Doing Open Social Science: Advancing Transparency, Reproducibility and Trust in Research

Source: Turing Way project 

• to speed up and encourage the re-use, re-analysis, iteration, and improvement of research over time and across researchers, so that social science evidence cumulates, instead of (as often now) being  frittered away by projects beginning again from naïve premises; 
• to connect this research more effectively to what Lindblom and Cohen called  ‘ordinary knowledge’ and developed bodies of expertise outside academic work (e.g. in  public policies and organizing businesses and markets effectively);  
• to democratize access to social science research for citizens and other potential users in civil society; and 
• to build public trust in the value of the discipline group’s scientific research. 

Enhancing the openness of academic work requires an important change of working culture (or organizational cultures) amongst researchers themselves, one that only they can accomplish, and that must encompass all the siloed subjects in the discipline group in similar ways. In particular, qualitative and quantitative work must both become more open in balanced and broad-front advance if distortions are to be avoided and if the often problematic meanings of quantitative research are to be assured via triangulation.  

Who Drives Open Social Science?

Funders and universities can usefully encourage open social science approaches. And academic-related professional support staff (in libraries, research offices, and PhD schools and academies) can help disseminate (general) knowledge on making research more transparent. But if the change is to be sustainable, OSS changes must be ‘owned’ and promoted by researchers, rather than appearing to them as new and burdensome forms of external regulation. It is researchers  themselves who must adopt diverse innovative initiatives and approaches, meshing them together with their existing research methods and practices. Social scientists will need to autonomously adopt ‘open’ practices for their own, good reasons – because ‘open’ works better to get research done, explained, accepted, re-used and acted on.  

Insights from Doing Open Social Science: A Guide for Researchers

In Doing Open Social Science: A Guide for Researchers (LSE Press, 2026, open access) Tim Monteith and I have drawn on an 18 month Civica I research project and the generously shared expertise of social scientists from across its seven participating universities to condense and explain the emergent (and yet still rapidly advancing) open social science best practice across our discipline group. 

Progress and Limitations in Quantitative Research

Existing progress down these avenues has been great amongst quantitative researchers using strong hypothetico-deductive models of investigation, for whom the issues involved in pre-registering research designs and at the end of research permanently depositing already anonymized OA numerical datasets have been less. Allied with changes that have made journal articles increasingly open access and equipped with initially working ‘replication archives’, it may seem that openness in these fields has been already solved. Yet in practice many research datasets still lack useful metadata (so are hard to find), have incomplete documentation (so are hard to make sense of), lack descriptive stats (essential if re-users are to understand their contents) and may require specialist software or esoteric analysis skills to understand, while ‘replication archives’ for journal often stop working after a few years. Recent estimates on quantitative projects show that at best half of published work in political science, economics or psychology is practically reproducible, and much less in fields like sociology and education. So Parts I and II of our book condense a lot of advice useful for any kind of research project on doing better at openness by starting well at the outset, documenting as you go, mashing or re-using other data carefully, compiling and following a data management plan (even for small projects) and fully documenting and providing data for each exhibits included in research articles or books. 

For qualitative researchers the barriers to openness have been greater, but we show enhancing reproducibility in many ways is fully compatible with a reliance of exploratory data analysis and research methods that are inherently grounded in understanding research subjects’ own experiences and meanings. Part III of the book also shows how many of the difficulties in better demonstrating the provenance of data gathered using interviews, case studies, fieldwork and systematic documentation or archival analysis can be creatively mitigated. Depositing anonymized data in straightforward ways may not be ethically feasible, better demonstrating the high quality and reliability of the evidence obtained and used to found analyses can be advanced in many innovative and already practicable ways. 

The Future of Open Social Science

The open social science agenda is one that will take a decade or more to fully implement. It seeks to establish more clearly than ever before the provenance of our research - in a kind of academic ‘show and tell’ that makes visible not just the final writing up of results in articles and books and their accompanying exhibits, but also gives direct access to the underlying evidence and data behind research claims in ways that can be easily checked, re-used and cumulated in the digital era. Being open about evidence, plus achieving open access across all research publications (journal articles and long-form books and chapters alike), can enable iterative improvements in knowledge via the ability to track changes and previous analysis through well-documented data sources and transparent publications. 

by Patrick Dunleavy and Tim Monteath 

Patrick Dunleavy is Emeritus Professor of Political Science and Public Policy at the London School of Economics and Political Science, and a Fellow of the British Academy and of the Academy of Social Sciences. In 2023-5 he lead the Civica I Open Science project. 

Timothy Monteath is Assistant Professor in Data Visualization in the Centre for Interdisciplinary Methods at the  University of Warwick, and previously was Researcher on the LSE’s Civica open social science project (2023-4). He also completed his doctorate in Sociology at LSE.