Query-based approach: Can ChatGPT help us generate themes without coding?

The coding approach has dominated qualitative data analysis since at least the introduction of grounded theory methodology in the 1960s. The arrival and development of generative AI with the most prominent example being ChatGPT by OpenAI has led to the debate about whether qualitative researchers can finally break through the dominance of the coding approach by having the technological power to potentially generate themes without coding. In this article, we develop a query-based approach that helps create, train and guide a custom GPT provided by OpenAI in order to conduct a thematic analysis of a public dataset provided by Lumivero. This query-based approach, as an alternative to the coding approach to qualitative data analysis, is a methodological procedure in which researchers use conversational prompts and follow-up questions to guide AI in identifying themes in qualitative datasets as well as checking on AI-generated outputs. Our goal is twofold: one, to see if this query-based approach has the potential to replace the coding approach; and two, to see if under the guidance of this query-based approach, the custom GPT can generate themes without coding.