Given our shared interest in big data and public policy, we wanted to find out what research was going on across the university that might link these topics, and in particular, what social science research was using big data to address public policy issues. Others in similar ‘coordinator’ or research management roles in universities and research councils may be asking similar questions within their respective fields. With an emphasis on interdisciplinary approaches in current large-scale research programmes, it is important to get researchers together to talk to each other about the work they are doing.
The relevance and contribution of social science researchers to big data studies was emphasized throughout the day; as one contributor put it, “social science data is big data” because social science already constructs complex models to try and map and understand the world and our place in it – how we work, how we move, how we interact as people. Social science models are now global social models that can map traffic flows, land use, but there are many times of social science data – spatial, population, social, behavioural, financial and political.
Developing models that attempt to understand these data is important because many of the most significant policy challenges are global; problems such as food security, conflict, disease, pandemics and climate change. Social science provides important context and theoretical insight to explain and understand big data. New insight into enduring social problems will most likely come through collaboration between researchers linking innovative method and data gathering with established knowledge and understanding of the problem.
To facilitate fruitful collaboration and make the most of the potential that big data offers for social science research and public policy, the workshop highlighted the following challenges for social science researchers:
- The challenge of big data causes us to reflect on what are social science research questions: the data do not ‘speak for themselves”.
- The importance of multi-method studies, an established tradition in social science – in order to ‘triangulate’ and verify findings.
- Make data as open as possible – this leads to improvements in the data; more studies carried out enables replication and verification of findings.
- Respond to data generating new research questions and models – be open to new learning from data.
- Social scientists and policy makers may not need to become computer scientists but do need to develop some computational skills in order to understand data and the decisions that are being taken with data.
Workshop participants learnt about new data sources (such as mobile sensors; location and behaviour data; Twitter feeds; satellite mapping; text mining); how new policy models are being constructed to include new data sources; the challenges of handling new data sources (real time, adaptive, behavioural data, whole population rather than sample data; social data; fine grained data; ‘tall and fat’ data; ‘messy’ data; multi-lingual data) and how best to integrate new data with existing hypotheses and techniques that have been developed to address public policy questions.
The research questions that were addressed through the day show the huge range of public policy questions that Big Data may have relevance to. Do police on the street make a difference? Can shipping data be used to detect piracy and illegal drug imports? Can you identify power theft in developing countries? Can we infer changes in mood from the way that people move? Will workplace technology mean the end of work? What are the levels of corruption in government procurement?
Big data challenges policy makers because it can offer real-time results that require a rapid, adaptive policy in return. Big data is often a rich data, offering refined data points and high quality observations that span different levels of analysis and the data is often fragmented, so researchers spend time trying to locate and access diverse data sets. The data requires translation – between languages, and also between disciplines. There may be missing data that causes concern for researchers, or ‘blind spots’ in data where significant sources are ignored in the research because they are not represented in the data sets. And social science researchers who connect with public policy are well aware that measuring is not a passive act. The application of big data represents significant ethical as well as practical challenges, not least the growing predominance of data driven policy and the effect of measurement on behaviour in the public sphere.
How best to capitalize on what big data has to offer social science and public policy? This was a recurring theme of the day.
Read our report here