Gathering and analysing big data for deliberative democracy: lessons from a case comparison

Deliberative empirical research is increasingly occupied with questions of scale, inspired by closely-related and long-standing concerns about the impact of small-scale deliberation on wider political processes, and the more general contextual concerns of deliberative systems approaches. This shift in the scale of deliberative democracy’s empirical view is in turn inspiring searches for data that themselves capture aspects of large-scale action, and methods to analyse that data. However, generating and analysing big data sets involves many analytic choices and careful interpretation. While some techniques can give a more inductive view of the data than others, we should not think for a moment that big data is natural and neutral. This paper outlines some of these issues and choices in the context of a comparative study of online political talk over two years in two cases: the Scottish independence referendum of 2014 and a campaign to ‘Recognise’ Aboriginal and Torres Strait Islander peoples in the Australian constitution that ended in 2017. The comparison reveals much about the hard choices that have to be made when generating and analysing big data sets.