While the Comparative Constitutions Project will likely have general applicability to a broad range of research questions, the investigators are motivated by a very specific set of empirical and theoretical questions that we can classify roughly as the (1) origins, (2) implementation, (3) duration, and (4) conceptualization and measurement of constitutions. We expect that the craft of constitutional design will improve both as an indirect result of improved scholarly analysis, and as a direct result through analysis of the data by constitutional framers themselves.

Origins: Diffusion and the Process of Constitution-Making

What factors lead to the adoption of one constitutional component over another? Surprisingly, we know little about the process of constitutional design. Indeed, Elster (1995: 364) is unable to identify even “a single book or article that considers the process of constitution-making in its full generality.” The investigators are concerned, in particular, with one aspect of the process: how, and to what extent, constitutional designers incorporate foreign constitutions as models. That constitutional actors engage with and utilize material from foreign sources at some level is a fundamental assumption of theories of diffusion, an enduring theoretical concern for the investigators.

We conceive of diffusion as a general set of processes by which one government's adoption of a constitutional component affects the probability that another actor adopt the same component (Elkins 2003; Elkins and Simmons 2005). Our basic expectation, stated quite generally, is that the constitutional provisions of peer countries have a decisive impact on the decisions of constitutional engineers regardless of a state's particular institutional needs and constraints. Context and fashion, in this line of reasoning, matter as much as a government's particular propensity for a given institution.

Exactly when and how context matters, of course, is the critical theoretical question of interest. The investigators – working independently – have developed their theoretical expectations based on in-depth case studies of constitutional design in Latin America (Elkins 2003) and judicial review in Asia (Ginsburg 2003). Through this work, the investigators have identified various plausible diffusion processes, each of which implies a set of hypotheses testable with the dataset the investigators propose to construct. For the most part, these are learning theories according to which the cognitive and research capabilities of constitutional designers are constrained in predictable patterns. One pattern, for example, is that designers pay disproportionate attention to the constitutions of states that are highly “available,” in the sense of Kahneman and Tversky (1982). So, we should expect that the constitutions of countries with high levels of interaction and communication (measured by levels of trade, linguistic compatibility, telephone flows, or mail flows) will be more accessible and, so, more similar. On the other hand, it might be the case that constitutional learning follows a pattern that might be describe as analogous modeling. Under that framework, constitutional framers model their institutions after countries that are similar to theirs in important ways (such as size, ethno-linguistic fragmentation, or security environment), reasoning that what is good for an analogous state will be good for them. Finally, it might simply be that a state's success – measured by political stability or wealth – inspires imitators.

We suggest these three hypotheses as examples of the sort of empirical implications that derive from the models of learning that seem most plausible to us. Certainly, a rich literature on policy diffusion more generally (for a review, see Strang and Soule 1998), and constitutional “borrowing” and “migration” more specifically (e.g. Epstein 2003; Chaudhry, forthcoming; Boli et al. 1987) suggests other mechanisms of interdependence. However, an adequate test of any of these hypotheses is not possible without the data we describe here. A central component of our analytical design is the development of measures of similarity among constitutions with respect to their provisions, structure, and style. Using spatial and network methods, we can compare these similarities with measures of social, political, and economic distance among countries to understand the migratory patterns of constitutional ideas, and make inferences about the diffusion mechanisms at work (Simmons and Elkins 2004).

The investigators' diffusion framework, although complementary to traditional approaches to constitutional design, marks a distinct departure. Traditional explanations of constitutional design focus on interests, passions, and the public good (Elster 1996). Interest-based accounts (e.g. Mueller 2000; Cooter 2003) focus on rational choice of self-interested designers. Other scholars, however, emphasize emotions and constraints on rational choice (e.g. Elster et al. 1997), or historical processes that can lead to reactive constitutional drafting (Scheppele 2003). A third school of thought, found in traditional legal scholarship, describes constitutional drafters as somewhat disinterested engineers motivated by higher purposes such as efficiency and the interests of the public. Our diffusion perspective does not contradict these causal stories, some of which we have ourselves documented and developed (Ginsburg 2003). However, it does introduce an altogether different set of factors, which appear to play an important explanatory role.

Implementation: De Jure versus De Facto Constitutional Law

A second research objective concerns the implementation of constitutional provisions. There is a certain cynicism regarding the value of constitutions. The cynicism stems, in one part, from the suspicion that constitutions - with their lofty ideals and sometimes inscrutable language - are often at variance with actual political practice. The poster child of this problem (at least in comparative constitutional texts) is the former Soviet Union, whose constitution has been lauded as the most democratic ever written (Finer et al. 1995). Enough authoritarian governments have written “democratic” constitutions, however, that the complaint is by no means limited to the Soviet Union. The Democratic People's Republic of North Korea, for example, has a constitution that guarantees an extensive set of human rights. In addition, constitutions will sometimes call for democratic institutions that are never instantiated. For example, the Cambodian Constitution of 1993 calls for a constitutional court that was not set up for many years.

