Session 1: Innovative uses of software
Saturday 30 September 2000

Chair: Clare Tagg
Time: 11.00 - 12.30
Nunn Hall, Level 4

Clive Seale, Goldsmiths College, University of London, UK
Use of NVivo and Concordance software to analyse newspaper stories

I will describe a recent project combining use of several software packages to analyse news articles both qualitatively and quantitatively. The articles, downloaded from a commercial database of world wide news stories, concerned the reported experiences of people with cancer. Analysis involved a series of coding and retrieval operations using NVivo, supported by word concordances, which provided both quantitative estimates of word frequencies and stimulus to develop new coding categories. The flexibility of NVivo in pasting results of searches into Word, which could then be subjected to word concordances proved very useful in comparing subgroups of the articles. The project also involved the use of Word macros to organise data and to edit output, as well as limited use of SPSS to analyse attributes produced by NVivo. The result so far has been three papers currently under consideration by journals, reporting messages put across by newspapers about appropriate ways of managing the experience of cancer. Themes have included gender differences in the management of emotions and the display of character, the prevalence of religiosity, and the use made of sporting and military metaphors to describe cancer experiences. The reporting of data seeks to exploit the advantages of a combined qualitative and quantitative approach.

Silvana di Gregorio, SDG Associates, UK and USA
Using NVivo software for literature reviews

Reviewing and analysing the literature for a research topic is not often recognised as a form of qualitative analysis. However, articles and books are unstructured/semi-structured data. You analyse this 'data' by looking for common themes, contradictory arguments, conflicting opinions etc. Then, you need to organise this data to help you construct one's own argument about the state of the literature. NVivo offers many useful tools to help with this process. It does not replace bibliographic software, which you need to produce citations and generate bibliographies in different house styles. But it helps you in the analysis process and in building your argument. This use of NVivo will be demonstrated by a literature review which is in the process of being analysed and developed in NVivo.

Junko Otani, London School of Economics and Political Science, UK; United Nations University Institute of Advanced Studies, Tokyo, Japan
Older people's experiences after the Kobe Earthquake in Japan: the application of QSR NVivo in a non-English context 

This paper explores the use of QSR NVivo in my ongoing doctoral project examining the experiences of older people affected by the 1995 Great Hanshin Earthquake in Japan. My research data consists of various data types: TV videos, newspaper articles, interviews of key informants such as reporters and public officials, surveys, and field work with ethnographic approaches such as participant observation and unstructured interviews at the research sites. 

This paper focuses on the multimedia aspects of this research project. As the total length of videotapes is 67 hours, and as NVivo cannot directly code such multimedia documents as videos, but only link to them, I prepared transcripts from the videos in English so as to code them as documents.

By coding and searching for individuals' names in all documents, I was able to compare how what was said and shown differ in tone in the various kinds of data - for example when the same person appeared in TV video transcripts, newspaper articles, personal interviews and observations in my ethnographic field notes.

In the course of data handling in this research, several Document Sets were made. For example, the Media Document Set includes TV video transcripts, newspaper articles, and interview notes with a TV reporter and a newspaper journalist. This grouped the various types of data on media in one set.

Lastly, this research project will be one of the first applications of NVivo in a non-English context, and is probably the first in a Japanese context. Japanese is a language that often skips subject and object, especially in conversational use. It is noteworthy that when a sentence or phrase of conversation dialogue is cut out and separated from the paragraphs that represent its context by coding, it is more clear in English translation than in Japanese what the word or sentence means in its context. The implications of this for my data handling strategy are outlined.

Session 2: Software and the focus on coding
Saturday 30 September 2000 

Chair: Lyn Richards
Time: 11.00 - 12.30
Elvin Hall, Level 1

Sarah Delaney, National Women's Council of Ireland Millennium Project, Ireland
The impact of qualitative software on the problem of 'coding fetishism'

 This paper aims to investigate the impact that qualitative software has on the problems of 'coding fetishism' as described by Lyn Richards (1998, 1999). The paper is written from the point of view of a researcher originally trained in social anthropology, who now works in the area of commissioned research. The software referred to is QSR NUDIST4, as this is the package used by the author.

 The author argues that the problem of 'over-coding' is exacerbated by a number of factors; difficulties in the training of students in qualitative theory as well as methods; External pressures from both academic and non-academic sources; and an over-emphasis on categorisation inherent in many academic disciplines.

