• Choose a category
• All
• Classic-Fiction

A second problem is that statistical methods are not well suited to testing causal mechanisms in the context of particular cases. These methods are optimized for assessing correlations across cases or among data points within a case, rather than for testing whether every aspect of a case is consistent with a hypothesized causal process. In contrast to statistical methods, if process-tracing shows that a single step in a hypothesized causal chain in a case study is not as the theory predicts, then the variable in question cannot explain that case without modification, even if it does explain most or even all other cases. If, for example, we find a case in which a democratic public clamored for going to war, the hypothesized propensity of democratic citizens to avoid voting upon themselves the cost of war cannot explain this case, even if it might explain other cases. Conversely, if a complex hypothesis involved one hundred steps and ninety-nine of these were as predicted in a case, a statistical test would confirm the hypothesized process at a high level of significance, but a case study analysis would continue to probe the missing step.

Third, the relative infrequency of both wars and contiguous democracies presents a sharp methodological limitation for statistical research. Given the small number of potential wars between democracies, the existence of even a few wars between democracies or the omission of a single relevant variable could call into question the statistical support for an interdemocratic peace.101 Because there are at least twenty hotly debated potential exceptions or near-exceptions to the assertion that democracies have never fought wars with one another, the results of statistical studies remain provisional despite the emerging consensus that an interdemocratic peace exists.102 For case study researchers, this is an opportunity rather than a problem: it is easily possible for the field as a whole to intensively study every one of the possible exceptions to the democratic peace and to also include a number of comparative cases of mixed dyads and nondemocratic dyads.

The Second Generation: Case Study Contributions

As one researcher argued in the 1990s, “generalizations about the democratic peace are fine—we have many of them—but now is the time to explore via comparative case studies the causal chains, if they exist.”103 The limitations of statistical methods as applied to the democratic peace were greatest precisely where case study methods had the most to contribute.104 Case studies on the democratic peace in the past decade illustrate the comparative advantages of qualitative methods and offer commendable examples of alternative research designs.

One of the main advantages of case studies is their ability to serve the heuristic purpose of inductively identifying additional variables and generating hypotheses.105 Statistical methods lack accepted procedures for inductively generating new hypotheses. Moreover, case studies can analyze qualitatively complex events and take into account numerous variables precisely because they do not require numerous cases or a restricted number of variables. Case study researchers are also not limited to variables that are readily quantified or those for which well-defined data sets already exist. Case studies on the democratic peace have thus identified or tested several new variables, including issue-specific state structures, specific norms on reciprocity and the use of deadly force, leaders’ perceptions of the democraticness of other states, transparency, and the distinction between status quo and challenger states.106

Second, process-tracing can test individual cases regarding the claims made about causal mechanisms that might account for a democratic peace. Miriam Elman, for example, asserts that The quantitative empirical analyses that find that democracy is associated with peace are correlational studies, and provide no evidence that leaders actually consider the opponent’s regime type in deciding between