Format

P274. Big and Creative Data in Public Policy and Public Administration Research: Opportunities, Methods and Practices

Methods & Research Design
PANEL CHAIR(S)
Vassilis Karokis-Mavrikos
University of Surrey
Main chair
Maria Patouna
University of Oxford
Second chair
CATEGORISATION
POLICY TOPIC
Methods & Research Design
SECTOR
—
KEYWORDS
big data computational methods
GENERAL OBJECTIVES, RESEARCH QUESTIONS AND SCIENTIFIC RELEVANCE

The data available for studying public policy and public administration have changed fundamentally in scale, variety and accessibility, and with them the questions the field can ask. Governments publish legislation, consultations, parliamentary records, budgets and procurement data in machine-readable form; administrative records are increasingly accessible for research; and technology firms and digital platforms are releasing large datasets on how citizens, organisations and officials behave online and use new technologies. At the same time, computational tools, from text-as-data methods and machine learning to open-source transformer and large language models, make it feasible to analyse such material at a scale and granularity previously out of reach (Lazer et al., 2009; Grimmer, Roberts & Stewart, 2022). Much of this is repurposed, or “readymade”, data, generated for purposes other than research, which invites creativity in connecting it to the questions that matter to the field (Salganik, 2018).
Political science has embraced these developments for some time (Wilkerson & Casas, 2017), and public policy and public administration scholarship is increasingly doing so, such as through applications of text analysis and machine learning to agency priorities and organisational reputation (Hollibaugh, 2019; Anastasopoulos & Whitford, 2019). Advances in large language models open further possibilities, from automated text annotation (Gilardi, Alizadeh & Kubli, 2023) to the study of how citizens and officials themselves use AI in their dealings with the state. As these opportunities multiply, there is value in bringing together researchers working with big and creative data across policy and administration research to share approaches, compare designs and reflect on what these data make newly visible.
The panel asks: What new questions about public policy and public administration can big and creative data help answer? How can such data be connected to established theories, concepts and debates in the field? Which data sources, research designs and computational tools are proving most productive, and under what conditions? And what ethical, practical and epistemological considerations accompany their use?
The panel explores four propositions. First, big and creative data bring into view actors, interactions and stages of policymaking and administration that have been difficult to observe at scale, from citizen-state encounters to implementation and service delivery. Second, these data allow established theories and concepts in policy process, public management and governance research to be examined in a new light, while also opening avenues for new theorising. Third, open models and data repositories lower the barriers to computational research, supporting broader participation, replication and comparison across countries and languages, and the building of shared, systematic knowledge bases for the field. Fourth, big and creative data can strengthen the dialogue between research and practice, providing timelier and more granular evidence for policymakers and public managers.
The panel’s scientific relevance lies in creating a dedicated forum at IPPA for data-intensive research in public policy and public administration, and in building a community of practice around its opportunities, methods and responsibilities.

CALL FOR PAPERS

We invite papers that use big and creative data to advance research in public policy and public administration, broadly defined. The panel is quantitatively and computationally oriented and welcomes both substantive empirical studies and methodological contributions, from any policy sector, level of government or country. We are particularly interested in papers that show how new data sources and tools connect to substantive questions and theories in the field. Papers may use these data to test, refine or extend existing theories, or to develop new concepts grounded in the patterns they reveal. Contributions may draw on public policy, public administration and management, political science, economics, computer science or data science, and interdisciplinary collaborations are especially welcome. Topics include, but are not limited to:
•          text-as-data approaches to legislation, regulation, consultations, parliamentary debates, court decisions and media coverage;
•          applications of machine learning, open-source transformer models and large language models to classify, annotate, scale or compare policy-relevant material;
•          administrative, transactional and linked data in the study of public services, implementation and performance;
•          digital trace, social media and platform data, including data on how citizens and public servants use generative AI;
•          creative and repurposed sources such as web archives, images, audio, satellite imagery and open datasets released by technology firms;
•          network data on policy actors, coalitions, lobbying and intergovernmental relations;
•          big data approaches to policy process questions, including agenda setting, stakeholder influence, diffusion and policy change;
•          geospatial and sensor data on the distribution, accessibility and delivery of public services;
•          the use of big and creative data to generate timely evidence for policymakers and public managers;
•          the construction of new comparative and longitudinal datasets and shared research infrastructures for the field.
We welcome mature papers as well as work in progress with a clear data or methodological contribution, and especially encourage doctoral and early-career researchers, as well as papers that make their data, models or tools openly available.