The Second International Conference on ‘Strategies in Qualitative Research – Issues and Results from using QSR NVivo and NUD*IST’, organised by the Institute of Education, was held on Friday the 29^th and Saturday the 30^th of September, 2000. I have volunteered to voluntarily write a brief report on the proceedings.

My overall impression of the conference was that it was a great success, bringing together a wide range of people from different disciplines and with varying levels of experience of qualitative research and the associated computer aids to qualitative data analysis.

I flew to London with some apprehension for a number of reasons:

1. I was flying with a very cheap airline that is notorious for selling off your seat two hours before departure.
2. I was giving a paper with two other people I had never met in the flesh.
3. I was giving a paper to a bunch of people I had never met in the flesh.

The first concern was justified in that I only just managed to save my seat on the plane (the journey home was even more surreal). However on the second two, I need not have worried at all. From the moment of my arrival the atmosphere in the conference was friendly, open, interested and interesting.

This welcoming ambience continued right throughout the conference and made for a refreshing change. It was enhanced by good wine and lovely food, leading to great conversation and information sharing. In fact, in a post-conference meeting held by four delegates (those protagonists can ensure my silence by sending me lots of money) in the pub afterwards, everyone commented on the lack of elitism and competitiveness which are all too often the hallmarks of conferences like this. On the contrary, everyone seemed to be open and honest about their difficulties, willing to share suggestions or take advice, and above all, had an enthusiastic and enquiring attitude to proceedings. I personally was delighted to meet such a nice group of people! It certainly made up for the fact that my flight home was delayed by an hour because, and I quote, ‘the captain phoned in sick’. This is true.

Another refreshing aspect of the conference was the fact that Lyn and Tom Richards of QSR International really engaged with the questions and concerns voiced about QSR software, and their responses were honest and professional. Risking a plug, the QSR sponsored dinner was free and delicious, and there was lots of it. Great idea.

The conference had a truly international flavour to it. Although the majority of delegates were from the UK, (perfectly understandable seeing as the conference was located in London!), people also attended from the United States, Canada, Australia, Portugal, Israel, Spain, Belgium, Sweden, France, Norway, Switzerland, Japan, and last but definitely not least, one particularly interesting and attractive delegate from the Republic of Ireland. Let’s hope that this trend is encouraged and continues into the future.

PLENARY SESSION: PROGRAMMING PROGRESS.

Although the session was entitled ‘Programming Progress’, the focus of debate centred on methodological issues in qualitative data analysis generally. This proved to be a fascinating discussion, and introduced a ‘thread’ which ran throughout the rest of the conference.

Clive discussed the problems associated with researchers claiming to used a ‘grounded theory’ approach, when in fact they are doing something very different. He referred to some research he had done into a number of articles that were supposed to be using grounded theory. In fact only a very small minority of these had indeed used a grounded theory approach. The majority had employed a technique that Lyn Richards has previously referred to as ‘pattern analysis’ – a concept she explained in more detail in the parallel session held on Saturday. He emphasised the importance of raising the methodological sophistication of researchers to complement the capacities of QDA software.

Pat raised the question of whether technology has replaced traditional ethnographic skills, and Sylvain echoed Clive’s concern about standards in qualitative research at the moment, and felt strongly that qualitative research must be grounded in it’s historical and philosophical context if full understanding is to take place, and before researchers turn to qualitative software to aid them. He went on to argue that the problems associated with QDA software also rests with the training of people to use the software, pointing out that ‘making sense of data can never be handled by computer’ alone.

Debate after the speakers was interesting and challenging. Many people continued the methodological issues raised and the debate centred on these problems and the effect of qualitative software on them. The general consensus was that the software itself was not the problem but it had exposed a pre-existing weakness within qualitative methodology, which needed to be addressed.

PLENARY SESSION: EFFECTS OF SOFTWARE ON ANALYSIS STRATEGIES.

Tom Richards re-affirmed a need he felt for research training and practice to be made more explicit, accountable and grounded in the philosophical context, and for the methods used and represented in publications to be made more explicit. Software cannot function to replace good methodological approaches to research and analysis. In this context, he felt it essential that CAQDAS must be grounded in it’s own historical narrative.

Overall, what emerged was the continuity of three things that were always central goals through all the versions of NUD*IST - the symmetry of documents and nodes, and the insistence that nodes are things, to be played with, ordered, reviewed, and browsed live; the emphasis on searching that is cumulative –i.e. the results of searches are more data - as well as restricting searches to particular data; and the goal of growing the software in response to researcher demand and research change.

PARALLEL SESSIONS:

The 'innovative uses of software' session contained three papers which illustrated the variety of research and data types where NVivo is used. Clive Seale described the analysis of newspaper articles, Silvana di Gregorio the use of NVivo in literature review and Junko Otani the experience of older people following the Kobe earthquake in Japan. Each paper was noted for the useful detail about how NVivo was used. Clive illustrated how useful Word macros are for preparing data and concordance software for highlighting less obvious themes. Silvana showed how literature reviews can be developed using proxy documents and memos in NVivo. Junko discussed her decision to translate her interviews from Japanese into English and the consequences of that decision.

For Lyn a highlight was the session that qual-software members were waiting for - on coding and how to control it. Sarah, who started it all with the famous "my head hurts" message, placed the pressures to over-code in a political context. She showed a disturbing fit between the contribution of qualitative software to the problem of ‘coding fetishism’ and external pressures from both academic and non-academic sources; that contributed to an over-emphasis on categorisation. Pressure to show rigor & thoroughness can, she argued, dominate thinking about relationships with data. Harriet Meek, from Illinois, combined roles of researcher and therapist to discuss the place of the unconscious in qualitative research, relating the problems of overcoding to other ways of "getting stuck", and providing a range of solutions. Anja Declercq from Belgium gave a dramatic account of cutting through grip of clerical duty, reducing excessive coding to get on with analysis. She vividly illustrated her story of the late stages of a project with before and after accounts of her interpretation of a passage: we shared the excitement of "seeing" what mattered when coding was not an end in itself but a means to emerging ideas.

