Research Methods

Thematic Analysis Coding Laboratory

Six fictional interview extracts about asking for help at university. Code them, then set your reading beside three other defensible ones over exactly the same words.

Fictional — invented extracts written for teaching, not transcripts

Learning objective

By the end you should be able to treat coding as an analytic act rather than an extraction, and defend a coding choice as one of several readings.

About 30 minutes. Anything you type stays in this browser tab and is not saved or sent anywhere.

There is no answer key here

This tool does not mark your codes and does not compare them with anyone else's for agreement. In reflexive thematic analysis the researcher is part of the analysis, so two competent analysts coding the same extract differently is expected rather than a failure of reliability. What is assessed, in real work, is whether a reading is traceable to the data, coherent, and explicit about its own assumptions.

Answer the opening question to unlock the laboratory.
Read the extract, then write a code of your own for it.
Optionally tick codes from the bank and read what each one does.
Compare with three coding passes, then answer the reflexive prompt.

This part needs JavaScript

The laboratory presents extracts and assembles coding passes in your browser, so it cannot run with JavaScript switched off. The rest of this page is readable without it.

Two researchers code the same interview and produce noticeably different codes. In a reflexive thematic analysis, what does that tell you? The coding is unreliable and they should agree a scheme and recode One of them has misread the data Very little on its own — it is what you would expect, and the question is whether each reading can be defended That coding is subjective, so any reading is as good as any other

Key terms

Code: A short label recording something the analyst has noticed. It is a decision with an author, not a category the extract falls into.
Semantic and latent: Whether a code stays with what was said, or reaches for what is being done in saying it. The two are ends of a continuum rather than separate boxes.
Coding pass: One route through the whole dataset. A second pass over the same words can be a different reading rather than a mistake.
Reflexivity: Attending to what you brought to the reading, and what that made available or hard to see.

Write your own code

A short label for what you notice in this extract

Stays in this tab. Nothing is saved or sent anywhere.

Would you call your code semantic or latent?

Or borrow from the bank

Six candidate codes. None is the right one; each does a different job, and ticking it shows what that job is.

Extract 1 of 6

Coding is a decision, and decisions have authors

A code is not sitting in the transcript waiting to be lifted out. It is a short account of what the analyst has noticed, written in the analyst's words, from a position the analyst occupies. That is why the same extract in this laboratory yields three different sets of codes without any of them being a misreading.

Semantic and latent are a continuum, not two boxes

Staying close to what is said is a real analytic choice with real costs: it keeps you honest and it can leave the pattern invisible. Reading for what is being done, or for the setting that makes a thing sayable, reaches further and owes the reader more justification. Most analyses move along this line rather than picking a point on it.

Agreement is not the criterion

In reflexive thematic analysis, two coders reaching the same codes is neither necessary nor especially reassuring: it can mean the coding frame was tight enough to prevent anyone noticing anything. What a reader is entitled to is data extracts that support the claims, an account that hangs together, and a statement of the position it was written from.

Software organises; it does not analyse

Qualitative analysis packages store codes, retrieve extracts and count them. Every interpretive act in this laboratory happened before any of that would be useful. A tidy codebook is a record of thinking, not a substitute for it.

What this does not demonstrate

Six short extracts are not a dataset. Real coding happens across whole transcripts, where a phrase means something different because of what came twenty minutes earlier.
The extracts are invented. They were written for teaching. No participant, interview, institution or study is being represented.
Three passes are not the three possibilities. They are three that make the range visible. A narrative, psychosocial or feminist reading would produce others again.
Coding is not the analysis. Themes are developed later, across the whole dataset, and a good codebook does not guarantee a good theme — that is what the next two tools in this module are about.
Nothing here is a measure of you. The tool does not score your codes, store them, or infer anything about you from them.

Thematic Analysis Coding Laboratory

Learning objective

There is no answer key here

First, a judgement

The laboratory

Asking for help — six fictional extracts

Extract 1 of 6

What the codes you ticked are doing

Three coding passes over the same words

A reflexive prompt

Challenge — which of these are codes?

What this demonstrates

Coding is a decision, and decisions have authors

Semantic and latent are a continuum, not two boxes

Agreement is not the criterion

Software organises; it does not analyse