[FIELDS] Thematic Analysis

reflexive thematic analysis

• Method under the qualitative research paradigm for systematically constructing, organizing, and offering insight into patterns of shared meaning (themes) across a data set 

  • To construct themes that are relevant to answering a particular research question

  • A relatively accessible and flexible method to doing qualitative analysis

• works with codes and themes

codes

• Code: provide a tentative label for a feature of the data that is potentially relevant to the research question

  • Basic blocks of interpretation by the researcher

  • Codes are constructed by the researcher into themes

themes

• Themes: a theme is a pattern of shared meaning underpinned or united by a central organizing concept

  • Constructed from codes and the data set

earlier iterations of TA

• In earlier iterations, Thematic analysis focused on accessibility and flexibility

  • As a method of analysis, it is not tied by default to any ontological + epistemological worldviews and schools of thought 

  • Can be used as inductive (Data-driven) and/or deductive (theory-driven) approaches 

    • Inductive: objectivist + postpositivist

    • Deductive: constructivist + interpretivist

  • The quality and validity of your TA rests on how consistent you are with your standpoint

what makes reflexive thematic analysis reflexive?

• Have to be cognizant with what that stance is 

• Less connected to post positivism 

• Postpositivists don't think that they have to be cognisant to be post positivist - the only way to value stance 

1. differentiates itself from coding reliability TA

2. differentiates itself from codebook TA

3. reflexive means valuing

4. researcher’s role in production

RTA differentiates itself from coding reliability TA

• Postpositivist; creates themes after phase 1

• Uses multiple coders and statistics to determine reliability and validity 

• Preconceived themes -> analyze the data using these

• Not emergent

RTA differentiates itself from codebook TA

• Applies a framework based on quali research philosophies. Focused on the framework 

• Also generally creates themes after phase 1

• Use thematic analysis to try to analyze it 

• In between but they already have a school of thought 

reflexive means valuing

• A qualitative paradigm (rather than quantitative) 

• Researcher subjectivity

• Organic and recursive coding processes 

• Deep reflection on and engagement with data

• Take a step back and apply school of thought and use their own but still have the sensibility of TA

• Not strictly -> getting the skills that is learned from doing TA

• Means valuing several things 

• Personal process

• Admit that you are not a detached scientist 

researcher’s role

• The researcher’s role in knowledge production is at the heart of our approach 

  • fully aware of the framework they are using and its assumptions 

  • ook at the assumptions they are making when interpreting the data

  • Social constructivism 

  • know the decisions they need to make; engage in the decision they need to make

  • Results are stories co-created by the researchers and the data using the framework

RTA process

1. familiarizing yourself with data

2. generating initial codes

3. generating initial themes

4. reviewing potential themes

5. defining and naming themes

6. producing the report

familiarizing yourself with the data

• Recommended: listen to the audio at least once (as a team), then read the transcript several times (at last twice)

• While listening to the audio, make/write individual notes

• Make separate notes for each time you read the transcript or listen to the recording 

• Compare notes across readings/listening sessions 

• Make notes but don’t make in depth codes yet

[P1] questions to keep in mind

Questions to keep in mind

1. How does this participant make sense of their experiences?

2. What assumptions do they make in interpreting their experience?

3. What kind of world is revealed through their account?

generating initial codes: code

• Codes: provide a tentative label for a feature of the data that is potentially relevant tho the research question

  • Building blocks of analysis 

  • Semantic codes:  a summary of some portion of the data  -> summary of life

  • Latent codes: identifies participant meanings, assumptions, tensions that lie beyond the surface of the data interpretation

semantic codes

• a summary of some portion of the data  -> summary of life

latent code

identifies participant meanings, assumptions, tensions that lie beyond the surface of the data interpretation

generating initial codes

• Another transcript -> make quotes on the side

• Point more clearly where interpretation comes from

• Read the transcript again, generating a mix of latent and semantic codes per block of text

  • Block of text here can mean a phrase, a sentence, or an entire chunk of data

  • Make sure you code one black in its entirety before moving onto another. No codes within codes

• In this phase, code as much as you can

  • Multiple codes can be made for one line at this phase

generating initial themes: theme

• theme: A pattern of shared meaning underpinned or united by a central organizing concept

  • A main idea that several codes relate to, formed out of the researcher’s interpretation of participant meanings 

generating initial themes

• A theme captures and represents patterned response or meaning within the data set in relation to the research question 

• A theme’s saliency is not fully dependent on how often it is seen in the data set

• In relation to research question

[P3] themes are not just summaries of data

• Themes are not just summaries of data domains; there is a focus on shared meanings, eve n when highlighting differences 

