skip ahead to Chapter 5 on the detailed analysis of prisoners of war.

I have broken this chapter into sections to allow the reader to move directly to particular topics of interest. The order of the section generally matches the order of Chapter 4, although this chapter also includes analyses, such as that of declarations of war, that were not discussed in the companion chapter. The presentation here follows the standards of statistically oriented political science, as it is written for that audience.

This chapter presents the central statistical results from the dataset, those underlying the graphics of the previous companion chapter and demonstrating that they are robust. In doing that work, I explored many other possible specifications of the statistical models. Space, and the interest of most readers, precludes me from presenting all these combinations in these pages. Dina Zinnes once said that she figured that for every regression published in an article, ninety-nine others went into the trash can. I do not think my ratio of specifications tried and discarded to those reported is quite that high, but my statistical trash can is hardly empty. In the spirit of allowing the interested reader to rummage through my statistical trash can, I have created a Web site (http://sitemaker. umich.edu/jdmorrow/lawsofwar) that contains a wide range of supplementary analyses to those reported here. I refer to some of these estimations in the text and footnotes.