In developing our sample, we have established sampling guidelines with respect to three dimensions: time, countries, and constitutional events. In general, the existence of historical constitutional texts throughout the world encourages us to be expansive. At the same time, we are cognizant of practical limits of time and resources. Nonetheless, our pilot study suggests that it is reasonable to err on the side of inclusiveness at this point. The current sample includes more than 700 new and interim constitutions as well as more than 1500 constitutional amendments.

Time - We include constitutional events from 1789 to 2005. 1789, of course, marks the effective year of the United States constitution, the widely reputed "first" document of its kind. Certainly, other documents of quasi-constitutional status predate the US document (e.g., the Magna Carta). Nevertheless, such documents are sufficiently rare and ambiguously "constitutional" that 1789 makes for an appropriate and practical beginning for our study.

Countries - Our intention is to include all independent states, of sufficient size, that have existed for at least five years during the sampled timeframe. Having undertaken cross-national research projects, we recognize that defining states and their existence across time is not a simple matter. In our estimation, Ward and Gleditsch's (2006) list is the most authoritative and careful accounting of state births and deaths. As such, we base our sample of states on Ward and Gleditsch (2006) list of states (including microstates), except where otherwise noted. We do not, however, include constitutions of subnational units in federal systems.

A special problem concerns the small number of countries that either lack written constitutions (e.g., the United Kingdom) or have written quasi-constitutions composed of a small number of politically entrenched laws (New Zealand, Israel). Our focus on written constitutions forces us to exclude countries without any written constitution whatsoever (the United Kingdom, Saudi Arabia until 1993). For those countries that have only quasi-constitutions composed of one or more pieces of legislation, we will search for scholarly consensus from area specialists on exactly which documents are to be considered constitutional. For example, in the Israeli case, the constitution may be composed of the series of Basic Laws passed by ordinary parliamentary majority. Fortunately, at least for analytic purposes, formal constitutions are the norm and defining a state's constitution is largely straightforward. Nevertheless, we should be clear that problematic cases will be included in the dataset only with clear interpretative guidance concerning exactly which documents were coded, and an easy mechanism for users to exclude these cases from their analysis.

Constitutional Events - At this point we have identified 1792 events, or roughly 10 events for each country. As we describe in the section on source material, we have located the constitution and/or amendment text for most of these events and are hopeful that we will locate material for most of the other events. We derived this list by consulting comparative sources of constitutional history. For example, the Constitutions of the Countries of the World series includes documentation on the constitutional history for each country while other, less comprehensive, sources provide notes on the dates of major amendments and drafts for a selection of countries (Fitzgibbon 1948; Maddex 2001). Of course, during the coding process we will review the constitutional histories of countries in our sample more closely. Through that process, we will very likely find events that were overlooked as well as some amendments that can be excluded because they do not modify the constitution on any of the dimensions we are coding.