L8: Introduction to qualitative research Flashcards
why do qualitative research in health?
The real world is complex with many interacting factors
Certain human cognitions and behaviours cannot be identified by a QUANT research design
These obscured factors can be clinically relevant
Deeper insight can inform theory as well as practice, and generate hypotheses
Examples:
Patients’ experiences of receiving advice from clinicians
Psychological impact of treatment
Experience of a care pathway
Barriers to the provision of quality care
Psychological benefits of physical activity in a specified patient cohort
commonly used data collection methods
Interviews
Recorded conversations with people of interest, sometimes with structured questions.
Focus groups
Recorded group discussions with groups of interest, sometimes with structured questions.
Ethnography (less common)
Immersion of the researcher in the physical environment of interest (e.g. a urology clinic), often over a long period of time.
quantitative vs qualitative
Qualitative
sampling: Aim for deep and relevant insights – sample only as appropriate for this aim
procedures: Some study procedures can be fluid and evolve over time
data: Text*, produced organically
premise: Seeks perspectives and subjectivity – coherent data insights provide meaning to analysis
objective: To generate fresh insight (can be hypothesis generating)
analysis: Logical and theoretical interpretation to produce conclusions
quantative
sampling: Aim for large, reliable, and generalisable sample
procedre: Rigid study protocol, for consistency of data
data: Numerical values, measuring preconceived variables
premise: Seeks objectivity – numerical values give meaning to analysis
objective: To confirm a preconceived hypothesis
analysis: Statistical inference to produce conclusions
types of sampling
Purposive sampling
Actively seeking individuals who have specific characteristics
Convenience sampling
Sampling based on who is available/accessible
Snowball sampling
Enrolled participants suggest further participants who may be suitable
methods of analysis
Thematic analysis
Framework analysis
Grounded theory
(these 3 are the most common in health research)
Interpretive phenomenology
Conversation analysis
Discourse analysis
Narrative analysis
coding
Turning raw data into conceptual units
Different methods of coding, depending on analysis method
Inductive (emergent) or deductive (assessment)
Can require researcher’s interpretation, depending on analytical approach
Coding Pragmatically by Segments
What it is:
You break the data up into meaningful “segments” (chunks, sentences, phrases, paragraphs) based on content — and then assign a relevant code to each.
How to do it:
Read through the text.
Identify where one idea or meaning unit begins and ends.
Code just that segment, based on what it’s about.
Example:
If someone says:
“I don’t really trust doctors, but I still go to the GP when I feel ill.”
→ You could segment this into:
“I don’t really trust doctors” → Code: “Distrust in healthcare system”
“but I still go to the GP…” → Code: “Utilizing healthcare despite reservations”
✅ Use this when your goal is clarity and focus on meaning in manageable pieces.
🔹 Deductive Coding
What it is:
You go in with a pre-existing coding framework, usually based on theory, research questions, or literature.
How to do it:
Create a codebook before you analyze the data (based on CFIR, HBM, TPB etc.).
Go through your data and apply those preset codes where they match.
You can note extra codes if something new pops up, but your focus is on applying the framework.
Example:
If you’re using the Health Belief Model, your deductive codes might be:
Perceived Susceptibility
Perceived Severity
Perceived Barriers, etc.
→ Then you look for parts of your data that relate to those concepts.
✅ Use this when testing a theory or comparing to a known model.
🔹 Open Coding
What it is:
You start with no predefined codes and let the data speak for itself. It’s inductive, exploratory, and totally open-ended.
How to do it:
Read through transcripts.
Mark anything that seems interesting, relevant, or repeated.
Assign a label that describes the content — you’re making up the codes as you go.
Example:
If someone says:
“I didn’t get the vaccine because I was scared of the side effects.”
→ You might code: “Fear of side effects”, “Vaccine hesitancy”, or “Risk perception”
✅ Use this when you’re exploring new topics, developing themes, or grounded theory.
thematic analysis
Thematic Analysis (TA)
✅ Why it’s so common in health research
It’s flexible — you can keep it simple (e.g., identifying just a few main themes) or make it complex (e.g., deeply analyzing latent meanings).
It works with a variety of research designs and data types (interviews, focus groups, open-ended surveys).
Doesn’t require allegiance to a specific theory (unless you’re doing it deductively).
🧠 Purpose of Thematic Analysis
To identify, analyze, and report patterns (themes) within data.
To present key elements of what people are saying, in a way that makes sense to others (including policy makers, clinicians, etc.).
Not just summarizing — you’re trying to interpret meaning and capture the essence of what’s being said.
