Qualitative research is something I was completely new to at the start of my PhD. Before the PhD I was used to collecting data in the lab, and then making graphs and running statistical analysis on it to see if I could find that illusive significant P value. Now I’ve got to grips with this new method of collecting data, I much prefer it to quantitative work. Selfishly, it gives me a license to be nosey; it encourages me to ask more and more questions instead of tracking data in the hope of finding the answer. I know a lot of the science world is quantitative, so I want to use this blog as a platform to explain a bit more about qualitative research; how I deal with it and any tips I’ve picked up along the way. An earlier post, ‘What is Qualitative Research?’ will shed light on these methods for those who aren’t sure what I’m talking about, I recommend you read that and then come back to this post.
Qualitative research is used to gain an understanding of underlying reasons, opinions, motivations and experiences. It can provide context to quantitative data, it can be used to provide insights into a problem, and it can be used to help develop ideas for future research. Qualitative methods, just like quantitative methods, are varied – there are different types of interviews, document analysis, even analysis of social media conversations from places like Twitter.
I’m using semi-structured interviews it to explore trial recruitment – semi-structured meaning that I have a rough structure of the interview in my head, the questions I might ask and the topics I want to cover, but if I get into the interview and the participant says something off topic, I can go ahead and explore that too. I’m interviewing two groups of people; people who actively recruit participants into trials, and people who design the recruitment strategies for those recruiters to implement.
I interviewed 23 participants, and interviews were an average of about an hour in length. I audio recorded all of those interviews, and paid for them to be transcribed so that I could analyse them. (Side note – if you have the budget to pay for this service, please do. If I’d had to transcribe all of my interviews I don’t think there would be a chance of me finishing this PhD!) That resulted in a huge number of words. Naturally, I left all of those words alone for a few weeks because they were a bit overwhelming and I had other projects that I could work on, but eventually I had to get started.
My first task was to familiarise myself with the data – basically get to grips with it, understand what’s going on, and what I could pick up from each interview; this involves a lot of listening to your audio (and cringing at your own accent), a lot of reading through your transcripts (again, cringing at how many times you’ve said ‘er…’ or ‘like’), and an absolute tonne of highlighting & note taking. I also used this as an opportunity to check the accuracy of my transcripts.
After that I got started with coding. Coding is the process of putting your data (words/phrases/sentences) into categories so that it’s grouped together in themes. For this I needed to have a thematic framework – an outline of the categories (called themes) that my data would slot into – the familiarisation really helped with this part. It meant that without looking at the transcripts I could already name broad themes that my data fitted into. After a bit more of a detailed look at transcripts I had a draft framework which I went over with my supervisor. Next up, applying that framework to all of the transcripts – i.e. coding the data.
What I use to code
One thing I’ve learned is that everyone does qualitative data analysis slightly differently, it’s a very personal process and you need to try a few methods before you settle on your most effective way of working. I went with NVivo. NVivo is a software package that allows you to look at your transcripts, create themes, and then literally click and drag chunks of text into those themes. I found this to be a really good way for me to work, it meant that I could see where everything was, it was all neat and compact and I couldn’t lose data. Other people might do coding by hand – printing out transcripts and writing on the page, highlighting sections, adding post it notes etc. That would have fried my head and I’d probably have lost pieces of paper/post it notes etc somewhere between uni and my flat, so NVivo was my best option.
How I got to grips with coding
Coding is not an easy thing – at first it felt ‘too simple’, and that’s because it was. I wasn’t thinking about it enough, I was just happily clicking and dragging sentences into themes and wondering why people had told me this process was going to take ages. After a few hours of doing this I knew I was missing something. I printed out the themes I’d added data to, and it was pretty clear that I just wasn’t thinking about the data enough. I gave myself a few hours away from it, and came back. Something had clicked and it was going better. Eventually I’d coded all 23 interview transcripts – this literally took weeks – and I had my data in a manageable format.
This process was super helpful because it allowed me to condense my data. In an hour long interview not every word is going to be useful, through coding I cut out all the waffle and made a number of coherent piles of data that I could then further analyse.
After coding comes the process of really understanding your themes; finding where and why different stakeholders have different views (or not), understanding how themes link together, getting to grips with how experiences differ between people based on their role/age/gender etc etc. There are so many layers to this process that you don’t think about if you’ve never analysed qualitative data before.
Keep an eye out for an upcoming blog post that’ll give you an insight into how the process of writing up by findings is going..