Research Process: Analysis of Qualitative Research Interviews

Research process with qualitative research interviews

Methods of analysis/methodologies vary

• what is the world like (ontology)

• how should we learn about it (epistemology)

• what can we ask about it? (Research question - RQ)

Realism

Reality is ‘out there’ and discoverable through the research process; aim of research is to get an accurate picture of human psychology (e.g.,via hypothesis testing; some form of qualitative analysis as some form of TA)

Ex. of RQ

What cognitive factors influence athletes’ confidence?

Critical realism

Reality is ‘out there’ but only accessed via participants’ interpretations of own experiences and researchers interpretation of participants interpretations - direct access to reality is never possible. Aim of research is to get at participants personal ‘sense-making’, to get a picture of human psychology as accurate as possible.

Ex of RQ:

How do suspected criminals experience guilt?

Constructionism/social constructionism

Different constructions/representations in different contexts: research’s aim to get at how specific event, phenomenon or activity is constructed in a specific context.

Ex of RQ:

How do parents persuade their children to eat vegetables?

How to approach/analyze qualitative research interviews

Thematic analysis (TA)

• TA offers a meaningful analysis of data by focusing on the meaning (or what people are saying) and patterns

• TA is essentially a method of data analysis - not methodology - for identifying and analysing patterns in qualitative data

  • theme is coherent and meaningful pattern (of what is said) in the data (i.e., pattern which co-occurs in a meaningful and systematic way)

  • concerned with searching, identifying and analyzing relevant meaning or themes in the data

• provides foundations for other methodologies, names Interpretative Phenomenological Analysis (IPA) and Grounded Theory (or GT) (resly upon thematic coding)

Research questions and different approaches in TA

Types of qualitative research interview & data collection

• often used with qualitative research interviews (such as semi-structured interviews) and focus groups data (but also with other types of research generated data such as open-ended questions in questionnaires)

(NB: can be used with any data that can be examined for meaning - i.e., qualitative data)

• small and big data sets

• sampling is usually purposive or convenience

Transcription analysis in TA (recap)

• transcription: through orthographic transcription (often without punctuation) of what is said (sometimes together with transcription of basic non-linguistic cues such as ((pause)), stresses, (mm), [laughter], ‘reported speech’

• step by step analysis but not single way of doing TA (ex: Howitt, 2016; Braun & Clarke, 2006, 2013, 2021)

• TA enables to identify a limited number of themes/patterns of meaning which adequately reflect the data

  • analysis based on coding the data and forming themes from the codes

Analysis in TA (recap cont.)

• codes and themes can be deductive (codes and themes reflect pre-existing theories - i.e., code and themes are generated by the researcher), inductive (codes and themes are generated from the data and aim at staying as close as possible to the meaning in the data - i.e.,’emerge from the data’) or both (deductive and inductive)

• Realist TA = deductive coding and themes; critical realist TA = inductive coding and themes or both (deductive and inductive); social constructionist TA = inductive coding and themes

• so themes identified depend on the research question, types of TA (approach) and data

• so before analysis: decide upon your research question, identify the approach you want to take and thus type of TA

TA: step by step analysis by Bruan & Clarke (2006:87)

Step 1: familiarizing with the data

• read and re-read the data (the transcriptions of the interviews)

• active reading of the data - read in curious and questioning way

• make notes of initial observations and ideas about the data

• ask the following questions (Clarke, Bruan, & Hayfield, 2015: 231-1)

  • why are they making sense of things in this way?

  • what assumptions underpin this account?

  • what type of world-view does that account imply or rely on?

  • how would I feel in this situation?

  • what implications does this account have?

Step 2: generating codes

• coding involves identifying and labelling key things about what is going on in the text (= basic units of meaning)

• codes are used to label manifest/explicit meaning - can use different types of labels (e.g., descriptive or in vivo labels with minimal interpretation)

• codes can also label latent/implicit meaning (such as labelling assumptions, world views that lie under the data surface)

• not everything is coded: identify relevant aspects of the data that are relevant to your research question(s)

  i.e., depending on the research question & data: code line by line, every two or three lines

Coding: examples

• same segment of text can be coded with different codes

• for example, coding can be (Saldana 2016):

  • descriptive (i.e., labels that summarise the topic in the data)

  • in vivo (i.e., using words or phrases from the text: this coding involves minimal interpretation of what’s going on in the text)

  • categorize labels (i.e., categorize the data for what participants are talking about, e.g. individual values/attitudes/beliefs, emotions, etc)

• is this a problem? no, but the codes used need to reflect adequately what is in the data

Coding: example from Bruan & Clarke (2013:208)

Step 2: coding

• because the same segment of data can be coded with different codes

• apply the codes systematically - decide whether to use an existing code or develop a new code - and thoroughly (i.e., apply the codes to the entire data set and do it at least twice)

• coding is not a static process so initial codes can be revised as the researchers proceeded through the transcript

• develop a codebook (but not for reflexive TA)

Step 2: collating data relevant to each code

• check wording of codes and be systematic (i.e., researcher might notice that they used two or more initial codes for the same thing; adopt one and apply it systematically)

• collate the coded data (ex: copy and paste the relevant extracts coded under the same code into a new document)

NB: At this stage, the objective is not to develop broader themes but to capture the essence of a segment of the text through codes

Step 3: searching for themes

• a theme captures cluster of related codes

  • theme are essentially obtained by joining together (or collapsing together) several of the codes in a meaningful way

  • far fewer themes than codes

  • themes thus reflect higher level of analysis (abstraction) than coding (more interpretative step than coding as themes are meant to make connections between codes)

• different levels of themes: overarching themes (capture an idea underpinning a number of themes), themes (capture the meaning related to a central concept/idea) and sub-themes (capture different aspects of the same theme)

Step 4: reviewing themes

• checking whether the themes work in relation to the coded extracts (i.e., check whether the candidate theme is a good fit with the mean of the coded data)

• check whether the themes work in relation to the entire data set (i.e. whether they capture key meanings and patterns in the data)

• at this stage: researcher can change themes, discard themes or restart the entire analytical process (coding of data & searching for themes)

Step 5: defining and naming the themes

• understand the scope of each theme and how the themes relate together

• generate a thematic ‘map of the analysis’

• and provide short descriptions of each theme

• provides names to your themes

Producing the analysis (step 6): example for theme ‘exercise is evil’ (Braun & Clarke, 2013: 255)

Criteria for good TA (Braun & Clarke, 2006:96)