Familiarise yourself with the data
Listening to audio, reading and re-reading transcripts and notes, making notes
Write summaries of each case: participant characteristics; interview context; demographic information
- Identify provisional themes
Text analysed as pragmatic segments
Segments compared with other segments, to identify similarities or differences
Deductive: themes pre-determined, ‘applied’ to data
…or…
Inductive: themes conceived from analysis of the data and refined
Code the data
Develop a coding scheme
Apply coding scheme to more data (but not whole dataset)
Refine / ‘firm up’ codes
Think about relationships between codes at this point
How rigid your coding scheme is depends on your approach / research question - Organise codes and themes
Horizontal analysis (across cases)
What was said about each theme?
Deeper analysis: relationships, typologies, meaning – what does the data ‘say’ a
framework analysis
Geared toward developing policy or practice-oriented findings
Less emphasis on theoretical or abstract analysis, more focus on the surface level
Rigid framework developed
2D framework populated with summaries of data
Populated framework analysed x against y
Steps 1, 2 & 3 are similar to thematic analysis
4. Analysis facilitated by ‘charting’:
This is what makes Framework Analysis special.
You summarize data from each transcript under relevant themes (usually in a matrix or table).
Each row = one participant or case
Each column = one theme or subtheme
You “lift” relevant data from the transcripts and drop them into the right cells.
This allows for systematic comparison across cases and themes, which is really useful for drawing out patterns and differences.
- . Mapping and interpretation
Relationships between factors observed and documented
Often done using diagrams and tables to physically explore relationships
Analysis works towards a specific aim
(e.g. designing a questionnaire, or informing a particular aspect of clinical practice)
grounded theory
Aims to fracture data into the smallest constituent parts, before reassembling these into a coherent whole (a grounded theory)
Designed to construct a mid-range theory, grounded in the data
Strong emphasis on findings emerging organically from the data
Iterative and cyclical process, guided by emerging data
- Open coding
Line-by-line analysis of transcript
Designed to avoid closing off avenues of enquiry, to generate as many codes as possible
Useful to look for in vivo codes – those that show how the participant divides up the world
e.g. “I’m just not someone who exercises” tells you that the participant divides the world up into ‘people who exercise’ and ‘people who don’t exercise’.
Not to be applied to the whole dataset – too intensive
2. Axial coding
‘Putting data back together again’
Finding relationships between codes / categories
A good method is to ask questions of each code:
e.g. if code = ‘pushing to change exercise habits’
Is pushing to change habits influenced by ‘normative beliefs about PA’?
Can ‘pushing to change exercise habits’ be a manifestation of ‘exhibiting defiance’?
What psychological prerequisites promote this type of willpower? Do other codes give us a clue?
Are the codes ‘pushing to change exercise habits’ and ‘embracing apathy’ mutually exclusive? Can they occur within the same individual?
3. Selective coding
Takes place following theoretical saturation
Categories and codes are accounted for, no new categories are emerging
More abstracted, analytical
Informed by wider theory in the field (e.g. social cognitive theories)
Intends to generate ‘core categories’, which will form the essential structure of your theory
Core categories are related to most other categories
A grounded theory is formulated and described – this is your ‘Results’
Essential components of grounded theory analysis:
Constant comparison
Memo writing
Theoretical sampling
Seeking deviant cases
A cyclical approach that aims to iteratively refine and test codes:
appraisal of qualitative research
Credibility
Are the study findings believable based on the analysis reported? Is there a logically consistent thread from data to analysis to findings?
Confirmability
Are the study findings and conclusions supported by the data presented?
Transferability
Can the findings be transferred to other situations, considering both the findings themselves and how they were established? How useful are the findings to others?
Trustworthiness
Can the methods used by the authors to establish credibility, confirmability, and transferability be trusted?
Triangulation
How was triangulation applied in order to cross-reference information and insights? Were there multiple methods, analysts, or reliability checks of sources? Did these corroborate?
Content validity (related to the concept of ‘saturation’)
To what extent can you be sure that the findings represent the totality of information relevant to the research question being studied? What is the likelihood that key insights may have been missed? Are the conclusions suitably aligned/caveated?
Reflexivity
Are the authors transparent about how they may have personally influenced the data or findings? Are the authors cognisant of their own characteristics (e.g. gender, ethnicity, education, profession, class) and how these may fit into existing sociologies of power which might be relevant to the data?
Theoretical base
Is the research question and/or analysis situated within an established theoretical framework? If yes, has the framework been applied appropriately? Is the application conceptually faithful?
