Seeing beyond the datasets: the unexploited structure of electoral and institutional data

We propose relational data modeling as a tool for replacing the ad hoc and uncoordinated approaches commonly used throughout the social sciences to gather, store, and disseminate data. We demonstrate relational data modeling using global electoral and political institutional data. We define a relational data model as a map of concepts, their attributes, and the relationships between concepts developed using a formal language and according to a set of rules. To demonstrate the methodology, we design a simple relational data model of six concepts: countries, parties, elections, districts, institutions, and election results. Furthermore, we introduce a data model to solve the particularly vexing issue of party discontinuity (party splits, mergers, and alliances). We show how the solution facilitates computational tasks, such as the calculation of core measures of political phenomena (ex: electoral volatility). Ultimately, a relational data approach will play a central role in collective investments to develop advanced data capabilities, and thereby advance the accuracy, pace, and transparency of scholarship in the social sciences.