Computational Mapping of Alternative Dispute Resolution Institutions
PI: Cyanne Loyle (Political Science)
Personal department research funds allocated for graduate student tuition funding.
This project aims to develop a comprehensive global database of Alternative Dispute Resolution Institutions (ADRIs): formal or informal organizations that provide services for resolving disputes outside state legal systems. The project would immensely benefit from the use of advanced computational approaches to identify, classify, and analyze these institutions across multiple data sources, creating the first systematic dataset of its kind in social science research. ADRIs represent a critical but understudied component of governance, particularly in regions with limited state capacity and armed conflicts. They include lineage-based institutions, customary law institutions, religious councils, councils of elders, chieftaincies, armed non-state actors, civil society organizations, and matrilineal groups, among others. These institutions can have either complementary or competing relationships with state authorities, significantly affecting governance and political stability.
Key sources for the data include preexisting data collection efforts, such as the Varieties of Democracy project’s country-year expert commentaries, Armed Conflict Location & Event Data’s non-violent event records, and the GDELT event databases. While these three sources have the potential to include ADRIs, none seek to definitively code their presence. Additional data can be sourced more broadly from Lexis Nexis legal and news archives. These data sources offer varying perspectives and coverages of ADRIs, from recognizably powerful bodies to those that operate in competition with or in the absence of the state. Furthermore, each source comes with its own biases, such as overrepresentation of armed or nationally significant ADRIs, or underrepresentation of peaceful, peripheral institutions. The computational challenge involves extracting mentions of ADRIs across different naming conventions and languages, classifying them by type and their relationship to the state, and tracking their political relevance over time. LLMs are particularly well-suited for this task, as they can process unstructured text and standardize categories across sources. The goal of the project is to generate country-level indicators of ADRI presence and country-year measures of their political salience, producing a standardized dataset for use in social science research. Crucially, this work would benefit from a Junior Researcher with dual expertise: machine learning and domain knowledge in political science and alternative sources of governance. Domain expertise is essential to fine-tune LLMs for meaningful classifications of ADRIs using detailed social science literature on hybrid governance systems. The researcher needs to have a deep understanding of the potential biases present across data sources, in the LLM training data, and propose theory-grounded solutions. The researcher needs to fine-tune and interpret model outputs ensuring that they accurately reflect the proposed ADRI definitions.
This work aligns closely with the ICDS mission of catalyzing interdisciplinary, computational and data-driven discovery to address questions of scientific and societal importance. By integrating social science expertise with computational tools, the project will provide foundational data for the social science community. Incorporating an understanding of these structures has implications for theorizing the origins and foundations of state-building, political order, and rule of law. The data will provide critical insights into the role of ADRIs for structuring legitimacy, demand for state institutions, and the resilience of political order.
Project objectives
- Demonstrate feasibility of data collection procedure to stimulate future funding activity
- Submit a peer-reviewed article
Long-term goal – Develop new research agenda on Alternative Dispute Resolution Institutions based on data collected through the proposal
Current engagement with ICDS The PI currently serves on ICDS’s faculty advisory board. Loyle has previously earned a RISE seed grant which was used to scrap web data on civilian engagement with authoritarian governments. Loyle is also the dissertation advisor to three students in the dual degree SODA program which engages the study of social data analytics in Political Science.