Week 8 - Thematic Analysis Flashcards
What is thematic analysis?
According to Braun & Clarke, 2006:
- One of simplest methods of data analysis
- Flexible -> compatible with any research paradigm
- Provides a rich and detailed, yet complex account of data
- Its a method for identifying, analysing and reporting patterns (themes) within data. It minimally organises and describes your data set in (rich) detail
Inductive vs Theoretical (deductive) TA
Inductive (data driven)
- ‘Bottom up’ approach
- Themes strongly linked to data themselves
- Don’t necessarily link directly to interview schedule themes
- Driven less by researcher’s interests or prior reading of topic
- Process of coding the data without trying to fit it into a pre-existing coding frame (research question can evolve through coding)
Theoretical
- ‘Top down’ approach
- Driven by researcher’s theoretical or analytic interests in area + prior reading of an identified gap in literature
- May link more closely to interview schedule focus
- Rich description of data and more detailed analysis of some aspect of the data
Choice between inductive or theoretical is based on how and why your coding the data
-> can either code for a specific research question which maps onto more theoretical approach
-> or this specific research question can evolve through the coding process and is data driven which maps on the inductive approach
What are the two types of analysis?
Choices:
1) A a rich thematic description of entire data set, so the reader gets a sense of the predominant or important themes (complexity is lost but rich description is achieved)
-> themes you identify, code + analyse would need to be an accurate reflection of thee content of the entire data set
2) A more detailed and nuanced account of one particular theme or group of themes within the data
What are the 6 stages of analytical procedure?
Reflexive TA (Braun & Clarke, 2022)
1) Familiarisation of dataset
2) Coding
3) Generating initial themes
4) Developing and reviewing themes
5) Refining, defining and naming themes
6) Writing up
Phase 1: Familiarisation
-> involves a deep, intimate knowledge of your dataset - immersion (will be reading + reading and listening repeatedly to your data)
- Critically engaging with data
Questions:
- How does the person make sense of whatever they are discussing?
- Need to be acutely aware of your own input into your analysis
Need to ask yourself questions: - Why might i be reacting to the data in this way? (do i have any previous experiences which impacts my understanding?)
- Note ideas around data
- Hand scribble on hard printed copy
- Voice recognition software to comment
- Additional document for each interview, research diary
- Brief systematic overall familiarisation of whole dataset
- May spot potential patterns in this stage
Phase 2: Coding
- Coding forms the building blocks for your analysis
- Codes capture specific and particular meanings within dataset of relevance to your research
- Succinct labels that evoke data content
- Code label = summary of analytic idea
- Codes can be summative or descriptive to more conceptual
- Coding is a systematic process
- Process involves reading each data item closely, line by line, section by section tagging all segments of the data where you notice any meaning that is potentially relevant to your research question and give it a code label
- Some segments might be tagged to different codes - a number of different meaning are evident
- Insight, rigour - avoid cherry picking + skipping parts of transcripts
- Codes should connect to more than one segment of data - idea is to capture repetition of meaning
- However, a code can be useful even if it only occurs once as themes can be developed from multiple codes
- Subjective process as shaped by what we bring to it
- Its a process of interpretation + meaning-making
- One coder is normal practice, multiple coders can be useful to gain richness but this isn’t essential
Inductive vs deductive data coding
Inductive (data driven)
- Dataset as starting point
- We always shape what we notice about the data (Fine, 1992)
Deductive (researcher or theory driven)
- Dataset provides a foundation for coding and theme development but reflect theoretical or conceptual ideas the researcher seeks to understand through dataset
- Existing theories or concepts provide a lens through which the researcher makes sense of the data
An analysis could have a blend of both inductive and deductive coding
Difference between semantic coding and latent coding
Semantic:
- Ppt driven
- Exploring meaning at surface level of data
- Semantic codes capture explicitly expressed meaning (ppt expressions)
Latent:
- Researcher driven
- Deeper, more implicit or conceptual driven
- Sometimes quite abstract from obvious content
Initial coding is often semantic or descriptive
-> later on researcher can add their own interpretation and add depth - codes become more latent
How do you code?
