Handle with Care

I haven’t blogged for a while as I have been immersed in the process of analysing my qualitative data, and trying into pull this together into something that might just be clever enough to call “findings”. However, I thought the process of dealing with qualitative data would be a useful topic to blog about.

I have quite a large data set, 25 interview transcripts in all, each about one hour in length so the logistics of handling such a data set required some consideration. The first decision to make was around transcription of the data itself. Now, many authors suggest that transcribing interview data yourself help “immerse” oneself in the data. I did do this for my masters research which only had seven interviews but studying for a doctorate part-time whilst working full-time and having a young family meant that for me, this was not an option. You see, I was a bit is a tomboy at school, preferring to spend my time drawing side elevations of houses to scale and making a garden trowel in the technology department rather than learning to type with all my girlie peers who aspired to be secretaries. Oh, how I now lament my lack of typing skills. I know that to type out 25 hours (plus) worth of interview data would have taken up an enormous amount of my time and that I would only be able to concentrate on accuracy of the data rather than the content. So I chose to pay a transcription company to do the transcription for me. Having someone do the transcription does not negate listening to the audio along with the files. This still meant I have to listen to each transcript again, going through the transcript making changes for accuracy, but I found it was much more time efficient and a  more productive way of immersing myself in my data.

Once I had the transcripts, I had to analyse the data. Here came another decision point, and one which has raised quite a bit of discussion during my supervision sessions. Should I use a computer assisted analysis product such as NVivo, or process my data manually. As someone who teaches technology enhanced learning, you might assume this is a no brainer; NVivo all the way! I did use NVivo but didn’t like working with my data in this. I found I was jumping ahead and not quite getting the depth of analysis I felt I wanted or needed and felt I could not easily visualise my data. So, instead I reverted to MS word and developed my own system. I even used highlighter pens, paper and flip charts!!!

I am challenged by my supervisors to justify my position in choosing not to use NVivo, yet try as I might, I have found no PhD regulation that states I must show that I have used it, equally in the theses I have read, unless the author specifically states they have used it I would not know if they had used crayons or not. It was good to be challenged in this way though, as it did make me think through exactly what the issues were and why I chose to do something in a certain way. I am happy working with my system and feel confident I can defend my decisions around my analysis yet I do feel as though I have failed slightly in not using NVivo.

My data is organised into theme tables for each participant, detailing the invivo codes from each, interim themes then master themes. They are then brought together in superordinate themes that provide overarching themes for the entire data set. It is detailed, perhaps a little laborious but, for me it works. I consider the analytical process more of a priority and I have used interpretative phenomenological analysis as the approach. The authors of this approach encourage a more manual approach to analysis like the process I have followed, so I am reasonably happy I have been true to the theory.

I would be interested to hear from readers of this blog what challenges they have faced when working with qualitative data, and if they have come across any requirement to use computer aided analysis. Equally, I would be keen to hear of those who have used it successfully.


8 thoughts on “Handle with Care

  1. Thanks for sharing this. I am currently recording interviews, will end up with about 16 hours of talk, then have to transcribe them (myself, probably, as I am afraid of the costs of having it transcribed) and then it’s off to MAXQDA for open coding and content analysis. Next round of interviews then in March 2013.
    It’s encouraging to read about your NVivo abstinence – if MAXQDA turns out too time-consuming I won’t bother either. I hope you’ll get some feedback from people who have been there as well.

    • Thanks for your comment Daniel. The costs weren’t as bad as I thought. About £1 per minute so they varied from between £50-70 each and as I didn’t do them all at once were reasonably affordable. I am working full time though and not on a studentship. Maybe getting some done and doing some yourself would help reduce cost and keep up efficiency.

  2. Thank you for sharing this, Karen! I’m in a similar position in that I have around 30 interviews between 30-60 minutes. Like you, I chose not to use NVivo, but I was also never encouraged to use it (no-one else in the office uses it). Instead I made notes at the side of the transcript and created mind-maps. It probably means that my data is less “neat” than others, but it makes more sense to me.

  3. Hi Karen, I enjoyed reading your article, I’m interested that your say 25*1 hour interviews is a lot, as I’ve repeatedly had supervisors/panel staff suggest that my 18 interviews are insufficient but I feel that I do not have time to analyse any more. Have you found any useful material to cite relating to the volume of data collected?
    I’ve found the analysis stage to be a sticking point, I had no desire to learn a new technology (and no one has pressed me to do so), but didn’t really know where to start, so I have been ‘thinking’ about it for too long now. I am looking largely for anecdotal evidence from my interviews to support a separate textual (media) analysis so I have ended up not fully transcribing my interviews, just extracting the relevant information from them (I started transcribing but only got 4 done in full). I’ve been writing up notes and transcribing quotable sections into an excel spreadsheet which allows me to gather information by theme.
    I’m studying part-time and looking after 3 little people the rest of the time, I’m in awe that you manage to fit full-time work in there as well.
    Good luck with your studies.

    • Hi Esther

      Thank you for reading and commenting on my blog. To answer your query about number sof interviews, I have found that my data set is more than adequate so far. I recruited 10 people newly diagnosed with MS and invited them to nominate a ‘support person’, most commonly, this was a spouse or partner. Some were interviewed together and some apart and I interviewed them twice, over a period of between 6mths to a year apart. What I have found is that in relation to my research questions there is little new in the follow up interviews, they are confirmatory with occasional new insights but I think I would have equally addressed the research issue with the first data set. In my opinion, qualitative research is not about numbers or interviews but about depth of analysis. I have used interpretative phenomenological analysis which is a very in depth idiographic approach to analysis. I have read other phd thesis with lesser numbers than I have used but like you I was guided to get a decent number for a doctorate level study. I would advise you to have a look at other thesis using similar approaches to yourself, within your discipline. I got most of my examples from the institutional research repository but there is a uk database of theses called EThOS.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s