I’m currently writing up a paper for submission to the Internet, Politics and Policy 2014 conference to be held by the OII in September. My paper – which draws substantially on interviews conducted as part of the Sloan Foundation-funded project of which I’m part – asks whether and to what extent the measurement of public opinion has been transformed by the new availability of socially-generated sources of big data, such as social media postings and search queries, and the tools which allow us to analyse them.
The paper itself will focus primarily on the grounded experiences of the interviewees – all of whom have used these newly available data sources – to offer insight into the novel challenges to validity posed by this new and distinctive form of research. In particular, I’ve isolated three widespread challenges – around reliability, representativeness and replicability – which emerged from the interview data. (I’ve blogged about two of these already.) But the paper will begin with a literature review offering a very brief overview of the thousands of years of rich debate concerning what exactly we mean by public opinion. Brushing up on the literature in this area has been really rewarding, touching on thinkers as diverse as Tocqueville, Tarde and Bourdieu.
What sticks out the most from this reading is that public opinion is a concept highly mediated by the history of technology in society. At the most basic level, this is captured by two shifts: the increasing size of the political unit and physical dispersion of those within it, (from the ancient agora to the modern nation state), and the development of ever more sophisticated technology to enable communication to pass over these greater distances (from Gutenberg to Zuckerberg).
For most of history, from antiquity onwards, the pace of the first trend has outstripped the second: societies grew faster than communication technologies could accommodate effectively, at least as viewed through an (admittedly inadequate) modern, western perspective. But since Gutenberg’s invention of the printing press, the emergence of various ICTs has enabled more people to communicate, faster and further, than ever before. This is to some extent summed up by Lewis Mumford’s oft-cited aphorism (1934, p.241), said about radio but which can be used to describe numerous ICTs before and since:
Plato defined the limits of a city as the number of people who could hear the voice of a single orator: today those limits do not define a city but a civilization.
Mumford here blurs the distinction between a city-state, whose authority could extend as far as a voice could carry, and a modern nation state, in which technology plays a mediating role in communicating its authority. Yet this analysis is misleading for a couple of reasons. Firstly and obviously, sophisticated communicative technology is not a pre-requisite for large-scale civilization. Secondly and more pertinently, by tying together the physical with the virtual, Mumford’s quote misses one of the crucial aspects of Athenian democracy. As John Durham Peters notes,
Modern publics differ importantly from ancient assemblies – above all, in the power of the hearers to shape the flow of public discourse … every Athenian citizen had the theoretical option of contributing … This kind of civic cocreation of public discourse seems impossible for the modern “public” (1995, p.15).
ICTs such as the newspaper, radio and television are thus imperfect substitutes, enabling a predominantly monodirectional flow of communication from writer and speaker to reader and listener, where the writers and speakers are typically part of a political/media elite. (Notable exceptions to this include ‘Letters to the Editor’ and radio phone-ins, but these are conspicuous by their limited column inches and airtime.)
The emergence of these forms of communication have affected our understanding of public opinion, both in theory and practice. In theoretical terms, thinkers from Alexis de Tocqueville to Benedict Anderson have sought to elucidate an understanding of one ICT in particular – the newspaper – as a symbolic representation of the public as an actor. In this view, the public has the potential for action, but (because of the monodirectionality of the medium and the physical dispersion of the public) the newspaper affords a member of the public the ability to imagine her fellow citizens, rather than see or talk to them as in the Athenian polis. (It might be hypothesised here that it is the very constraints of the medium, via the homogeneity of the media elite and the high cost of content creation, that promote this sense of togetherness amongst the audience: everyone is reading, watching or listening – not only at the same time, but to the same thing.)
On the empirical question about how we might attempt to measure public opinion, scholarship here, too, has adapted to the constraints of the traditional mass media, which as noted offer no scope for large-scale expression of substantive opinion. And since the primary method of widespread political expression in a representative democracy – the general election – is both rare (occurring only every few years) and narrow (offering only a few choices), researchers have had to construct their own methods for gauging public opinion in more frequent and finely-grained ways. For this, they developed the opinion poll. Conducted around the world on a regular basis, opinion polls done well provide a statistically rigorous and replicable foundation for making substantive claims about the views of a population generally.
Yet even when it is well executed, the process of opinion polling raises serious epistemological questions. It has been claimed, for example, that the results of a poll are neither strictly ‘public’ (with interviews taking place in private to ensure anonymity) nor strictly ‘opinion’ (with responses not expressed spontaneously but in response to a specific set of questions.) Perhaps the most effective criticism of opinion polls was offered by Pierre Bourdieu, who in his article ‘Public Opinion Does Not Exist’ (1979), put forward three false assumptions that their use inculcates: that everyone has an opinion; that every opinion has the same value as every other; and that everyone agrees on the question and its need to be asked.
The emergence of the Internet and particularly social media, I argue, offers a new and fundamentally different understanding of public opinion, both in theory and practice. Most crucially, this ICT facilitates a far larger and more meaningful role for members of the public to communicate on the same basis as traditional elites; in fact, the core structure of the Internet permits no particular technical distinction between ‘public’ and ‘elite’. Combined with the exponentially smaller costs of content production, this means that billions of people around the world have the ability to express themselves online, every day, in myriad ways.
In purely technological terms, this system brings us tantalisingly close to a fuller, truer realisation of the Athenian notion of public opinion by participation, as opposed to the representational paradigm of the mass media age. In reality, of course, the situation is different: old inequalities are replicated and new ones invented on the Internet, as for example my research has shown. Yet the sheer potential for the Internet to transform how public opinion is created and acted upon is immense.
A parallel transformation concerns how public opinion is measured scientifically. Fortunately, as the millions upon millions of expressions of opinion now occurring online have stacked up, so our technical ability to capture, store and analyse this data has also increased sizeably. This has transformed our ability to measure public opinion in the Internet era. Where before we would construct a small sample, now we can look at a population as a whole. Where before we would devise a set of questions to ask (imparting this, no doubt, with our own biases and interests), now we can judge objectively what people care about through their unprovoked expressions. And where before we would inquire into opinion privately, we can now observe opinion publicly and unobtrusively. Notably, all three of Bourdieu’s critiques of public opinion research are far more answerable in the age of social media: we can accept that not everyone has an opinion, we can use tools like network analysis and sentiment analysis to ascertain the differential ‘value’ of different opinions, and we can accept that not everyone agrees on the right question by not asking questions in the first place.
This brings us, of course, to the fresh limitations and novel issues around public opinion research conducted online, which as noted are the subject of my forthcoming paper for IPP 2014. But as this post has argued, these new challenges need to be located in the rich and longstanding debates which have characterised our understanding of public opinion for many millennia.
Bourdieu, P. (1979). Public opinion does not exist. Communication and class struggle, 1, 124-310.
Peters, J.D. (1995). Historical tensions in the concept of public opinion. In Glasser, T.L. and Salmon, C.T. (eds.) Public Opinion and the Communication of Consent.
Mumford, L. (1934). Technics and Civilizations. New York: Harcourt Brace Jovanovich.