3 reasons why you cannot find your themes / Thematic analysis in qualitative research

TL;DR

Themes require usable data that directly supports the research questions; misaligned interview questions or overly open interviewing can make theme development impossible.

Briefing Cornell Notes

Briefing

Qualitative researchers often get stuck on thematic analysis not because themes are mysterious, but because the prerequisites for building them are missing—either in the data itself, in the coding foundation, or in the clarity needed to turn codes into a coherent story. The most serious failure mode is straightforward: there may be no usable data for the research questions. If interview questions don’t elicit what the study needs—because the interview guide was poorly designed, questions were off-target, or the interview was too open so participants covered unrelated topics—then the analysis cannot produce themes that credibly answer the research questions. A related problem is changing the research questions after data collection in ways that the existing data cannot actually support, or shifting them before collecting data without ensuring the new questions align with the instrument used to gather evidence.

A second common bottleneck is coding that isn’t done in a way that can support theme development. Coding is described as the foundation of analysis: codes break the data into manageable parts, and themes are built from those parts. Without detailed, well-structured coding, it becomes impossible to create themes that are convincing and valid, since no researcher can reliably remember every participant statement across a full dataset. The transcript emphasizes that codes should function like a “table of contents” for the data—detailed enough to capture what matters, grounded in the data rather than overly abstract interpretation. When coding produces too few codes after only a small number of interviews, that’s treated as a warning sign. Overly theoretical or anticipatory coding—where the researcher starts “thinking ahead” toward themes instead of capturing concrete descriptions—can cause important details to be missed, forcing a return to recoding.

The third reason is more subtle but extremely common: even with the right data and the right coding, researchers can stall when they lack clarity about what their findings are and what narrative they want to present. Themes aren’t treated as something that simply “emerges” fully formed; they are developed. That development depends on deciding what the codes reveal in relation to the research questions—what is now known that wasn’t known before the study. Once that clarity exists, the researcher can reorganize the code set: sorting, rearranging, renaming, and deleting codes to reduce redundancy and sharpen the analytic structure. The transcript also challenges expectations about how themes appear, arguing that confidence in theme construction comes from making deliberate decisions about which codes belong together and which should be removed.

Overall, the path to themes is framed as a sequence: ensure the instrument produces data aligned with the research questions, code in a detailed and data-grounded way, then use the resulting coding structure to decide the findings and build themes as a coherent presentation for readers. If any step is missing—usable data, a solid coding foundation, or analytic clarity—theme development stalls.

Cornell Notes

Themes in qualitative research don’t “pop out” automatically; they’re built from what the dataset can actually support. If the interview guide or data collection doesn’t generate information that answers the research questions, themes can’t be developed credibly. Even with good data, weak coding blocks progress—codes must be detailed and grounded in the data, acting like a table of contents that makes theme-building possible. Finally, researchers need clarity about their findings: what the codes teach in relation to the research questions. With that clarity, codes can be reorganized, renamed, and reduced to form a coherent thematic narrative for readers.

What does it mean to say themes “aren’t there,” and how does that prevent thematic analysis from working?

The transcript treats themes as representations of findings, and findings must exist in the data. If the interview guide doesn’t produce responses that address the research questions—such as poorly designed questions, questions that miss what the study needs, or an overly open approach where participants talk about unrelated topics—then the dataset lacks the evidence required to answer the research questions. Changing research questions can also break alignment if the collected data can’t support the revised questions.

Why is coding described as a prerequisite for themes rather than an optional step?

Coding is presented as the foundation of analysis because it breaks the dataset into parts that can be examined systematically. Since researchers can’t realistically remember every participant statement across the whole dataset, codes provide a structured record of what each segment says. Themes are then built from these codes; without a solid coding structure, theme development lacks the necessary grounding for credibility and validity.

What coding mistakes can lead to being unable to find themes even when data is correct?

A key warning is coding that becomes too abstract or interpretive too early. Instead of using codes to capture concrete descriptions and summaries of what participants said, researchers may start “thinking ahead” toward themes. Another red flag is having only a handful of codes after coding more than a small number of interviews, which suggests something is off—either important details are being missed or coding is too broad. In such cases, the transcript recommends revisiting and recoding.

If the data and coding are both right, what typically causes the final stall before themes?

The transcript points to lack of clarity about the findings and the narrative to present. Themes are described as something developed, not something that simply emerges. Researchers must decide what they learned from the data by looking at codes alongside the research questions—identifying what is now known that wasn’t known before. Without that decision, it’s hard to know which codes to keep, combine, rename, or delete.

How does “clarity” translate into concrete actions during theme development?

Once findings are clear, researchers return to the code set and reorganize it: sorting and rearranging codes into candidate themes, renaming them to reflect meaning, and deleting redundant or irrelevant codes. The transcript emphasizes that many datasets produce lots of codes, so reduction is expected—but reduction requires confidence grounded in the earlier clarity about what the codes reveal relative to the research questions.

Review Questions

What alignment checks should be performed between the interview guide, the research questions, and the data before attempting to build themes?
How can you tell whether your coding is detailed enough to function as a “table of contents” for your dataset?
What steps would you take to create the “clarity about findings” needed to reorganize codes into themes?

Key Points

1
Themes require usable data that directly supports the research questions; misaligned interview questions or overly open interviewing can make theme development impossible.
2
Changing research questions must be handled carefully to preserve the ability of the collected data to answer them.
3
Coding is the foundation for themes; without detailed, data-grounded codes, themes lack credibility and validity.
4
Overly abstract or anticipatory coding can cause important details to be missed, often resulting in too few codes.
5
A common final barrier is not coding quality but lack of clarity about findings and the narrative to present.
6
Once findings are clear, theme-building becomes an organizing task: rearrange, rename, and delete codes to reduce redundancy and form coherent themes.

Highlights

Themes are treated as representations of findings, and findings must exist in the data; if the instrument doesn’t elicit answers to the research questions, themes can’t be built credibly.

Codes function like a table of contents for the dataset—detailed coding is what makes it possible to develop themes systematically.

Even with correct data and coding, researchers can stall without deciding what the codes reveal in relation to the research questions; themes are developed through deliberate organization, not automatic emergence.

Topics

Thematic Analysis
Qualitative Coding
Interview Guide
From Codes to Themes
Research Question Alignment