Sorting out the difference between the theory and practice of government is important for understanding the health of democratic governance. Well-functioning democracies generally demonstrate a tight fit between the two. When they do not - that is, when higher law is violated or inconsequential - citizens lack any understanding of when they can and should enforce their democratic (or otherwise constitutional) rights (Weingast 1997).

In order to investigate the correspondence between de jure and de facto law, one needs an accounting of both. Our interest, we remind, is in measuring the former. Conveniently, a fair number of existing datasets already capture the latter, whether it be in the area of rights (e.g. Freedom House; Amnesty International), constraints on the executive and participatory institutions (Marshall and Jaggers 2000), the role of the military and religious organizations in politics, or the transparency of government affairs (International Country Risk Guide). With these two sets of measures, the general analytic strategy is to ascertain how well the provisions of the constitution in a particular area predict actual practice.

Constitutional Duration

How long do constitutions last and what factors explain their duration? Our normative background assumption is that durable constitutions are important for the maintenance of democracy. This assumption rests on the premise that constitutions serve the critical function of tying the hands of government, a function that presupposes the document's endurance. If constitutions are fleeting, leaders feel no obligation to abide by their constraints and citizens, having minimal attachment to a fleeting document, will not punish those who transgress its boundaries (see Weingast 1997). Of course, this preference for endurance should not be construed as trumping otherwise legitimate reasons for change. Certainly, some constitutions should be rewritten. Nonetheless, notwithstanding an appropriate degree of tinkering, stability in the rules of the game is an important component to democratic rule, and underlies normative theories of constitutionalism. Not only does stability promote limits on government that become self-enforcing once they are consolidated, but a stable set of rules allows institutions (most importantly, parties) to develop. With these normative preferences in mind, we aim to investigate the empirical patterns underlying constitutional duration as well as the conditions under which constitutions come and go.

The dataset that we propose is a critical ingredient in answering these questions. In order to understand constitutional lifespans, one needs a comprehensive chronology of constitutional changes in each country in the sample. Such a chronology should identify the year of new constitutions, as well as amendments to, suspensions, and reinstatements of constitutions. Of course, this sort of chronology also defines the universe of constitutions, a critical first step in the creation of our dataset (see sampling plan below). As such, we have already completed the chronology, a fairly laborious task of historicism in which we referred to over 100 cross-national and country-level sources.

We are thus in a position to describe some of the basic patterns in the birth, death, and lifespan of constitutions. According to our data, for example, their median lifespan is about 17 years across the world since 1789. Of course this varies across generations and across regions (for trends, see Elkins et al. 2006).

We intend to investigate the sources of constitutional durability by combining the lifespan data with risk factors associated with the political and economic structure, international and domestic crises, constitution-making in neighboring states, and crucially, internal characteristics of constitutions themselves. For example, we can test hypotheses regarding the relationships between constitutional duration and a state's democratic tradition, its age, and its level of economic development. We can also, controlling for these external factors, investigate whether internal features of constitutions such as regime type and the flexibility of amendment procedures tend to facilitate constitutional survival (Lutz 1994). As such, the dataset makes for a useful resource to study institutional evolution and adaptation within states across time, and assess what constitutional configurations lead to documents that stand the test of time.

Towards Better Description: Conceptual Refinement of Constitutional Configurations

Quite apart from its explanatory value, the dataset we propose promises to pay significant dividends with respect to the conceptualization and description of constitutional structures. We submit that such a contribution will be an invaluable resource for both research and teaching in two important ways. First, the constitutional characteristics that we will record will allow for the refinement and elaboration of many of central features of governing institutions. Such refinement will allow for the more reliable test of important institutional theories that invoke constructs such as executive-legislative relations, federalism, and judicial review. Second, the dataset will dramatically enhance our understanding of how constitutional texts vary over time (Arjomand 2003) and across regions, cultures, and political systems.

Improving the Practice of Constitutional Design

As we assert at the outset, the science of constitutional design would benefit from the resource that we propose to develop. Our assertion stems from a study of cases of constitution making, discussion with academic advisors to constitutional assemblies, as well as our own modest experience in constitution-making. Ginsburg has had involvement in constitutional reform exercises in Mongolia and Afghanistan, and has closely followed drafting developments in several other Asian countries. In general, constitutional reform tends to proceed with the help of carefully selected experts, who in the best of circumstances have broad comparative knowledge and experience. But without a central resource documenting the various options taken by other countries, design processes must continuously reinvent the wheel in terms of informing the local decision-makers about the options available to them. Local lawyers who may be staffing and advising constitutional drafting commissions are at a relative disadvantage in gathering information on how other countries' constitutions look. In addition to the micro-data that we will make available to researchers, we will also provide a user-friendly diagnostic web tool for non-technical users to access the data.