The most common reason why you are STUCK in your thematic analysis (with examples)

The most common reason thematic analysis stalls isn’t a lack of coding—it’s failing to decide the “core story” that the results must communicate. After producing many codes, researchers often face an overwhelming question: which codes should become themes, which should be merged or deleted, and how the final structure should look. The bottleneck, according to the transcript, is that the core finding hasn’t been chosen yet. Without that central narrative, there’s no reliable way to judge relevance, and the analysis turns into an endless loop of uncertainty.

The transcript frames thematic analysis as a responsibility of the analyst: codes help reduce data volume and make patterns visible, but themes still have to be generated. That means the key decision comes first—what is the main point of the findings, grounded in the research questions and what the data ultimately needs to communicate. Only once that core is set can researchers sensibly organize other themes around it, decide what to keep, and determine what to discard. The transcript also normalizes code deletion: deleting many codes is expected and even desirable, but it becomes possible only after the core narrative is defined.

Two examples illustrate how the same set of codes can lead to different thematic structures depending on the chosen focus. In a hypothetical COVID-19 study of nurses, one dataset might contain 15–20 emotion codes (anxiety, fear, stress, sadness). If the research question is about what emotions the pandemic evokes, the analyst would likely keep those emotion codes and build a theme that lists them. But if the study’s real focus is the pandemic’s impact on nurses’ psychological well-being, those many emotion codes may be merged into a smaller set—such as a theme or subtheme about negative psychological impact—because listing every emotion isn’t the central message.

A second scenario involves international students describing motivations for choosing a university. If the study aims to list motivations, those codes can be organized as motivations. But if the study instead examines perceived benefits and challenges, the same motivation codes can be reworked through renaming and reframing—e.g., “reputable institution” becomes a perceived advantage, “high-quality teaching” becomes a benefit, and “networking opportunities” becomes another advantage. The transcript emphasizes that the shift isn’t only about deleting or merging; it often requires renaming codes and reshaping themes so they fit the narrative.

Finally, a practice dataset on barriers to mental health access shows how thematic frameworks can be built from the same raw codes using different organizing logics. When the framework is oriented toward “ways to improve access,” codes about learning from TV, reading books, and online materials are not treated as separate recommendations; instead, they’re merged into a subtheme about raising awareness. In an alternative framework focused on “factors that benefit access,” the same items can be reorganized into internal versus external factors, with the wording adjusted so the structure matches the story being told. The throughline is consistent: without deciding the core narrative first, researchers can’t make defensible decisions about relevance, merging, deletion, or renaming—and they risk circling without progress.