Combining towards effective research design in digital sociology

I am currently trying to devise a workable research design for a two and a half year project. Its absence, needless to say, isn’t stopping me jumping into the field for what I call the ‘landscape mapping’ phase of the project – which I half suspect will last longer than the part that’s directed by a robust and clear research design.

The project is at once a sociology of a profession (journalism), an organisation (a media group) and a distributed practice (blogging and online discussion). It is my ambition to take an action research approach and allow the agenda to be influenced by the needs and interests of respondents / organisational members, hopefully without sacrificing either scientific robustness or theoretical relevance. To put this another way, at the end of the journey I want to be able to write a book for an academic audience, but I also want to find time for plenty of detours and stop-offs along the road to enable me to engage very intensively with actors in the field about the emerging findings, and to interpret them collaboratively.

That in itself is a difficult combination to achieve. However, the combinations that are now giving me most pause for thought are more of a technical, methodological, interpretive and analytical nature than about subjects, objects and audiences. Like Ruppert, Law & Savage (2013), I am becoming aware of how “social knowledge is more visibly non-coherent than it was in the recent past”: that digital data is increasingly ‘self-generated’, ubiquitous and lively, generated outside the academy and less dependent on the capacity of actors to learn to behave as the kinds of publics experts like us construct them as in the design of our research instruments. It follows, they argue, that “we will need to vary the magnification as we explore the chains of relations and practices enrolled in the social science apparatus” and “explore fields of devices as relational spaces where some devices survive and dominate in particular locations while others are eclipsed, at least for the moment” (Ruppert, Law & Savage 2013: 19-20). Never has it been more relevant to strive for ways of combining views of the social from above and from below. Or, better perhaps, to configure the social successively and iteratively from different angles of attack, equipped with devices that afford different conceptual and technical preconceptions.

But what, exactly, am I trying to combine in my own project? I’m going to have a mixture of data and metadata: interview transcripts, samples of text from blogs and online discussion, detailed fieldwork observation notes and a range of pseudo-constitutional documents on the one hand; user databases and various forms of scraped online social data (Marres & Weltevrede 2013) on the other. I’m taken by the idea that such a heterogeneous corpus of texts can be analysed as … a corpus as long as you find a way of opening up an intermediate space for its active and iterative interpretation, as Chateauraynaud argues: “one can re-equip the social scientific laboratory by installing an intermediate workspace between the Web and the confined universe of sociological research … What we need is a mediator or an intermediary placed between the traditional world of knowledge and the digital world, without being configured entirely by one or the other, and which carries sufficient autonomy to be able to produce cognitive events of a new genre.” (2007: 3) The point is really to use that space to construct an intelligent corpus by continually interpreting the dossier you’re amassing – describing it, indexing it, annotating it, calibrating it, testing its coherence. To find new ways to responsibilise the datasets we construct that are as good as the techniques used by qualitative researchers to understand and mark the transformations unavoidably set in train when working with traditional documentary sources (Hakim 1993). One purpose of techniques like framework analysis, for example, is to format a dataset for one’s own study purposes, marking it as different from ‘raw’ data sources so that one is forced to take ownership of it by acknowledging that one’s own purposes are different from those of the authors of the documents.

So I’m looking for a digital methods equivalent of framework that can help me construct and analyse a corpus combining all of the following elements.

Technical level

Big data approaches

Mass observation techniques

Metadata and self-generated online social data

Ethnographic approaches

Following the actors

Rich textual discursive data

Methodological level

Disintermediated and collaborative data-gathering

Theory-based and expert data-gathering

Interpretive level

Interpretation through action research where knowledge is framed pragmatically by its use and relevance to a media organisation and journalists

Interpretation that engages with the academic literature on professions, organisations and the Internet / eParticipation in order to generate knowledge

Analytical level

Field analysis at the macro-level of an organisation and a profession

Discourse analysis at the micro-level of speech acts and communicative interaction, intermediary objects and their translation, localised regimes of justification


Some of these combinations are very difficult to do: field theory is generally held to be so bequeathed to structural explanation as to be incompatible with pragmatic and ethnomethodological sensibilities I insist on practising as a researcher in the terrain. Nevertheless, my hunch is that my research will be the better if I can pull them off. More than that, my hunch is that I need to pull them off to do this particular research properly. To be sure that the results are worth something.

