9.1.13, 9.1.14, 1.2.6 - Qualitative analysis and research reports

From qualitative evidence to codes

Qualitative analysis investigates meaning in material such as interview transcripts, open questionnaire responses and observation notes rather than reducing every response immediately to a score.

Start with a focused question, such as “How do adult learners describe choosing a place to study?” Record data with appropriate consent, remove identifying detail carefully and retain enough context to interpret what a speaker means. Where interviews are recorded, an accurate transcript preserves what was said. A pause, tone or surrounding statement may matter; a selected sentence is not always self-explanatory.

For the original illustrative response “I go to the library because my flat is noisy”, possible codes include “avoiding noise” and “choosing a quieter setting”. More than one code can be appropriate if it captures distinct meanings. The analysis should explain its approach instead of pretending there is one universally correct code. Inductive coding is directed mainly by the data; deductive coding uses prior concepts or a theoretical question. Both still require checking that interpretations fit the actual material.

Coding is analysis, not just highlighting favourite quotations. Read the whole dataset, retain the link from a code to its extract and inspect exceptions. A codebook can specify rules for a shared coding approach. In reflexive analysis, the analyst also examines how their assumptions shape interpretation; an interpretation is constructed through engagement with evidence rather than discovered without any human judgement.

Counting mentions can answer a frequency question, but a frequently used word is not automatically the most important meaning. “Noise” could describe distraction, company or reassurance in different responses. A qualitative account should preserve these differences and avoid claiming that a convenience sample's statements estimate the prevalence of a view in the entire population.

Developing and testing themes

Thematic analysis develops patterns of meaning across a dataset to answer a research question, with themes broader than individual codes.

Researchers familiarise themselves with the material, generate codes, develop candidate themes, review them against extracts and the full dataset, define and name them, and write an evidence-supported account. These steps are iterative: a theme may need splitting or revision after returning to a contradictory response. Different forms of thematic analysis have different assumptions; Braun and Clarke's reflexive approach explicitly recognises the analyst's interpretive role. Guidance from the method's authors.

Here is a small original illustrative dataset, created for teaching rather than collected from real participants:

The first theme concerns managing conditions for concentration; the second concerns feeling comfortable enough to start. The last response prevents a simplistic conclusion that everyone needs silence. It also shows that “quiet” should not be treated as a fixed universal preference. The boundaries are provisional: with more data, concentration and comfort might overlap in ways that justify refining the themes.

A theme is a pattern organised around an important idea, not merely a topic heading such as “rooms”. Define what belongs in it, what differs from neighbouring themes and how it answers the research question. Use well-chosen extracts to support an analytic explanation. Preserve negative or contrary cases; do not choose only vivid quotations that make the interpretation look persuasive.

Thematic analysis can capture experiences missed by closed response options and use unexpected meanings to challenge assumptions. Its flexibility also creates a risk of unclear or selective analysis. Transparent decisions, reflexivity and close links between extracts and claims strengthen credibility. Inter-rater agreement can be useful for a shared coding frame, but it is not a universal quality requirement for every form of reflexive thematic analysis. Code and theme distinction.

Building a grounded theory

Grounded theory aims to develop an explanatory theory from systematic engagement with data, rather than stopping at a description of themes.

Researchers collect and analyse data together, using constant comparison to compare one incident with another, codes with other codes, and new evidence with emerging categories. They write memos about possible relationships, questions and uncertainties. Theoretical sampling means seeking further relevant evidence because the developing account needs testing or elaboration; it is not random sampling to estimate a population percentage. They refine categories and relationships until further relevant data no longer substantially develop the account, often called theoretical saturation. A small fixed classroom dataset cannot establish that saturation has occurred. Charmaz and Bryant's methodological account and original methodological guidance.