This week’s #phdchat topic was the process of the literature review. There was talk about using NVivo to handle this integral aspect of the thesis, and I mentioned that I’d had a little experience using it in this way, and that I had written a review. It was written for a compulsory research training module I did last year (hence being a tad formulaic), but a few people expressed an interested in it so I will put it up here. An extract of the review portion is below, as this may be the most useful to you. You can see the full paper (including the bibliography) here if you like.
Reflections on NVivo
This early in my experience of coding I find it difficult to reflect on the process except to outline the general apprehensions I attempted to counter. While coding I had a constant worry that too concentrated a focus on the minutiae would obscure the running themes and reasoning processes of the theorists. To counter this I used general nodes at which to summarise the essence of a source and memos into which I pasted important points. In this way NVivo was able to complement my usual manual process by enabling me to ‘zoom’ in and out of the data as necessary.
From an analytical point of view, my tendency was to code an extract at an existing node and only create new ones when it was absolutely necessary. During the first pass of coding I began to feel certain data were being forced into categories rather than fitting more easily into new ones. When I reviewed the initial coding and created structure nodes I found a balance between these opposing techniques and saw much more descriptive patterns emerging.
Because the software displays quantitative information alongside qualitative I was conscious of the temptation to use numbers to substantiate claims. While figures were useful in highlighting areas of interest, I often noted to myself how ‘popular’ certain nodes were. To avoid drawing conclusions that weren’t necessarily accurate I often checked nodes to get a feel for how often an author may repeat the same argument or quote another author’s similar point of view. This, however, seems more a concern with my beginner-level understanding and use of the software, rather than an inherent problem of the programme, and it certainly encouraged constant reflexive engagement.
The detail with which the software allows one to sort through data cannot fail to be useful for handling large amounts of raw data, when used appropriately and without reliance on the technology to do everything. My overall conclusion on the use of NVivo though is concerned with efficiency of time. This was my first experience of formal coding and I found the process slow-going. While I am led to believe by colleagues and authors (Dey 1993; Richards 2005) that this is necessarily the case, I cannot help but conclude that for a literature review of so few sources my usual manual processes would have been more efficient. The structure of the programme was useful, particularly the memo function and ‘see also’ links, and this leads me to recommend NVivo as primarily an organisational tool, secondary to its function for analysis, when working on a literature review.
 I often found myself wishing there was the capacity to link a selection of text to another selection of text in another source, rather than linking a selection to a whole item. There may be a way around this need using nodes, but I haven’t yet discovered it.