Code: Meaningful piece of transcript/data
- Need to code data line by line, section by section
- Each time you spot something meaningful/interesting/relevant, tag it with a code
- Consider if an existing code applies, or if you need to develop a new code
What are the technologies of coding?
-> these are personal choice
- Some prefer to work with printed transcripts, others prefer coding in digital form
- You can handwrite code labels on printed transcripts (wide margins on transcripts)
- Sticky notes on printed transcripts
- Each code can have its own document, copy and paste pieces of transcript (quotes) into appropriate code documents - make sure to label quotes properly
- Typing codes onto electronic version (using a comment box)
- Software (computer assisted qualitative data analysis software)
What is a theme?
A theme captures shared meaning, united by a central organising concept
Central organising concept = expression of shared, or similar, ideas or meaning across different contexts (semantic or latent level)
- Be aware that some codes might on the surface appear quite dissimilar, or contradict but might form basis for there
What is one problem when developing themes?
Be careful of mistaking topic summary for a theme
- Topic summary is a summary of everything ppts said about a particular topic presented as a theme
- Often happens when you follow answers to specific questions on your interview schedule
- They unite a topic rather than a shared meaning or idea (e.g. reasons people had fertility treatment = topic summary; fertility treatment as deviation from socially accepted timeline of parenthood = theme)
Phase 3: Generating initial themes
- Themes need to be developed, reviewed and refined
- Series of choices -> there’s no one right route
- Generative, circular process
- Explore areas where there is some similarity of meaning - cluster together potentially connected codes into candidate themes and explore initial meaning
-> look over list of codes + try to cluster them into meaningful groups - Initially consider all codes + explore whether any broad ideas of number of codes that could bee clustered together
- If there is a core idea, but also variation, likely to be a theme
- You are trying things out -> end up with a number of provisional themes
- Consider what story your provisional themes tell you about your dataset in addressing your research question
- Good themes are distinctive + separate from others
- if you feel there is a separate construct within the theme, but is somehow distinct, you can develop a subtheme
What are thematic maps?
- Using visual mapping for theme generation, development and review
- Thematic maps may help is you are a visual person
- Helps identify themes and subthemes
- Theme tables are a common way of visually presenting your themes and subthemes in the analytical section of your report
What are 5 key things in theme development?
1) Themes do not have to capture everything in dataset
-> some codes might not be included if they are irrelevant to research question
2). Each theme should have a central organising concept
3) Don’t get too attached - candidate themes
-> you will be changing your themes and subthemes for the final version
4) Can have more themes than you end up with
5) Try to avoid question and answer orientation as they constrain ability to notice patterned meaning across dataset and prevent exploring those not immediately obvious
Phase 4: Developing and reviewing themes
- Thematic mapping helps to review tentative themes
- Identify boundaries
-> Is there enough meaningful data to evidence themes, are they nuanced, complex, diverse? - Are data contained within each theme too diverse, wide-ranging
- Does the theme convey something important?
- If not, rework and discard some
Phase 5: Refining and naming themes
Theme definition - write a few sentences that clarify your theme (take home point): what is it about? what is the boundary? what does each theme contribute to overall analysis?
- Naming themes - should convey gist - short phrase that captures essence of theme (not one word)
- Poorly named theme misrepresents the data
- Problems: themes might be too descriptive (won’t capture deeper meaning in them) or over-interpretive (they are not rooted in the data in your codes, quotes and transcripts)
How do you write up your analysis?
- Might not have space to present all themes in a report
- Best way to learn is to read over thematic analysis papers
When should you link analysis to wider literature?
- Some argue that early reading can narrow your analytic field of vision, leading you to focus on some aspects of the data at the expense of other potential crucial aspects
- Other argue that engagement with the literature can enhance tour analysis by sensitising you to more subtle features of the data (Tuckett, 2005)
- There is no right way to proceed with reading and incorporating that into your analysis
- For TA, although a more inductive approach would be enhanced by engaging with literature in early stages of analysis
- A theoretical, deductive approach requires engagement with literature prior to analysis
-> also depends on approach you take for your TA