This chapter turns to the statistical analysis of communication. It examines the signaling hypotheses and a range of fundamental questions about international affairs. In the last two chapters, we saw that private encounters influence the calculations of state actors in some cases, but are these the exception or the rule? Do public statements regularly convince in a way that private encounters do not? How does the context of alliance relations and material power influence the credibility of demands? Do the judgments formed from diplomatic signals or material factors merely confirm inferences drawn from the other domain? In other words, is one set of factors a leading indicator of state intentions? These and other questions are addressed using first-of-its-kind data on the public and private statements of diplomats and state leaders. The analyses of this data bear out the predictions of the theoretical models, in some instances in provisional fashion, where the data are limited, and in other instances, where the data are plentiful, with substantial confidence.

DEMANDS, OFFERS AND ASSURANCES DATA

To examine the signaling hypotheses statistically, data were collected on all demands, offers and assurances of which the British were aware, made by European great powers to other European great powers between 1900 and 1914. Like the inference dataset described previously, these data were drawn from the Confidential Print of the British Empire. The data comprise 955 unique statements, 83 percent of which were made by diplomats and leaders away from view of their publics. Slightly more than half of these statements were offers or assurances and the rest were threats or demands. If the same demand or assurance was made by the same state and to the same state on multiple occasions, this was coded as a single observation. More detailed information on data collection procedures and coding rules for all of the variables discussed in this chapter can be found in Appendices B and C.

In combination with the inference data, this data enable a more precise test of hypotheses than has been possible in previous studies. Not only has most scholarship focused on public signaling, it has also nearly exclusively examined the connection between statements and whether an adversary makes a concession. A finding that certain factors are associated with concessions is taken as evidence that those factors are associated with the credibility of threats.