Hello,
I have 78 semi structured interview transcripts. There were three of us who had conducted the interviews and the questions are not always framed in a similar fashion. They are asked mostly in the same order but sometimes in a different order. The probes differed as well.
I wish to conduct qualitative analysis on Nvivo.
- I started by manually going through each transcript and creating nodes along the way. Is this the way to go or are there better first steps.
- I created an excel sheet with participant demographics information and other specific details such as office building number, years of experience, number of breaks etc and then imported them into case classifications. The reason was to see if there was a relationship between say sentiment of people of a particular gender or building OR a relationship between a node regarding sunlight and age.
But now the cases are only respondent ids. And I am not able to select gender or company name or anything else while trying to create a relationship.
- I am not able to understand how the word frequency can be used since the words in the questions also get counted. (The transcript is all 'transcript body' in word)
- How do I do thematic analysis ? Can I use 'autocode> themes' for the nodes I created. I am not sure that makes sense though.
Your responses are much appreciated.
Going through the data and creating codes amounts to an "inductive" approach, which is just fine. The other alternative is to have a prior set of "deductive" codes that come from the literature. If you continue with inductive coding, you might want to consider a systematic approach to your analysis, such as Braun & Clarke (2006) Thematic Analysis.
In NVivo, the term for your spreadsheets is "attribute data," which can be matched to each participant. Here are a couple of links on the subject from NVivo's help discussions:
http://help-nv11.qsrinternational.com/desktop/concepts/organize_your_demographic_data.htm
http://help-nv10.qsrinternational.com/desktop/procedures/work_with_attribute_values.htm
Dear Vinita,
I agree with David that it is a good idea to start by reading and annotating your transcripts then creating a categorisation. The lack of consistency in the questions and answer order or formats is not a particular disadvantage in inductive research as you are looking beyond the language used in the questions to deeper constructs across answer categories.
Once you have analysed a sufficient amount of transcripts your categorisation should stabilise (a process known as saturation). You may also wish to construct a theory and a model of the relationships between your categories. At this stage I would recommend considering some automated techniques to confirm and generalise your interpretation to your whole data set. NVivo can help with this.
Finally, if some of your answers are short, you may wish to consider a quantitative technique such as synonym frequency analysis. Again, NVivo can help with this process.
Dear Vinita,
You don't say what you're hoping to achieve with your analysis - in part, the answer to your question will depend on what you want to get from the data you and your colleagues have collected.
The links in David's answer will show you how to ascribe demographic characteristics to the transcripts which will enable you to compare those in one group with another.
However, what I really wanted to caution against is rushing into automated forms of 'analysis' - there really is no substitute for knowing your data very well and that will come as you read and re-read the transcripts themselves. You should, in my view, be looking to manually code and re-code expecting to go down some blind alleys before you reach analysis that you are happy with. Of course, the other complication is that you are working with other people so you will need to also think about moderating your coding work (comparing your coding with that of others).
Good luck!
Dear Vinita,
Your questions have already given you what you may be expecting from the data. Hence, arranging and coding the data as expected will give you a good analysis. @Peter and will have said it all. Good luck.
- Badges
- Science method
- Similar topics
- Methodology