Developing Social Science Data Methodologies

Prince Philip House, Carlton Terrace, 11th December 2015

Main organisers: Rob Procter, Celia Lury, Noortje Marres, Robin Wiliams, Hannah Knox, Philip Treleaven, Helen Margetts, Robert Doubleday

By exploiting new sources of naturally occurring data, social data science has the potential to provide a hitherto unmatched means of understanding populations, behaviour and social life. For example, the explosion of social media has given rise to a form of mass, self-reported data about users’ daily routines, perceptions of, and opinions about, events, as well as opportunities for implementation of new experimental research designs. However, if these opportunities are to be realised then social data science must find solutions to a number of important methodological challenges.
This workshop brings together academic and industry experts to identify key methodological issues, and emerging strategies for addressing the challenges. Through a series of carefully selected case studies it will examine how social data science can exploit data generated through social media. The workshop will build on these case studies to deliver a set of exemplars that will promote sound methodological practice in social data science and contribute to developing a robust, inter-disciplinary data science research culture.
The workshop will focus on four key themes:
1. Harvesting: sampling strategies for social media streams so that data collected is relevant to research questions.
2. Interpretation: assigning meaning to social media data, accounting for interactivity effects in digital networked environments and spurious correlations.
3. Methodological synthesis: building an inter-disciplinary methodological culture that bridges data- and hypothesis-driven traditions
4. Capacity building: training social scientists to use data science methods appropriately and sensitizing data scientists to social scientific concepts and methods


The Foundations of Social Data Science

British Library, 14th December 2015

Main organisers:Patrick Wolfe, Helen Margetts, Melinda Mills, Rob Procter, Robin Williams, Cecilia Mascolo

The vast quantities of heterogeneous data now available about people and their actions will revolutionise our understanding of human behaviour (Lazer et al, Science, 2009). These forms of data also challenge the underpinnings of modern mathematical and computational methods for the analysis of large-scale data. Together these two open problems motivate a new field: social data science. Here we propose a workshop that brings together experts in key relevant areas – the social sciences, the mathematical sciences, and computer science – to map out the foundations of social data science.

Social data science represents an exciting opportunity both for the Alan Turing Institute and for the UK science base more broadly. Until now, the field has been largely led by US researchers, under the auspices of computational social science. Given the large quantities of well-curated social data in the UK, we have a great opportunity to contribute to the leadership of this field. Building up social data science into a vibrant stream of data science research, which builds our understanding of human behaviour and informs interventions geared at changing it, will provide the Institute with a unique competitive advantage.

In order to pursue this goal, the workshop will focus on three key themes:

  1. It will explore ways that large-scale data can be modeled to form our understanding of individual human behaviour, and identifying the novel mathematics and algorithms to do so;
  2. Using and developing state-of-the-art quantitative methods, it will explore the complex relationship between large-scale social networks of influence & interaction, and behaviour;
  3. It will identify new experimental methods and designs that can bring causal inference and inform interventions at scale in the setting of social data science.