Codes vs themes in qualitative analysis: what is the difference?

A code labels one idea in your data. A theme is a pattern of meaning built from several codes that together answer part of your research question. Students who confuse the two end up with a findings chapter that lists topics and never makes a point.

Virginia Braun and Victoria Clarke, whose 2006 article in Qualitative Research in Psychology is the source most committees expect for thematic analysis, put it directly: a theme is not a topic that appears in the data. It is an idea about the data, developed through analysis.

What a code is

A code is a short label attached to a segment of data. It names something that is there. In a study of mathematics anxiety among engineering students, interview extracts might receive codes such as "avoids asking questions in class," "guesses to avoid being wrong," "compares self to classmates," and "panics when timed."

Codes are small, specific, and close to the participants' words. You will have many of them, and some will overlap.

What a theme is

A theme gathers related codes around a central idea and says something about the data. From the codes above, one theme could be "Fear of being seen as wrong limits risk-taking in mathematics." That sentence makes a claim. It joins "avoids asking questions," "guesses," and "compares self" because each shows the same thing.

Compare it with a topic label such as "Anxiety." A topic names a subject. A theme states what is going on with it.

A quick test

If you can replace your theme name with a single word, you probably have a topic. If your theme name is a sentence, or can become one, you probably have a theme.

How to build themes from codes

1. Code the whole data set, closely and in detail.
2. Group the codes that point to the same idea and note the ones that do not fit anywhere.
3. Draft a theme for each group, and write one sentence stating what it says.
4. Check each theme against the data. Does it hold across participants? Does it contradict any extracts?
5. Check each theme against your research questions. A theme that answers none of them is either a different study or a code that has not finished developing.
6. Name the themes with phrases that carry the claim.

Our guide to presenting themes clearly in Chapter 4 shows how to write each theme up once it is built.

Why themes read as loose

When a discussion section does not hold together, the cause is usually at this stage. Themes overlap, or two themes are really one, or a theme is a topic and has no claim. The fix is to write one sentence per theme and test whether each sentence answers a research question. Johnny Saldaña's The Coding Manual for Qualitative Researchers describes coding methods in detail, and many programs list it for this stage.

What to write in Chapter 3

State the analysis approach, cite its source, and describe the steps you took. Your readers must be able to see how codes became themes. See Chapter 3 and reporting interviews.

Not sure your themes hold together?

The Two Dissertation Moms offers a Dissertation and Thesis Chapter Review that tests whether each theme makes a claim, answers a research question, and carries into your discussion.