Codes, Categories and Themes - Understand the difference

Codes, categories, and themes form a hierarchy—starting with the most specific “codes” and moving up to broader “categories” and the most inclusive “themes.” The practical problem is that researchers and supervisors often use these labels differently, sometimes even treating them as interchangeable in write-ups of findings. That mismatch matters because it affects how results are organized, how readers interpret the analysis, and how easily a student can defend their structure during supervision.

In the clearest, “ideal world” version, analysis begins with codes: small, specific labels attached to particular pieces of text. Codes function as analytic tools during data sorting—often created in qualitative software as the researcher tags excerpts. As analysis progresses, related codes are grouped into more inclusive units. Categories sit above codes, bundling several codes under a broader, less specific concept. Themes sit at the top of the hierarchy, representing the most abstract, overarching ideas that organize the study’s findings.

Yet published work doesn’t always follow that clean structure. Some authors use all three terms when presenting results, and some supervisors may push students toward categories and themes (and sometimes codes) at the findings stage. An example given for online education challenges illustrates the hierarchy: a theme like “challenges students face in online education” could include categories such as “personal challenges” and “institutional challenges.” Those categories, in turn, could be supported by codes such as “financial challenges” or “lack/shortage of teachers.”

Drilling into why confusion persists, the key distinction is timing and purpose. Codes are typically meant for the analysis stage—summaries of what participants say, attached to extracts, and used to build a coding framework. Once the final structure is ready for the dissertation or paper, the write-up should focus on themes (and whatever sub-levels the researcher uses), not on presenting raw analytic ingredients as if they were the final findings. To make that shift intuitive, the transcript uses a cooking analogy: codes are like ingredients while working (e.g., yeast, sugar, salt), but the final product is the theme (e.g., pizza). In other words, codes matter internally for building the framework; themes matter externally for reporting.

Where categories fit depends on the researcher’s chosen terminology. The transcript emphasizes consistency over universal correctness: different methodologies and authors may adopt different label systems. In Dr Kriukow’s own approach, categories are intentionally avoided in favor of themes and sub-themes. “Benefits” and “challenges” become themes; “personal” and “institutional” challenges become sub-themes; and further detail is handled either as additional sub-themes or as “sub-themes” at deeper levels rather than introducing “sub-sub-themes.” The takeaway is not that any one approach is inherently wrong, but that students should use one coherent set of terms, apply it consistently, and be prepared to justify it when supervisors request a different vocabulary—especially because the same underlying concepts can be described with different labels across studies and traditions like grounded theory.