 It is suggested that the research community need to 'take a step back' at this point in the development and expansion of qualitative software and revisit the reasons behind the development of qualitative methodology in the first instance. The author proposes that qualitative research was developed as an antithesis to the imposition of external categories of meaning and the over-emphasis on categorisation inherent in quantitative approaches. Qualitative software aids in the coding and categorisation of software and therefore should only be used with caution and in the context of the use of a range of holistic analytical approaches.

Harriet W. Meek, Governors State University, Illinois, USA
The place of the unconscious in qualitative research

We know our research design, how data is structured, the questions we ask, methods we use and many other factors allow some meanings to emerge and obfuscates others. We speak as though these decision-making processes are logical, but an intuitive leap is often necessary. The writer maintains that unconscious mental processing is a necessary part of qualitative research.

We need breaks from coding or writing. It becomes necessary to gain perspective. We find ourselves walking the dog or (horrors!) playing solitaire. This may be essential when doing intensive work like coding but most of us think of such activities as 'wasting time'. Paradoxically, these activities are often ways of handing over the problem to our unconscious, allowing another part of our brain to operate. If we can acknowledge this is an ordinary, necessary, part of the process, perhaps we can also discover the real time-waster is the internal battle? Under topics such as "Focusing on one's own work", "Lying fallow", "Wasting time", "Stupid work" and "Getting stuck", the contributions of unconscious mental processing to qualitative research, particularly when using computerized methods, will be explored and discussed.

Anja Declercq, Leuven University, Belgium
Coding: the problem of making choices

This paper was inspired by the author's experiences and difficulties while coding data from observations and interviews. When we have gathered a large amount of data, coding can become a never-ending task. There are always new themes to be found and every sentence seems to be important in one way or another. Textbooks often tell us we should code until saturation occurs. But when the amount of data is large enough and many codes are possible, saturation may never arrive. Instead, we have to make choices as to what is really important for the research questions we want to answer and we have to choose between what we really want to know and what might be interesting, but is not essential for the time-being. It is easily forgotten that codes and coding are not a goal as such, but merely tools to think and work with. There does not appear to be such a thing as 'the one and only' set of codes. After working with a pure code-and-retrieve program, the author switched to NVivo. Its capacity as a 'code based theory builder' allows one to work in a more 'holistic' and dynamic way and to take account of the complexity of the issues at hand. While coding helps you to come close to the data, theory building forces you to regain distance.

Session 3: Mixed methods: software and integrated analysis
Saturday 30 September 2000 

Chair: Silvana di Gregorio
Time: 13.30 - 15.00
Elvin Hall, Level 1

Pat Bazeley, Research Support P/L, NSW, Australia
The evolution of a project involving structured analysis: from N3 to NVivo.

Developments in QSR software from N3 through to NVivo have impacted not only on the convenience of using the program but also on the capacity of the researcher to undertake interpretive analysis. This is particularly so for a project involving an integrated (qualitative and quantitative) analysis.

Keywords and command files used to code demographic data in N3 were replaced in N4 and later versions with the table import function, making for much simpler coding of base data. Similarly, the ability to export tables made it easier to create a numeric summary of the results of matrix searches, and much more feasible to obtain coding summaries for individual documents in table file format. Each of these facilitates integrated analysis.

The introduction of a live Node Browser in N4 made for a radical change in the approach to analysis of structured data. Analysis of issues was able to replace analysis by document, while still retaining a document structure which allowed for analysis by respondent characteristics.

NVivo brought with it gains and losses of convenience in "processing" structured data. These do not generally impact directly on the researcher's ability to interpret the data, except that the changes in the structure of matrix nodes now allows for numeric output from more finely scoped sets of data.

While interpretation of text is always the prerogative of the researcher, the design of qualitative analysis software can and does impact on the feasibility of particular kinds of analysis. Beyond that, the capacity of software to generate new forms of output encourages the researcher to investigate data in new ways, potentially adding to the richness of her interpretation.

Catherine Duggan,* I Bates,H K Åström** and J Carlsson**
*Barts & the London NHS Trust, London, UK; HSchool of Pharmacy, University of London, UK; ** Uppsala University, Sweden
Interfacing qualitative and quantitative data to explore information preferences of medical patients

 Introduction A 12-item quantitative scale, the Intrinsic Desire for Information (IDI) (Duggan and Bates, 2000), was originally extracted using factor analysis, from a larger 50- item questionnaire administered to 501 patients to explore patient's desires for medical information. The IDI scale was further validated through a novel method of interfacing quantitative and qualitative data, to develop a potentially valuable tool to predict the specific drug information needs of individual patients.

