Qualitative resources

Here is a a whole list of resources for a research student to use to get to grips with methodological approaches. The list includes a link to a video of Alan Bryman talking about conducting mixed methods research; it has helped me get a better idea of how people use the word 'paradigm' in the context of methodology.

I've now made the arguments that I want to make in my methodology chapter, though I daresay my supervisors will have plenty of critical comments to make. They can only be helpful to sorting out my confusion of terms. I shall submit a full draft of the thesis to them in a few hours, and then must wait three weeks to get their feedback.

But hey! a full draft!

Saving time with Nvivo

Qualitative data analysis software can save time (Silverman) is my answer to Minh's assertion here that Word is so good. He writes some provocative blogs, worth reading.

I use NVivo 8 to store, search and retrieve interview text from 30 transcripts. After transcribing I code it against:

1. general codes - process, structure, climate, facts, queries (see Checkland). For a query, I create linked annotations as to what the query is about and how I might follow it up.
2. Metaphors, stories and anecdotes or words that strike me. For example, in one case study, participants used the word 'gel'. IN another type of organisation, people, more than once referred to "skin in the game". The shard use of the terms may indicate the culture.
3. the theoretical framework for engagement, covering aspects of communication and knowledge-ability: mutuality, negotiation of meaning, environment, expertise, contribution, participation and adaptation.

Having coded, I use the NVivo query function. I'll search for a code to draw together all the text that I'd coded under facets of engagement. For a particular case study, I query all the relevant transcripts and documents and emails, copy and pate them into a Word document and then draw on them as reminders and examples as I write the analysis.

Another query I use is text query - I've heard a particular phrase - "Who said it?" I ask myself and search all the files to find it.

I use memos to create:

• a to do list
• a coding journal
• thoughts on each case
• thoughts one each code
• key issues that arise

To develop theoretical aspects, I link from a source file to one of the memos, though I'm not very sure yet of the technicalities. I think cross linking will become more important as I move into cross case analysis. Then Nvivo should be of help in developing the theoretical aspects of my study.

Matrix queries then will be the next facility to experiment with to see what patterns I can find with certain attributes.

---

CHECKLAND, P. (1981) Systems thinking, systems practice.
SILVERMAN, D. (2005) Doing qualitative research : A practical handbook, London ; Thousand Oaks, Calif., Sage Publications.

Representative qualitative research

I was thinking about how to judge the goodness of my qualitative research - see chapter 10 of Miles and Huberman.

Are my informants representative? They are to some extent self-selected in that their organisation's management accepted my approach asking to research them. The management then suggested which IT project or programme to study. In that sense, the informants didn't volunteer, though the organisation did. So informants and organisations may not be representative.

But representative of what? They are representative case studies of successful IT projects, not failures. That's clear because:

1. the organisations made the cases available - I'd assume they'd not make failures so available
2. several of the cases have been submitted for awards
3. the pride is in the words and terms that the informants use

Informants tended to be articulate and elite in management, but outliers were available for interviews at very short notice, such as a clerk, a techie chap and a data input person. So does that mean I've got something representative - both management elite and operational grade informants?

---

Miles & Huberman in Miles, M. B. and A. M. Huberman (1994). Qualitative Data Analysis : An Expanded Sourcebook. Thousand Oaks, Sage Publications.

Qualitative data analysis

I was talking with a fellow OUBS post grad about qualitative data analysis (QDAS). As she hadn't been using Nvivo, but Decision Explorer, she wanted to know if Nvivo would be more useful to her. We talked about its linking mechanisms, annotations and memos - those are what make NVivo really useful for me.

I demonstrated some of the ways I now work after doing that Silvana di Gregorio workshop.

• time stamp (Cntl shift T) - useful for journal
  
• memo for a coding journal
• right hand view instead of bottom view (view>detail view> right)
  
• Alt-F1 to remove / show left hand column
• creating and reading annotations
  

We received a manual at the workshop, which is very useful reference. Do the course to get the manual too. Silvana di Gregorio also writes books on QDAS as well as giving workshops.

The linking mechanisms are what make NVivo really useful for me. I annotate as I transcribe, or listen the first time to a transcription, or whenever. I use memos to link to collections of notes on facets of engagement, so I'm linking what I find in the cases with what I've found in the literature. This will help my later analysis as I write up the case studies for the dissertation.