  • E,g, gender; summary of a data domain a feature

  • Theme: The gendering of emotions

• Some themes are 

  • Semantic: dependent on surface level answers 

  • Latent: looks at the underlying assumptions and ideologies beneath the utterance in the transcript 

[P3] semantic themes

• dependent on surface level answers

[P3] latent themes

Latent: looks at the underlying assumptions and ideologies beneath the utterance in the transcript 

[P3] searching doesn’t fully capture the essence of the phase

• Searching: doesn’t fully capture the essence of the phase 

  • We actually construct themes (remember constructionism) 

  • Reflexive TA assumes that the themes are not in the data but constructed by the researchers 

  • Builder 

[P3] the phase involves reviewing coded data

• The phase involves reviewing the coded data to identify overlapping areas/similarities

  • Codes will cluster together, and around a general idea and subthemes 

  • Codes will be discarded

  • Easier to make subthemes then themes -> compare to research questions 

[P3] always return to the research question

• Always return to the research question 

  • Some codes will be merged into subthemes 

  • Some codes will be discarded

[P3] phase involvement

• This phase involves reviewing the coded data to identify overlapping areas/similarities 

  • Begin exploring relationship between themes will tell an overall story of data as an answer to your research question

  • Think about how the themes will tell an overall story of data as an answer to your research question 

  • Assign discarded codes into a miscellaneous themes 

  • Recommended # of themes; 4, range of 2-6 

  • Set it aside dont completely throw 

  • End this phase with construction of an initial thematic map

reviewing potential themes

• Review the themes in relation to the coded data and the entire data set 

  • Is there enough meaningful data to support this theme? Is it a theme or just a code?

  • Does this theme directly answer my research question?

  • What does this theme exclude/include?

  • Is this theme to broad/narrow?

[P4] Possible things you can do when reconstructing: 

• Collapse weakly supported themes together

• Splitting a broad theme into two or more themes 

• Discard a theme if it doesn’t really answer the research 

defining and naming themes

• When defining and naming themes, ensure that 

  • The themes answer your research question 

  • You can summarize each theme in a 4-5 sentences 

  • Each theme has one central organizing concept, which you can explicitly sate

  • Make sure that the central organizing concept is reflected in the theme title 

  • The themes are related to other themes. But do not overlap

  • There are a sufficient number of quotable excerpts that represent the theme

  • When taken together, themes provide structure for a coherent story about the data (the story which you will tell) 

[P5] naming the actual theme

• The name is informative, concise, memorable and catchy 

  • You can use parts of quotable excerpts in the titles of the themes, if they are striking examples 

[P5] transitioning into writing

• Utilize the excerpts to tell a story about the data in relation to your research question 

• Tell the reader specifically what about quoted extracts are interesting and why; in relation to RQ

• Incorporate both descriptive and interpretative analysis of your themes and the excerpts you used

producing the report

• To ensure a good report structure for data storytelling 

  • Connect themes in a logical sequence 

  • E.g. past to present, public then family the self 

• Connect the themes in such a way that the next theme build on the previous theme 

• Build up an argument that answers your research question 

• Keep your writing concise and active

• Look at logical sequence 

general errors

• Adding data extracts/quotes without explaining them 

  • Only add exemplar quotes in the results write-up that you will explain 

• Using general topics as themes and thus theme names 

  • Does your theme name sound like interview question in topic form? = error 

  • Make names according to the content of data

• Make sure your data narrative/argument fits your basic worldviews/ontological positions

common errors in coding

Common errors in coding 

• Making your blocks of analysis too large 

• Coding too short or too long: stick to codes having at least 3 words and no longer than 18 words 

• container coding

• unanchored coding

container coding

• this is when you code for the more general topic rather coding for the actual content

• Ex. emotions towards relationship

• Code the actual emotion and some context

unanchored coding

• when the code made cannot be pointed to a set of words in the quote

  • More general

recommendations: contexts

• One of our many contexts as they researchers is that we are doing psych research 

  • Emotions 

  • External behaviors

  • Cognitive processes/internal behaviors

    • Rationalizing, problem solving, remembering, etc. 

  • Cognitions 

    • Memories, beliefs, attitudes, etc. 

recommendations

1. When a code closely applies to another quote, you can use the same code but try to add a new one

2. Avoid anecdotalism: when one or few instances of phenomenon as a reified as a pattern/theme

3. Make sure your data/narrative/argument fits your basic worldviews/ontological positions

   1. Review how social constructionists vs. objectivists approach research differently