The quantitative analysis in the working papers generated by this project uses information drawn from an original dataset. You may download an SPSS version of the dataset by clicking on the appropriate link below. Although there is no charge for the use of this information, please note that all material on this site is copyrighted and should be cited as indicated in our copyright page. Please read the data documentation carefully to avoid misuse of the data.
Constitution drafting is an example of what Charles Beitz has called “complex proceduralism.” The process embraces a bundle of distinct rules and procedures. The “process” dataset records over 130 items of information about the rules used in each case in this study, along with contextual information. Note that there are missing values in some of the cases. Additions and corrections are welcome.
The information on the dependent variable in this study, violence, originally came from two sources. The PRS Group produces monthly data on internal conflict and military involvement in politics (ICRG Table 3b). This dataset is available for purchase commercially. It is not available through our project. The ICRG violence data omit several countries in the constitution writing study and in some cases the ICRG series does not cover all the dates of interest. We developed a patch for the missing data, using information drawn from Keesings Archive and from Lexis-Nexis news service. Since the original study, better sources of data on violence have become available, and we recommend that you use those (for example, see the work by PRIO).
It is important to emphasize that the information contained in these databases is of a highly formal nature. The project records formal rules and written terms. It does not adequately capture informal political deals, social relationships, acceptance of the rule of law, or highly localized interpretations of constitutional language. In some cases we know that these elements trump the formal rules in importance.
Further, we advise great caution in using these data to draw causal inferences about the effect of process on outcomes. The data are most helpful for identifying particular procedures countries used and for looking at the proximate (immediate) impact of a given procedural choice on willingness to compromise or trust (not easy things to measure). Because constitutions are complex bundles of procedures, it is very hard to infer much about the effect of a process, overall, on an outcome like violence or inclusion. Further, in trying to understand such outcomes there are always many confounding variables and selection effects. Intellectual integrity may require we all use the data mainly for descriptive purposes and not for causal analysis.