Validating qualitative data

A senior PG colleague is completing his analysis. He's coded his work in Nvivo, but needs to validate his model. He printed and cut out loads of words from his texts and asked others to categorise them physically and manually. Now his wall is covered categories and codes.

He is pleased to find that others' categorisations match his work - that validates what he's done.

Analysis paralysis

Paralysis is what happens when you come to qualitative analysis. It's the weakest part of what they teach you - how to analyse qualitative data. As I work my way through with

• coloured pens,
  
• tables from Miles & Huberman,
  
• a data accounting sheet and
  
• diagrams to compare case study relationships with what I wrote about Peled's findings,
  

I think it all feels as qualitative an approach to analysis as the data that I'm analysing. And the analysis is only another layer on top of what is already there - it's my view of the interviewees' views.

I think I learned a more about analysis of these sort of soft systems when I was tutoring systems (T301) years ago. Checkland's approach included identifying climate, process, facts, structure and queries, but it was the climate and process that revealed the issues. Even Miles and Huberman have picked up on those two.

So I have a framework that gave me my interview questions from which I've elicited data, and the main thing now is to explain how I've found these themes, or found any findings. I have to make the links back and that's the hardest bit - writing coherent and logical links. Perhaps it's my writing that is paralysed.

---

Miles, M. B. and Huberman, A. M. (1994) Qualitative data analysis : an expanded sourcebook, (2nd Edn), Sage Publications, Thousand Oaks. 758
Peled, D. A. (2001) 'Outsourcing and Political Power: Bureaucrats, Consultants, Vendors and Public Information Technology', Public Personnel Management, 30 (4), pp. 495. 475

Qualitative analysis

I've just realised that I could use NVivo, could have been using NVivo, to extract and analyse data from my literature review.

Why did no-one point that out before?

• Because supervisors don't need to know much about software and aren't going to point their students at it.
• Because there isn't training in using software, but training that is either about the product (e.g. Word, NVivo, Endnote) or about the process, and not how to integrate the two.

The best integration the OU has provided in the last two years was through the Professional Academic Communication in English sessions (PACE) when one of the lecturers explained how she'd used Endnote to keep notes and print them off or insert them in Word.

Here are slides that discuss the training issue, from the 2006 ESRC research methods festival. And here's a paper from Silvana di Gregorio on using NVivo to analyse literature for the review. And here's a discussion on qualitative data analysis.

Data analysis & using NVivo

Third year student, Linda Wilks, presented on the role of habitus in festivals. That was interesting but of greater relevance and use to us students was her explanation of how she analysed her qualitative data. Here's what:

Recording the interviews

• Assess background noise
• Put your recorder on something soft

One of my colleagues had problems transcribing an important word when the interviewee spoke as he put his coffee mug down.

Transcription

• Use a foot pedal
  
• Contract some of it out but expect to have to check it thoroughly yourself too. Cost is £11 per hour of their time.
• Takes approx 7 hours to do 45 minutes

I thought this useful & practical. But it got more useful because she talked about organising the data to make it manageable by:

1. Designing a data classification system
• Based on research theory
• Related to interview questions
• Be prepared to modify it
• Transfer into NVivo under ‘tree nodes’
  


2. Import your transcripts into NVivo
• Highlight sections of the transcript and drag into the appropriate tree node


3. Export the tree node contents
• Make sure you tell NVivo to include the folder and hierarchical name
  

She gave us a screen dump of NVivo with nodes and folders and hierarchies. She explained with examples of sources and numbers of references in the nodes, and an analysed transcript.

• To get an idea of the important areas, assess how many references a node has.
• Print out tree nodes for analysis.

For the second level of analysis, she used Fairclough's critical discourse analysis and justified the use in relation to her work.

She had a check list to scribble on her texts, for example what styles, types of statement, intertextuality and transferred the notes to a spreadsheet.

The final step is to write up findings, without reference to theory at this stage and no language that precludes conclusions.

What I've noted above is all Linda's work, and if you want to know the detail you should contact her. I'm going to get NVivo and practise on my MRes data, and I'm going to look up Fairclough.