The Problem:
Electoral and institutional data are currently available and disseminated in a way which makes using the data very cumbersome and time-consuming. Dozens of purveyors of these data primarily use printed volumes, Excel sheets, and html tables to present the data. The process of taking these data and converting them into a format for computing important quantities of interest is tedious and prone to error, especially if one is interested in many countries over long spans of time. We aim to improve upon this state of affairs.

The Solution:
Electoral and institutional data are inherently systematic and relational. For example, information about political parties, national legislatures and executives, and the electoral rules which govern their selection  are not discrete bundles of information, but related bundles of information. Unlike Excel sheets and printed tables, relational databases are well-equipped to model the relationships that exist between the information. A relational database physically connects disparate data tables and maps according to their underlying logical structure by creating indices linking common variables from separate tables and maps.
A simple application would function as follows: a database of electoral returns is built by linking together a) a table of political parties; with b) a table of national electoral calendars; building c) a third table which incorporates the parties and the elections while adding the number of seats and votes earned by a given party in a given election; and d) allowing maps to display results. We could extend this simple example in two ways. First, we could include basic attributes of either the parties, the elections, and/or the election results. From there we could generate a wide array of measures which derive from the attribute. For example, if we add a single variable which classifies each political party on a left/right scale, we can derive the average ideological position of the legislature using a single formula written once and applied automatically to all the possible legislative sessions; or, we can calculate the share of seats controlled by parties of a particular ideological flavor, again by constructing a single formula and applying it globally. Second, if we include election results on both the legislature and the executive, we could build a simple formula to indicate whether there is divided or unified government. Both of these extensions are possible because the data on parties, executives, legislatures, and elections are all linking together in a web of relationships; like a web, one place in a database is associated, either directly or indirectly, with all other places in the database. Figure 1 below displays four tables and the linkages that exist between them for the variables used in this example.

Figure 1: Web of Relationships-A Simple Example

Furthermore, with our solution, all derivatives—which are simple aggregations of the basic data elements—can be broken down, with the formulas reported, thereby enhancing the transparency and replicability of the results. Ultimately, this example could grow to incorporate information on executives and their cabinets, central bankers, courts, and supranational parliaments and their selection mechanisms, or membership of political parties in international organizations, such as Socialist or Christian Democrat International. The accompanying demonstration version of this database shows this process in action.