study overall changes in the quality of governance in a society rather than measuring changes in particular state institutions. Existing databases also do not provide consistent cross-national measures of the different effectiveness, accountability, and rule of law of the military and police. While existing measures might be useful as an indicator of the quality of central government, they are insufficiently sensitive to identify the changes that occur during a foreign reform effort. For example, the World Bank Governance Indicators for Iraq barely changed from 2003 to 2011, despite significant changes in the country’s institutions: on a −2.5 to 2.5 scale, government effectiveness varied from −1.70 to −1.57; voice and accountability from −1.5 to −1.07; and the rule of law from −1.64 to −1.45.52 One possible reason for the small variation in the World Bank statistics is that they are designed to be globally comparable. Making the index more sensitive to reforms in a particular country could make it harder to compare Iraq with other countries with more stable institutions, such as Saudi Arabia or Italy. Other databases that might speak to the quality of central government, such as Polity, do not include measurements of the relevant cases, because the international presence created an interruption in government.53 Analysts have also questioned the conceptual consistency of these datasets.54
Because of these challenges, I propose a new scoring system for the quality of a state institution. The score system evaluates each of the three desired dimensions listed above—effectiveness, accountability, and compliance with the law—from 0 to 3, with 0 corresponding to the absence or complete dysfunction with respect to that dimension; 1 meaning that the institution has some basic framework or structure but little else with respect to that dimension; 2 meaning that the institution has some desired characteristics but does not meet regional standards for that dimension; and 3 corresponding to an institution that operates at or above the level of comparable institutions in neighboring countries with respect to that dimension. The appendix provides specific questions that the case studies use to make these assessments. By adding the results of the scores for the three dimensions, the scoring system yields an overall score for the quality of an institution ranging from 0 to 9.
The case studies repeatedly apply this scoring system over the course of the reform effort to evaluate the quality of an institution over time and thereby judge the success of reform following a rubric shown in table 1.1. The case studies seek to be as transparent as possible in making this assessment, although the result is inevitably subjective and some improvement or decline in quality may not alter the score. This scoring system offers a useful, if not always precise, metric for evaluating competing theories and for comparing the relative improvement of state institutions across different countries and regions.
TABLE 1.1. Change in Score and Associated Level of Success
| Change in 0–9 score | Judgment of success | Description |
| 0 or negative | Counterproductive | Violence or major nonviolent protest occurred, or the state institution was worse off than when the stage of reform began. |
| 0 | None/Minimal | There may be identifiable improvements in rules, practices, or organization, but an outside observer would likely not notice the difference between the quality of the state institution before and after. |
| 0–1 | Limited | There were some changes in the rules or functioning of the state institution, which might have lasting effect, but the general quality of the state institution remained similar. |
| 2–3 | Moderate | There were clear improvements to the state institution. |
| 4–9 | High | The state institution experienced rapid and dramatic improvement. |
PLAN OF THE BOOK AND RESEARCH DESIGN
The book proceeds with six chapters and a methodological appendix. In chapter 2 I provide the background and hypotheses of my domestic opposition theory, the international resource theory, and the path dependence theory.
Seven case studies in chapters 3–6 evaluate these hypotheses. Chapters 3, 4, and 5 each considers two reform efforts from highly resourced post-conflict interventions, with chapter 3 examining reforms of central government institutions; chapter 4, reforms of defense institutions; and chapter 5, reforms of police institutions (see table 1.2). The reform efforts studied in these chapters occurred in Bosnia, Kosovo, East Timor, and Iraq, which, among interventions with a strong state-building mandate, had the highest level of international resources (see the appendix for details). Following the international resource theory, if reform success is likely to occur anywhere, it would occur in these societies.55 Within these countries, I chose reforms that cover the full range of variation in threat to domestic interests and international resources and that provide useful comparisons for causal inference. The appendix also shows there is substantial variation in the preintervention level of development of these countries, which the path dependence theory expects would lead to different outcomes. Chapter 6 explores the wider applicability of my domestic opposition theory by considering defense reform in Ukraine, which had a very different context from the reform efforts examined in chapters 3–5. Ukraine was not a post-conflict society, the local government was the driver of reform rather than the international community, and Western organizations had far fewer resources to encourage the changes that they desired. The fact that similar dynamics of domestic opposition occurred in Ukraine suggests that my theory applies to a wide range of US and international institution-building efforts, as I discuss in the conclusion.
The case studies use three logics of causal inference to test the hypotheses. A first type of causal inference, congruence testing, simply compares the observed outcome to the outcomes predicted by the various theories. The case studies also use causal process tracing, which means to scrutinize the events of the reform to identify information indicating whether the logic of the process hypotheses is operating as expected.56 Finally, I use the comparative method to isolate the impact of different variables and rule out competing explanations. The case studies compare reforms of the same institution in different countries, different state institutions within the same country, and variation over time within stages of the same reform.57 The appendix offers additional detail on these methods.
The case studies draw on a wide range of official documents, news reports, think tank analysis, and academic work. The case studies on reform efforts in Bosnia, Kosovo, and East Timor make use of more than 160 interviews from over ten months of fieldwork in these countries. I also conducted additional in-person and phone interviews in the United States. The interviews offer a firsthand perspective and fill gaps in the written record.58 Because nearly all interviews
