And we would have been wrong. Before getting into anything of the kind, Gareth Harris instead took us on a highly interesting and thought-provoking journey through some of the epistemological issues and debates associated with the broader world of computer assisted qualitative data analysis software (CAQDAS – also, see the CAQDAS Networking Project). As it turns out, CAQDAS are much more than mundane research tools, but in many ways are at the fulcrum of contemporary debates about the interface of research and technology. Gareth kindly provided his slides used in this journey, which augment much of my reflections below, as well as provide some useful links.
Gareth began by pointing out the CAQDAS has proceeded through three fairly distinct generations:
1. Search and retrieval of text
2. The coding of multiple textual fragments, which can then be retrieved as coded themes or categories
3. Theory-building, in other words, looking at the relations between categories (e.g. though the use of visualisation tools) in order to build higher-order classifications and categories.
In the 1990s, there were quite a number of critiques of and debates about the uses of software for qualitative research, and Gareth pointed out that these were pitched almost completely in relation to the second generation – using software for simple data coding. As a result, there has been little debate, at least so far (Gareth estimates they may indeed kick off soon), about the emergence of third generation CAQDAS – using software for more complex theory-building. What is so interesting about this third generation, Gareth noted, is that it aligns software like NVivo ever closer to the inductive approaches of Grounded Theory. This, of course, it not necessarily a bad thing, but it does highlight the implicit incorporation of a fairly specific inductive methodology into a software package. This at least potentially raises rather more sticky issues than the second generation, which is little more than a faster and more secure way to code data. As in, it did little more than computerise that which one would otherwise have achieved through the use of such materials (technologies?) as a stack printed photocopies, multi-coloured highlighters, a pair of scissors, and a glue stick.
In this context, Gareth highlighted a very important question, harking back to the overall themes of the workshop, and particularly our introductory session. Does CAQDAS have its own ‘effects’ (positive and negative) on our research? In other words, is a CAQDAS package like NVivo a neutral tool of our autonomous methodological actions, or does it have agency and channel our research in some ways? On the one hand, one response is that it is indeed primarily a tool for our research practices and decisions. This is a response often made to counter critiques of CAQDAS – which suggest that it is mechanistic, decontextualising, a fetishisation of coding even (see Gareth’s slides for more) – with an argument that all this really depends on how CAQDAS is actually used. On the other hand, however, perhaps it is naïve to take this claim too far. After all, CAQDAS must channel research, in the same way that writing an essay using word processing software entails a fundamentally different process than writing by hand (e.g. one continuously edits, rather than in more fixed stages). So, it might be seen as technologies with certain capacities and constraints, but which also comes into contact with a researchers’ know-how, practical work and ethics in doing research. And in this process of contact, Gareth seemed to suggest, we find a tool which allows for research practice to potentially be much more transparent and accountable than that based on paper (rather different, Gareth emphasised, than any erroneous claim that CAQDAS makes qualitative research ‘reliable’, a concept with strong connections to positivism)
It’s worth mentioning an interesting tidbit Gareth pointed out early on in the workshop: there is a rather good chance that we might see a situation in 5 or 10 years time where the more extensive and thorough literature reviews are being conducted using CAQDAS. Assuming one has a library of electronic materials, this is easy to see. It would not only provide a very effective way to code, organise and retrieve text fragments of interest across several sources, but would provide interesting ways to compare how authors have dealt with similar concepts, to visualise connections between groups of scholarly communities and their ideas, and much more. Like any technological change, there would be drawbacks; but the advantages for large-scale literature reviews, on complex subjects, seem quite clear.
Now, you’ll remember of course, we were in the lab, ready for ‘training’. And we did spent some time taking NVivo 9 through its paces. Certainly, participants had a chance to gain some initial exposure to the software, see its overall architecture. But in having had such a good and intellectually interesting overview, I’d expect most participants likely left thinking about their exposure to NVivo within a much bigger picture of research practice and technology.
By Scott Rodgers