I'm using NVivo for this.
I have a few questions and would like to read the opinions of people that are more experienced than me:
- Do you code everything, or skip coding information that is not so relevant?
- I think the wisest would be to code interview by interview instead of ttheme by theme (the interview is semi-structured which means that each topic is sometimes spread out throughout the whole interview). Am I wrong?
- I think I'll have codes that fit several themes (or maybe they'll just be slightly different...). Should I repeat them under each theme?
- I have read through all the interviews carefully and I attempted to "build" a node tree in NVivo to be aware beforehand of the themes emerging from the interviews and the problems (like too much overlapping etc). Was this a good strategy?
Thank you very very much.
- It is not entirely clear to me whether you are using codes derived from the theory in your field, or themes emerging from the data. In the former case, you would do well to code strategically: i.e., code those parts of the interviews that inform your research questions, or those that relevant to your coding scheme. If your coding scheme emerged from the data, then one might argue that there are no parts of the interviews that are 'not so relevant'. Everything needs to be coded, and compared to other data, evaluated, etc. It depends on your research design.
- How you choose to code is entirely up to you. There are advantages either way.
- Use all the codes that apply to each theme.
If you are using a grounded theory approach, then your themes are emerging from the interviews and everything should be coded. Then you would use your emergent themes to aid in the following interviews. You would keep this up until you start seeing fewer new themes and more repeats of existing themes.
- Badges
- Science topic
- Similar topics
- Filing