#3 Useful nuggets that I learnt from 'Analysing interviews & focus group data using NVivo' training
- jennyrouth
- Mar 23, 2021
- 3 min read
...delivered online by Dr Christina Silver and Dr Sarah Bulloch
What is a code?
“… a word or short phrase that symbolically assigns a summative, salient, essence-capturing, and/or evocative attribute for a portion of language based or visual data” – Saldana, (2016) “The Coding Manual for Qualitative Researchers”
^^^ This is important to keep in mind with our node/code labelling
Remember coding is iterative! – a process, not an event
Codes can be thematic, but also structural (e.g. coding all annotated references), which can help with analysis.
Remember to describe your codes! – it will help you to remember what you mean, but is also important for the codebook.
Organising data
An important question to answer: what is my unit of analysis?
For me, because the participants are anonymised, and their talk is embedded within a group discussion, the unit of analysis is the group interview.
- We use codes to attach ideas to parts of the unit of analysis
- We use attributes to attach ideas to the whole of a unit of analysis (for me, this would be the demographic groups)
Do I need to use attributes in my project to capture information I have? – yes – the demographic groups (each group is homogenous in terms of background)
If I use attributes, will I attach them to cases or files? – this is to do with the unit of analysis…
- What is the unit we want to classify? – each group interview
- Is that unit equivalent to an NVivo file? – yes (each file = one transcript of one group interview)
- Then we use file classifications (remember – attribute = name of the variable (demographic group), attribute value = students, faculty, supervisors etc)
BUT ALSO
- What is the unit we want to classify? – interviewer vs participant
- Is that unit equivalent to an NVivo file? – no, interviewer and participants in each file
- Then we use case classifications
A case = a code with all the data from one person that you can apply attribute values to.
I can grab speaker identifiers – e.g. INT: me as the interviewer – and autocode them as cases. I can then exclude these from queries, if I want to.
As participants are anonymised in the transcripts, identifying them is less useful. However, if they weren’t, I could attach attribute-values to them (e.g demographic information). I could query data by those e.g. bring up only student cases, and use a crosstab query to find data coded to a specific code, by a type of case.
Precoding activities (akin to stage one [familiarisation] of Braun and Clarke’s six stages of thematic analysis)
Low and slow – annotations
Select the text in the transcript --> right click --> new annotation (turns blue) --> write an annotation – an idea for a code
- This reference will be listed in ‘annotations’
- Code this annotation to the node ‘annotations’ – this way they can all be reviewed (reference AND annotation) at the end to form the codebook v1.0
Word frequency queries
Select query --> right click in blank space --> word frequency
Remember: selecting “files and externals” include memos – might be better to select transcript files only
--> Save search
--> Run search
--> Remove uninteresting words (go into the stop word list – accessed under “project properties” from “file”)
--> Select Word Cloud
Remember: after word searching you can increase or decrease the ‘coding context’ by Cmd+A to select all data results, then right lick and selecting e.g. Broad, which gives you the whole paragraph.
Action points:
- Assign file classifications: demographic groups
- Assign case classifications: interviewer (INT) vs participant (RES1, RES2, RES3, RES4)
- Word frequency query and produce a word cloud
- Once finished listening and check transcripts, start annotations
Comments