Phase 3: Generating initial themes Here, you aim to start identifying shared patterned meaning across the dataset. You compile clusters of codes that seem to share a core idea or concept, and which might provide a meaningful ‘answer’ to your research question. Although we originally identified this phase as searching for themes, that language can be misleading: the process is not like an excavation, where meaning is lying there, waiting to be uncovered and discovered through the right search technique. Rather, theme development is an active process; themes are constructed by the researcher, based around the data, the research questions, and the researcher’s knowledge and insights. Where codes typically capture a specific or a particular meaning, themes describe broader, shared meanings. Once you’ve identified potential or candidate themes that you feel capture the data and address your research question, you collate all coded data relevant to each candidate theme.
Phase 4: Developing and reviewing themes. Here, your task is to assess the initial fit of your provisional candidate themes to the data, and the viability of your overall analysis, by going back to the full dataset. Development and review involves checking that themes make sense in relation to both the coded extracts, and then the full dataset. Does each theme tell a convincing and compelling story about an important pattern of shared meaning related to the dataset? Collectively, do the themes highlight the most important patterns across the dataset in relation to your research question? Radical revision is possible; indeed, it’s quite common. Certain candidate themes may be collapsed together; one or more may be split into new themes; candidate themes may be retained; some or all may be discarded. You have to be prepared to let things go! In review, you need to think about the character of each individual theme – its core focus or idea (the central organising concept) – and its scope. You also need to start considering the relationship between the themes, and existing knowledge, and/or practice in your research field, and the wider context of your research.
