Narrative-Integrated Thematic Analysis (NITA): How can LLMs support theme generation without coding?

Large language models (LLMs) have sparked debate about shifting away from the coding paradigm; yet many applications have focused on replacing human coding rather than facilitating transitions to alternative methods. This article introduces Narrative-Integrated Thematic Analysis (NITA) that allows qualitative researchers to design, train, and guide an LLM in conducting thematic analysis. As a nonpositivist, pragmatist approach, NITA positions researchers’ reflexivity, intellect, and judgment at the center of the analysis process. This approach combines a reflexive, iterative monitoring, evaluation, and learning procedure (PERFECT) with a conversational method for interacting with LLMs. We experimented with the LLM through six stages: planning an initial PERFECT procedure, preparation, generating candidate themes, constructing individual narratives, constructing meta-narrative, and writing up. Our findings reveal the transformative potential of LLMs to support researcher-led, noncoding data analysis while maintaining interpretative agency. We argue that GenAI enables researchers to develop an alternative mode of thinking in qualitative data analysis.