I have 27 interviews for a psychology research paper with 4 people involved in the project and willing to code using Thematic Analysis. Is it better to use a software programme to do that or to just do it manually ? I would be grateful for anyone else's experience with these programmes. In particular the budget is small so expense may make a difference. Finally does it take long to work out how to use them? Those involved are clinicians and may not have done research in some time.
In my opinion, manual coding of data is just too tedious and energy wasting exercise. I used to do manual coding but ever since I started using software I never looked back.
The software I'd recommend to you is NVivo. Though Atlast.ti is also good but I personally prefer using NVivo. All the best in your data analysis.
With 27 interviews, you are definitely in the range where using software will help you, especially when it comes time to search through your codes.
In terms of which software to use, all of the major programs can do thematic analysis equally well, including ATLAS.ti, Dedoose, MAXQDA, and NVivo. To make a choice, I recommend looking at their tutorial videos and choosing the one that gives you the best introduction for what you want to do.
Atlas.ti used to attract me due to the dedicated iPad app. However, I believe it hasn't been updated in years, which is why I now tend to use NVivo - which is good, but far from perfect.
I'd say have a quick go at coding a small bit of text on multiple software and see which one you find most intuitive.
Thank you very much everyone. I see that Braun and Clark are not fans of software but with a group of people coding this number of interviews the software looks like a good option. Could I ask a quick follow-up question? Do the same people need to code all the interviews or can they be divided out? I suspect the answer is it needs to be the same people, but it would be so much easier if we could share out the coding.
The only reason I can think to have over-lapping coding is if you want to report an inter-rater reliability statistic (e.g. a simple agreement rate or a Cohen's kappa, etc.) That is certainly not necessary if you are taking a constructivist stance, but if you are doing something like a content analysis, reviewers may expect it.
At any rate, even if you do want to assess reliability, you only need to have a subset of the interviews "double coded."