Pat Bazeley looked at the effects of the development of the software had on a particular project. The project was started in N3, moved to N4 and finally to NVIVO. When the project was started in N3 she discovered that it was easier to work with fewer codes. She reluctantly weeded down the number of codes developed from 40 to 12. In order to discern patterns in the data she had to manually construct matrix tables from the information N3 gave her. This was a tedious process.

N3 to N4: The first impact the development in the software had was that it was much easier to code and then, from the initial coding, to 'code-on'. This deepened the interpretation possible. She immediately decided to re-code the project and develop a more sensitive coding framework. The software also automatically produced the matrix tables she previously could only create manually. There was also a very simple interface with SPSS allowing matrices to be exported for statistical analysis. She described an iterative process in how the qualitative analysis informed a statistical analysis (in particular correspondence analysis) which then further informed what to look for in the qualitative analysis.

N4 to NVIVO: NVIVO did not have any further impact on the interpretation of the data. However, it was able to produce better tables. In particular, it was possible to do a global matrix (look at everything by everything) in order to do a cluster analysis in SPSS. Her mix of moving between qual and quant analysis informed the next steps in the analysis process.

Lyn Richards [Lyn offered a paper when another presenter had to withdraw at the last minute. This paper was entitled: Two Types of Analysis? The Perils of Pattern Analysis]

Lyn decided to pick up Clive Seale's comment at the Opening Plenary that in his survey of people using computers in qualitative analysis, many people said that they were doing grounded theory but very few actually were. Lyn pointed out that people need a label to describe the kind of qualitative analysis they are using and the term 'grounded theory' has been used as it was a label people recognised. There is not a label for the kind of research common in policy studies - which look for patterns among different sub-populations. Lyn has coined the term 'pattern analysis' to describe this kind of research. She argues that it is a perfectly legitimate form of qualitative analysis-in keeping with the objectives of that kind of research. The problem occurs when people try to dress up that kind of research into something which is not - by using the term 'grounded theory'.

Catherine and Sébastien reported on the use of N4 in two different Human Resource projects in France. They commented that in France there is a cultural emphasis on automating the analysis process. So they had difficulty in persuading their colleagues of the value of N4. They focussed their talk on the links between N4 and other software. In one study, with Decision Explorer and in the other study, with Excel/SPSS. In one study, they were frustrated at N4's inability to do a global matrix which could be the subject of a cluster analysis. They devised a way to do this via Excel (However, it was pointed out that NVIVO can do global matrices.) In the other study, they had to manually calculate the strength of the relationship between links between codes and display them in Decision Explorer. The following discussion focussed on the difference in strategy between Pat's use of mixing quant and qual methods and Catherine and Sébastien's use of mixing methods. Pat focussed on the qual analysis first to inform what quant analysis she need to do. This she fed back into further exploration in the qual data. Catherine and Sébastien's strategy was to use cluster analysis first to direct what they should look at in the qual analysis.

Overall, the session could have been better described as ‘New Things’. Both papers were quite narrowly focussed but each in their own way groundbreaking

Clare’s paper discussed her recently submitted CD multimedia thesis. On the whole she used her own software and codes to increase the reader’s contact, for example between her own multimedia or other source data and her conclusions. She explained the way she structured her thesis in this respect and suggested that although NVivo could be the software structure behind such a presentation of results, it had some way to go before a project could easily be read and navigated by the inexpert (in NVivo) reader.

Sylvain presented a new procedure for checking coder reliability in the team environment whilst using Nudist. It was based on the use of a combination of customised command files, run on identical, closed coding schema/index systems (or parts of) and their application to the same data files by different coders. Simple statistical analysis after running the command files allowed the similarities/differences between coders to be rated. It was suggested that QSR integrate the procedure as a formalised tool within Nudist software.

CLOSING PLENARY: WHERE NEXT?

Lyn Richards opened by saying that she had really enjoyed the conference and had learned a lot from all the delegates. Tom also said that he had learned a lot – in fact he was going away with a list of ‘things to do’!

Those who chaired the parallel sessions summarised the proceedings. Clare Tagg noted that not much reference had been made to projects using multi-media research techniques. It was suggested that this should be encouraged for the next conference. She also said that there were plans to publish the papers presented at the conference, on a CD-Rom version and in the International Journal of Social Research Methodology – Theory and Practice (as a special issue) and invited speakers to submit to one or the other.

Silvana di Gregorio commented that cultural and language differences could contribute to the further development of software and, also, in qualitative analysis techniques. In particular, Junko Otani's study showed that the structure of Japanese meant that coding had to proceed differently than in English to be meaningful. She pointed out the difference between high context specific and low context specific languages. This could have implications for the tools that can be used to analyse such languages. And these new tools could also raise issues for English language researchers and challenge assumptions about coding practices.

I would like to thank the organisers of the conference (Colm Crowley, Judy Ireson and all at the Conference Office) for a really refreshing and enjoyable experience. I’d also like to thank Lyn Richards for encouraging Harriet, Anja and myself to present papers at the conference. It was a great experience. Last but not least, thanks to Ann Lewins for running the qual-software list which has brought so many researchers together in email if not in flesh, and has certainly helped and encouraged me since I first subscribed.