Methods Patients in three London teaching hospitals were interviewed at the bedside (n = 299) using the 12-item IDI scale and a series of open questions concerning patients' information preferences. Responses were transcribed into QSR NUD*IST software for qualitative coding and analysis.

Results The derived factor scores were dichotomised as high and low scores, around the mean, and imported into the qualitative database for analysis. The "high scorers" emphasised the amount and quantity of factual information wanted, supporting the validity of the scale. The themes from the qualitative data described the amount and perceived purpose of drug related information. Matrices were formed between these themes and the "high" and "low" scorers. Whilst the "low scorers" sometimes wanted information, they more often expressed anxieties and wanted reassurance of the drug's benefits. This contrasted with the more pro-active tendencies of the "high scorers", who did have concerns, but generally not the emotional dimension of the "low scorers".

Discussion The methodology employed in this study involved importing quantitative summative demographic data into a qualitative database and re-analysing both the quantitative and qualitative data at this interface, to validate the quantitatively derived scales. The scale "the extent of information desired" may be valuable for targeting receptive patients, or identifying those who may be refractory to drug information. This is an appropriate and innovative way to validate quantitative data and develop such scales.

Catherine Voynnet Fourboul and Sébastien Point
IAE University of Lyon III, France
Two experiences of N4 in human resource management research and related recommendations for supplementary software tools and capabilities

 Two different experiences of using N4 will be discussed: one involving interview data (35 documents, 700 nodes and 12 major categories) the other involving company reports (80 external document and 350 nodes). Suggestions for improvement of the software and linked packages are offered, in the light of particular analysis strategies desirable in the field of human resource management. [full abstract awaited]

Session 4: Improving quality in qualitative research
Saturday 30 September 2000

Chair: Ann Lewins
Time: 13.30 - 15.00
Nunn Hall, Level 4

Sylvain Bourdon, Université de Sherbrooke, Quebec, Canada
Inter-Coder Reliability Verification using QSR NUD*IST

The use of Inter-Coder Reliability Verification (ICRV) of qualitative coding can be a labour-intensive, repetitive and error-prone process. But, used in an appropriate research design and within it's own epistemological limitations, ICRV can yield a great amount of information not available otherwise.

This paper presents a series of procedures making the ICRV process both fast and error free using N4's command file functionality and a word processor. It will also introduce the notion of "fuzzy" ICRV, a more "qualitative" kind of ICRV that allows for a certain degree of difference between coders. This fault tolerance is designed as a systematic way to eliminate the minor (and meaningless) disagreements typically encountered in the comparison of unstructured text coding. The use of N4's powerful search operators make this flexible and yet systematic procedure possible and quite straightforward.

Clare Tagg, Tagg Oram Partnership, UK
Using electonic publication to enhance research quality
 

Providing access to data and analysis in qualitative research is necessary to meet quality criteria but poses difficulties for presentation. The reader needs to be given sufficient access to data and analysis without being swamped by material. Unfortunately condensed qualitative data loses much of its richness and context. This paper suggests that these problems may be overcome via electronic publication of data and analysis. A particular example will be discussed illustrating the techniques using web technology. The paper will conclude by discussing the extent to which N4 and NVivo support the electronic publication.

Alan Reid, Keith Bishop, Sue Martin and Kate Bullock, University of Bath, UK
Analysing creativity, critical thinking and independent learning: does the medium match the message?

 Our ESRC-funded research uses NVivo to explore the potential of GCSE coursework in promoting teaching and learning strategies that foster independent learning, critical thinking and creativity. Although ill-defined, these constructs are highly valued by the education community and often invoked as the basis for a high skills economy and for developing capacity for lifelong learning. For the qualitative research community, similar expectations coupled with debates about quality criteria have also led to their operationalisation in discourses on qualitative research innovation, data management and 'good practice' (Reid & Gough, 2000).

In six case study schools, a team of University and teacher researchers have appraised current practice in coursework; probing students' understanding of their learning; and examining how teaching and learning styles are influenced by the coursework assessment framework. Grounding the data collection and analysis in practice, the team have developed and used an NVivo project in a variety of ways to identify indicators and dimensions of the three constructs, and to develop case studies of organisation, approaches and outcomes of coursework. Our paper takes a reflexive turn to appraise how these constructs are developed by researchers themselves through the application of computer-assisted qualitative data analysis techniques. Recognising our diversity of expertise and experience with CAQDAS and qualitative research, we draw on Gilbert's recent analysis of practitioners' reflections on these matters (AERA, 2000) to deliberate further on their effects on the quality of qualitative research.