What I am less sure about at the moment is precisely what leverage I get from each of the perspectives I introduce. How to weigh their importance (not least for practical purposes of time allocation)? And then, how do the combinations work – how do I make them perform work, i.e. make them stick together, as the actor network theorists would say? One idea is to think of the different views as different phases of the same process. For example, Aguiton and Cardon (2008) have suggested that you glimpse a socialising dynamic at work when your focus is on the small-scale or the physical, and an amplifying dynamic at work when your focus is on the large-scale or the virtual. They use this insight to understand processes of open innovation, but we could easily extend it to any case of emerging social practice that makes use of online communication. Another idea would be to take Pentland and Feldman’s proposal (2005) about the dialectical relationship between change and routine in organisations. Does one perspective make you more likely to notice the ostensive scripts that actors follow when they try to emulate an ideal, while another sensitises you to a submerged but active repertoire of performative scripts that actors invoke when they subvert (or repair) organisational routines?

Accordingly, I sense that the research design needs to be as iterative as possible in terms of chronology: zooming in and zooming back out as often as possible to alternate between perspectives that will contradict and interrogate as much as they complement one another.

Cooren and Fairhurst (2004) ask us to consider how the programmes of action of the groups we study intersect around mutual interests. If the aim of research is to attain an omniscient version of a story, this might therefore have a network shape centred around a key helper or co-orientation object. Once we then recognise ourselves as actors in that network, it is incumbent on us to go one step further and ask where and how their programmes of action coincide with our programme of action as researchers so as to identify potential boundary objects around which a joint action domain for ‘research in the open air’ might productively coalesce.

So my quest for a good study design is, I hope, not just another tilt at structure-agency windmills. It’s the necessary step to preventively screen or ‘proof’ my data collection techniques by truly appropriating the data and to recognise the parallels (which represent possibilities for collaboration) with the indigenous acts of recording, comparing, combining and interpreting that intelligent and interested members of the organisations I am studying perform on distinct but overlapping knowledge corpuses. And hence to open Chateauraynaud’s intermediary space.


Aguiton, C. & Cardon, D. (2008) ‘Web participative et innovation collective’ Hermès 50: 77-82.

Chateauraynaud, F. (2007) ‘Faust, la société de l’information et le village sociologique’, texte de la communication aux Journées Figures du lettré et Technologies numériques: une chimère contemporaine? Paris, March 2007. Available at:

Cooren, F. & Fairhurst, G. (2004) ‘Speech Timing and Spacing: The Phenomenon of Organizational Closure’ Organization 11(6): 793-824.

Hakim, C. (1993) ‘Research Analysis of Administrative Records’ in Hammersley, M. (ed.) Social Research: Philosophy, Politics and Practice, London: Sage: 131-45.

Marres, N. & Weltevrede, E. (2013) ‘Scraping the Social? Issues in live social research’ Journal of Cultural Economy 6(3): 313-35.

Pentland, B. & Feldman, M. (2005) ‘Organizational routines as a unit of analysis’ Industrial and Corporate Change 14: 793-815.

Ruppert, E., Law, J. & Savage, M. (2013) ‘Reassembling Social Science Methods: The Challenge of Digital Devices’ Theory, Culture & Society DOI: 10.1177/0263276413484941.

Simon Smith is Marie Curie Intra-European Research Fellow at the Institute for Sociology, Slovak Academy of Sciences, Bratislava. He can be contacted at

Categories: Digital Sociology

Tags: , ,

Leave a Reply

Your email address will not be published. Required fields are marked *